this is the link to download every drunk ranking
I’ll be diving in to rhe progress above soon today.
Awesome !! Sounds good it is a weight score based not necessarily ml based upon weights and if the max or min for each column is better is what the ranking Is based off of
@James Scott, so do these columns, sorted A to Z, with #1 being the "most negative stuff happening"?
• rankweightedProductsumNorminverse • rankweightedProductmaxNorminverse
Yes those are the rankings the column name is what kind of calculation it is
drug_types:
cardiovascular drugs, central nervous system agents, gastrointestinal drugs, respiratory drugs, hormonal drugs vaccines
Legal Injury to Compensation Model
- Death - $7
- Catastrophic - $6
- Severe - $6
- Minor - $5
- No Injury - $4 to $5
let me know waht you think, might have to have sme look into this for the correct grouping but from a stats perspective doing some checks thse seem to align well. i used a google bert model for tokenization maybe i could try a healthcare tuned bert one too which is different than a tf-idf vector space
@James Scott, what does Cluster Label mean? Count of instances in our dataset?
It is the cluster or group that each column name should belong to
Does the higher the ClusterLabel # the more severe or does it have no sequential meaning?
So we took 20845 adverse reactions, and the clustering put them into 49 groups of what it considered "related" by what means?
Just asking cause I filtered on "cancer" and see 311 cancer reactions spread across 34 of the 49 clusters
Yea I saw that too, that’s why I was referring to as a sme and I can see if I can put it into ChatGPT the stats portion of it checked out but when I looked it, it was ehh not how I would group them as a person. I can still work on it
the thing about this right is also eye testing its hard to validate, i am wondering why if anything is tehre a difference between bile duct cancer and breast cancer for it to be in different groups, like region based anatomy based?
cuz i see bile duct and blader are in the same cluster
because in a specifc cluster anal cancer stage iv bile duct cancer stage i bile duct cancer stage iii bladder cancer stage i bladder cancer stage iii colon cancer stage 0 colon cancer stage i colon cancer stage ii colon cancer stage iii colon cancer stage iv genital cancer male non small cell lung cancer rectal cancer stage 0 refractory cancer vaginal cancer stage 0
@James Scott, so sorry, holy crap, client and meeting gauntlet
Yeah, that seems to be the cluster from that sample, that is "trunk of body" region
I wonder what the Healthcare version would look like....
I think it’s interesting cuz body region isn’t necessarily what we are going for, more so like the stuff you mentioned previously I will try and see if I can adjust somehow based upon severity but would need a sme or hopefully a python library that can deduce that
WebMD has a pretty sweet "Conditions" taxanomy
https://www.webmd.com/a-to-z-guides/health-topics
https://github.com/AmirUCR/webmd_spider
Nevermind on last question, answered it myself. Health
Got a 1pm meeting then need to feed kids, but yeah, see if we can cluster by those health topics, that would help "dumb it down" for us but if you have an SME, what would that cost us?
I will see what I can do with this first
little update the dataset contains reactionoutcome codes that i think groups them buy everity if we want to use that as a grouping for reactions
but its almost impossible to find what they mean cuz i cant find it on there website yet lol
so if i find it and have the definitoins then this task would be completed cuz its the same code for the reactions just a different oclumn
he wouldnt. but i think this is something that i can ask as a question on there webpage or forum
What is this categorization called again? Reaction Outcome Codes?
yes on there site it said patient.reaction which is a list of objects within that object is reactionoutcome but they dont have it labeled as such
Great, I'm working on the cost estimate for our Thursday with Cameron for both AI and BI projects
I feel like with the Reaction Codes as a column, and the WebMD Health Topics as another column, we'll ahve a good classification as we look to bring a parent-classifier for those with some guidance from Cameron.
gotcha! yea the topic or type of what that drug is for would be good to have
For sure, does the FDA Approved Drug list have a data column for that?
so the approved drug list is just drug approval this is adverse events reaction type so it wouldnt be there i tried searching some readme stuff from the files and though extensive didnt have what i need
In the health topics, they do talk about drugs that are used for the symptoms in WebMD
looking over the cost stuff for the estimate and some of these prices sadden me like 50k-80k a year for looker is a lot to me lol would we get that much value from it?
Yeah; looker enterprise for sure. But if we were to build external tables in BQ, we may be able to circumvent the pricey External embed module. That is an option to discuss.
Especially if we go full app and just build our own HTML5 and JavaScript visuals
yea inn python now i know for a fact there is power/bi library that create visulizations free of charge, i think looking there is a looker api as well. is paying 100k for visulization or 80-100k that much better or worth it? i am not sure cuz i dont know how many people would use this as well as the overall impact in the company. if it is enterprise wide analytical tool that is fine and all our dashbaord for all departments people will use this tool fine. but if its a handful a people with no impact having or being savy enough for an api call in python with the same visulizations could suffice as well
Internally, we might have about 30 users, externally, though we will have potentially a couple hundred or more once our predictive model is in some type of application
So the external commercial product would be the only reason for Looker Embed.
So I see your insight as a great way to potentially save on licensing for building it in the portal and using python libs
ok that would make it worth it then, if it was me and you and nick or some other people i am not sure how much t would be worth it. take a look at this this is something that is normal for like analytics and ml models using streamlit or looker or something to create external products
interally and externally this is how our models were hosted using apps like these web a internal web address and you did the monitors and stuff on the backend with gcp and or aws services
Excellent, really appreciate this. I’ll dive in deeper all week.
So with that article, we don’t need the looker in bed access, just our local enterprise, API would let us share the data externally, I can’t ask Google, they would obviously tell us that we need both.
ahha yes i think i would or we would need to do some testing but it seems plausible to embed the data like that through a python or software engiener application/webpage
i dont know the exact use case that you guys are working on but from my understand so far yes
take a look, now i dont know what the columns 1-6 mean yet specifically but my hypothesis is that its by severity or reaction outcome
this is an article i found on there site
from looking at it the order changed slightly but not by that much
I’m at the office today, I’ll probably get very little focus time until this afternoon. But I will take a look when I have focus time.
Question, for your preferences, what do you think should be our internal meeting routine each week? I want and and working on this trip to carve out more "dev time" versus "doing other peoples jobs or working on things other people should be doing" tasks each week and day. Wanted your thoughts.
are you talking about the Aj meeting routine
Hmmm I mean once or twice a week is fine with me I have been pretty communicative on progress / what I need help in and our assitwnce how do you feel ?
Nope, the other way around. I wanted to make sure you didn’t feel isolated with our lack of meeting, because I need to spend more time on the project to help do the data schema and dictionary.
oh ahah no im great as long as i can chat when needed i am extremely self sufficent in that regard
no problem!! are u going ot be in the meeting earlier
*Thread Reply:* https://towardsdatascience.com/adam-latest-trends-in-deep-learning-optimization-6be9a291375c
https://www.analyticsvidhya.com/blog/2021/10/a-comprehensive-guide-on-deep-learning-optimizers/
the meeting is being rescehduled right
i have all the icd10 codes download now i am just creating the dataset to merge them all then create the bucket for them all
Hey Ryan I am traveling today won’t be able to make the meeting
Ok, me too. Maybe end of it but they’ll record it.
yea i know i was on the call yesterday and they seemed to have things under control
hey! i might have some time in the middle of the day to meet and talk, when my plan lands but i have to hop on another one can we have a sprint catchup when i land? i should land around 2 your time i think
@James Scott, yeah, our biggest client Dicello has me in probably my 300th hour of Camp Lejeune reporting today, so let me take a look at the calendar. We really need to get the thumbs up for the NDA. I'll bug Cameron more shortly. Safe travels.
Sounds good! I should be back to regular tomorrow and we need this signatures to proceed more
running into a little hiccup here with memory contrains to create this dataset do we have a company card or something i need to get google collab pro for 9.99 a month and see if this works for us
@James Scott, here is the SL CC for Dev projects:
woohoo i think it should get past the block with this, i should have the icd 10 dataset done today
I am not needed in this 5tran meeting right
so update, google colab isnt what was needed its good but i needed to create a machine instance to handle larger values of data that we have to manipulate, and now that i have that created now i can do this flawlessly, if all go well should send you or give u the query for the icd 10 codes with the adverse events
it keeps saying my quote/service limit has been reached are u familiar with this?
says have to increase the project quota
no i think its osmething on ur end to add quota could be a cost
Ok, we can check in the "Dev Team Admin" in GCP admin
We have some basic support or email Google gals if we get stuck
I'm working on a SQL change for Cam and investor...all week has been distractions for me.
ok i can take a look the dev team admin slack? i dont htink im in that
oh lol that ok ahha i can wait till then
oh no thats not icd 10 codes that how many reactions is in the adverse event dataset like 1 line per adverse event
i have a call coming up i will be 5-10 minutes late for this mening
i finally got this to work 77 million rows lol
thats for sure lol i will send u a link to the data in big query when im done with the fuzzy match joins
this is how ima group the icd 10 codes that we have
@James Scott , this is perfect for classifying the reactions.
RANGE TOPIC
• A00-B99 Certain infections and parasitic diseases • C00-D49 Neoplasms • D50-D89 Diseases of the blood and blood-forming organs and certain disorders involving the immune mechanism • E00-E89 Endocrine, nutritional and metabolic diseases • F01-F99 Mental, Behavioral and Neurodevelopmental disorders • G00-G99 Diseases of the nervous system • H00-H59 Diseases of the eye and adnexa • H60-H95 Diseases of the ear and mastoid process • I00-I99 Diseases of the circulatory system • J00-J99 Diseases of the respiratory system • K00-K95 Diseases of the digestive system • L00-L99 Diseases of the skin and subcutaneous tissue • M00-M99 Diseases of the musculoskeletal system and connective tissue • N00-N99 Diseases of the genitourinary system • O00- O9A Pregnancy, childbirth, and puerperium • P00-P96 Certain conditions originating in the perinatal period • Q00-Q99 Congenital malformations, deformations and chromosomal abnormalities • R00-R99 Symptoms, signs, and abnormal clinical laboratory findings, not elsewhere classified • S00-T88 Injury, poisoning, and certain other consequences of external causes • V00-Y99 External causes of morbidity • Z00-Z99 Factors influencing health status and contact with health services
yup! this was the plan its running now to populate the icd 10 codes
some of these are spot on some of these are ehh lol
the icd10 adverse event one row dataset is complete now ima look into the ranking with the icd topic and see how that looks
uploaded into our storage but havent made it to big query yet
Oh I was asking if we are still having it
Not yet, been hectic day with some Ben crump stuff popping up. I’ll take a look over weekend.
Thanks for the heavy lifting in data this week, between strep throat and other work stuff just glad to get some rest
No problem ! Just finally glad to get some traction on it
here is the upwork doumentation file that i created, i am familiar with upwrk so i know how to host projects and such as i also ahve a 5 star rating on there lol
I haven’t been on there for a while, so if you’re already set up, do you want to send links of the people and how many rows do you think we need people to review?
yes i can host the project, the question is what kind of cost should we do for it, 500 dollars?
there are some data scientist and stuff but thats not needed for htis we need icd experts in the medical feield
Nice.Hit them up and check price and availability please.
so the prive is ther ima interview them both today
had a meeting with one today they said they dont do one time projects but applied for it but was weird have my favorite canidate so far meeting tomorrow in the morning
Oh did I miss that meeting ? I didn’t get notified for some reason
so it would be 1000 this lady was good
Let’s do it. I’ll tell Cameron in our group chat, do you need the credit card?
i will retry in a little idk if its getting declined from something
@James Scott, have we charged anything to UpWork before this $1000 planned?
Like personally myself yes I have used it for clients before
And I have maybe about 3-4k from just freelance work before
No on the 9356 card, have we charged anything on UpWork prior to the planned $1000, Cameron was asking me about $1200.
And the previous charge I used it for was google collab but I canceled that that was 50 dollars
are u getting 2 different charges for upwork?
this is my upwork cost for the past 6 month
And the ones listed as Payment are from 9356 correct? Are the October 2023 ones from our account too?
From reading it, its 1155.95 from 9356 we've spent on data analysis in past 6 months correct, just need to confirm so I can tell Cameron not to worry.
no the october ones arent from your account i wasnt hired then lol
Is there any predictive analytics we can do forecasting with some of these metrics / dashboards
I was on the call, but let me reschedule, Crump team blew me up all morning
I didn’t know most places have holiday today I could joined but I disnt know if u we’re haha why I messaged but I can meet now or later
All good, tomorrow we can setup time for the MDL help from Michael Brunet, our Chief Legal Officer, and discuss IDC-10 project, and CDC data dictionary. Plus ping Shellie and Cameron to get Application Software Engineer resume out. Since we'll want to get our architecture stack setup for the early stage lite-AI product, we should discuss giving you a Sprint Board for it to manage with this person and people as the project grows, sound good?
The project and the retraining of your algorithm?
No just the project I am going to retrain and re run in the morning
Position Overview: We are seeking a talented and motivated Application Engineer/Software Engineer to join our team. The ideal candidate will have a strong background in Java, SQL, Python, JavaScript, and experience with various Google Cloud backend services. This individual will be responsible for developing both front-end applications and backend solutions, as well as implementing pipelines, monitoring systems, and contributing to the overall development process. Responsibilities: • Design, develop, and maintain front-end applications using modern frameworks such as React, Angular, or Vue.js. • Utilize Java, SQL, Python, and JavaScript to build scalable and efficient backend systems. • Implement and manage pipelines for continuous integration and deployment (CI/CD). • Develop monitoring solutions to ensure system performance and reliability. • Collaborate with cross-functional teams to drive product development and innovation. • Stay updated on emerging technologies and best practices in software engineering.
And JUST had a brainstrom with Brian, how does our heuristic sound if we call it "Tortelligence Impact Ranking"? We can mad-lib in Company Impact Ranking, Drug Impact Ranking, Reaction Impact Ranking and the word "impact" does not explicitly imply negative. @James Scott?
so from this aspect right is this going to be marketable to other companies, as in a weblink.google search right?
Cam will market it but yes, our star will be our Heuristic.
Tortelligence Adverse Events Drug Impact Ranking - something like this i like, i think just impact doesnt really explain what its about if i am google searching or viewing the page, that we people searching for drug impact, adverse events, fda, things like that owuld also be able to get our page as a hit
Technical Skills: • Proficiency in Java, SQL, Python, and JavaScript, with a strong understanding of object-oriented programming principles and design patterns. • Experience with modern front-end development frameworks such as React.js, Angular, or Vue.js, including state management libraries like Redux or Vuex. • Familiarity with backend development frameworks and tools, such as Spring Boot, Django, or Flask. • Hands-on experience with cloud computing platforms, particularly Google Cloud Platform services like Cloud Functions, Cloud Storage, BigQuery, and Cloud Pub/Sub. • Knowledge of containerization technologies such as Docker and orchestration tools like Kubernetes. • Understanding of microservices architecture and its implementation using technologies like gRPC, RESTful APIs, and GraphQL. • Proficient in implementing and managing CI/CD pipelines using tools such as Jenkins, GitLab CI, or CircleCI, with experience in automated testing and deployment strategies. • Experience with monitoring and observability tools like Prometheus, Grafana, ELK Stack, or Stackdriver for ensuring system performance, reliability, and troubleshooting. • Familiarity with relational and NoSQL databases, such as PostgreSQL, MySQL, MongoDB, or Cassandra, including data modeling and optimization techniques. • Knowledge of version control systems such as Git, including branching strategies and collaborative workflows. • Strong understanding of software development methodologies, including Agile/Scrum, and the ability to adapt to evolving best practices and industry standards. • Excellent problem-solving skills, with the ability to debug and troubleshoot complex issues efficiently. • Effective communication and collaboration skills, with experience working in cross-functional teams and delivering high-quality software solutions in a fast-paced environment. • Continuous learning mindset, staying updated on emerging technologies, trends, and best practices in software engineering.
Soft Skills: • Strong analytical and critical thinking abilities, with a keen attention to detail. • Adaptability and willingness to learn new technologies and techniques. • Effective time management and organizational skills, with the ability to prioritize tasks and meet deadlines. • Proactive and self-motivated, with a passion for innovation and continuous improvement. • Strong teamwork and collaboration skills, with the ability to work independently and contribute to a positive team culture. • Excellent communication skills, both verbal and written, with the ability to convey complex technical concepts to non-technical stakeholders.
there is os much data in this cdc lol i am trying to narrow my focus on mortality? with the icd 10 codes or dieases and birth waht do u think
Ugh a lot lol like kinda pick our poison but seeing the fields is hard
We want to learn about diseases / reactions that would have stats about the mortality, the locations, and other data that would give insight into the "who, when, where and what" of diseases that we can match back from our Adverse Reactions datasets.
Look into pacer / MDL websites to see mdl statistics and how to extract
Pacer api cost / use / data dictionary
We got your GCP Project setup in DBT, created a SL-BI GitHub for true dev, staging and prod code and connected it all. Next ByteCode session you should join so we can make sure we know how to use DBT to go from dev, staging and prod on the transformed tables that would make up the application tables used in Looker and the app UI/UX. We want to ensure SL, TIP, AI and all projects are setup for build the same way.
The email datascience@shieldlegal.com is a GW Group with you, me, Nick, Brian and Luke on it.
gotcha yea i saw the emails yesterday!! and i will deff join it
Great, we'll walk you through "what we know in DBT" so far but it aint much. LOL
i think i can crank out the census data to get populations of cities quick turn around on that
have the census data downloaded and uploaded to the storage going to make a table shortly but i think it might only go up to 2019, hwich we could use a regression or estimate currently populations, i ca look into that now that i know the fields i am going to look into the pacer mdl data to see if we can get data on that to join
idk if u say the pricing for this but i tried to get it to be personal lol but we most likely need enterprise what od u think
@James Scott, had total Lambda failures on both TIP and Shield daily financials today, and been helping Media Buyers get their tracking done. I'll reschedule our AI meeting. Will get to your questions shortly. 🚥
https://unicourt.com/courthouse/judicial-panel-on-multidistrict-litigation-1429
make datascience@shieldlegal.com a user so we have one user-service account on otp of yours
awesome i am going to be working on cdc stuff
just started to do a search for some of these phamaucitical companies
and for janseen pharama there is 105,000 court cases
we can even use this for indiviuals i am trying a trug search
Nice. That’s awesome. I spoke to one of our largest clients who was infatuated with what we are doing including future projects for media buying optimization.
@James Scott, if Cam gets on let me know, but he's up to date so we are good to skip today, been behind in 30 minutes
Ok that’s fine !! I don’t think he is prob going to get on
We had an amazing sales pitch with a power player yesterday on tortelligence.
No, it was the corporation that owns law firms that uses "tech and analytics" to optimize class action lawsuits, just getting into mass torts. LOL.
He asked me what I know about marketing optimization...LOL
Oh that’s good news to be already pitching to clients it’s very exciting !!
Future is bright....from our legal research platform to the marketing / operations platform. Once we get you and Nick together connecting the dots and providing those actionable insights and datasets, boom.
It's real and the pioneering spirit was heavy in the call.
I am excited as well!! I think we can do some really Cool stuff with this api as in co bing through it I think it has the meat of what these firms want to see. However I did talk to there support team today the subscriptions might increase because the premium doesn’t include the API which I thought it did
And Nick is leaning in to help Dicello get TIP systems synced with IDs into their in-house case CRM, so after he builds TIP and SHIELD pipes, we could use your help on SQL transformation (the hard ones) and Python Lambdas in AWS to ensure we code the SL and TIP business logic correct, while this gets you access to the datasets of lead-case-criteria for these drugs we focus on, like Ozempic.
So the cross over from AI, as we have portal being built for the initial AI models, you can help draw the gap from legal research to legal mar-ops-tech (aka consumer side).
Next 2-3 years will be fricking awesome (excuse my lame cuss word)
I have done and built lambdas in python all day long sometimes annoying as it may be I can do a lot of that ml ops side of things
And I kinda wanna start building out a looker dashboard with what I already have cuz it’s going to help me Brainstorm the datasets flow on how I need it to look and feel
I'ts more of teach the youngsters but learn the business logic implied in these lambdas to help guide darn-complex SQL
Awesome !! I will get with u tomorrow on a looker dashboard I can spin up
let me know if you ahve time for a quick call today
Yo, lets push our AI meeting to tomorrow's meeting, I'm slammed here at the LV office, as usual!
hey so i am getting access to the trial today
so i talked to tehre sales some states charge for documents
oh, that is just for "court scanned" docs for gross detail on a case facts & matters @James Scott?
yea they have to clarify that but it seems like if i do hey give me the number of court cases containing xyz then its just the normal subscription. but if i want to download the court cases then its the subscription + whatever the state charges, some states are free some aren't.
however the yearly subcription for the api is 27k
Ok, get a SOW and what about geo data? That in the datasets.
gotcha i will i have a 1 month free trial of it so we have time
he said he would give me a call i would talk to him botu the project adn the engineer correct
ah i had to ask for it but they have a whole python github on how to run queries and stuff this is awesome
finanly connected to the api in pyhton
Lets get into what location and demographics data we get from this case data.
Once you find the "join" fields for company, drug, etc.
writing these queries iwht the api is so annyoing lol
What's up? It's an object model API where you have to relate endpoints and cycle through?
Yes I’m wedding throw it it’s just kinda complex but I think I got it so basically to to funnel trough the layers so say I want the party pzier s a defendant right I have to hit that api return information then I have to search for that extracted information in another api then I have to take that information to get the final number of cases by a geo location
have a meeting on monday with unicort they will help me set up the quiers so next week we mwill hopefully start buidling out the dashbaords
*Thread Reply:* @James Scott, excellent, thanks for the update
Yes, figured I’d give you the week to get a hold of UniCourt, we meet Friday while I’m in Vegas.
oh i alrady talked to them have data and everything
wanted ot show u but we can meet tomorrow too if ur busy
this is an example of one of the documents downloaded
@James Scott, sorry, Luke is out with jury duty, so I'm still doing morning numbers today
But I added meeting to review it tomorrow morning.
hey do i need to do an out of office thing and if so how do i
hit a little snag the python in the google ai stopped working
We pay for light support in Google Cloud, but not live , on demand support.
can u join this meeting with unicourt a second
i forgot to ask what do u think about what unicourt offers and getting access to everythhing
I think because it’s the most important data source arguably for the model, that it’s a pretty fair price. As soon as you can get me the information on what the incremental cost during development would cost us, I’ll talk to Cameron. Whether we get it all now, or climb the steps to getting it all with him, we’ll get it.
i think its great to have accesss to the hwole data warehouse they also seem to have insurance information as well and if its healthcare insurance info than thats great we also have access to the full mdl too
@James Scott, what will our "Top 10 Drugs from your ranking" cost based on the 0.13 call overage rate at the $27K per year if you aggregate the all the related case-file data. If I know that that gives me the "middle step". Meaning, I can see us getting all of it but want to show a "path of cost so I can see what total investment over time" would look like. Plus, can you tell me how many of our "Top Pharma Companies" in your ranking have cases, and how many calls that would be per month?
Then a quick Google Sheet "calculator tool" would let us play with volumes. That would be very useful for my to understand the impact of $120K per year option and how to get there.
yes, so i can easily provide that information, however I think that i am at the limit again for API, i will discuss on the call with them today getting that information on alaytics of how many of them have cases, we saw that pzifer has at least 20k cases, how many ongoing? im not sure. i will have more information later, more so i also asked for a data dictionary of what we would be getting for the sql side of things to enhance the investment purposes
Sounds good, but also do you know why they said 25,000 calls per month at first, then switched it to 10,000? Meaning is there another tier that we can pay for that doesn’t jump from 27,000 to 120,000?
i highly doubt it, it seems like its always been 10k i thought it was 10k as well. but even so our data downloading of these cases is going to easily exceed 25k as well as the document download.
Sounds good, all right just let me know on the cost estimate, and have them send me that side-by-side and we’ll get it in front of Cameron.
I have a doctors appointment, but I can talk while I’m driving there shortly. Send me your mobile phone.
they said the 27k would be credited towards the 130
waht is shield legal address is it phillips road
i dont have any more api calls until we sign lol i can do some data and project manamgement until we get the sign off
@James Scott, I will get the $27K invoice paid today.
thjank goodness for fireflies they went back and viewed our transcipts and we were told 13 so they said they will adjust it to 13 cents !
yea right now i am adding things to gitup updating the end to end process so it runs smoothly in the mean time while i wait for the new contract
*Thread Reply:* call my cell or u wanna talk on here
*Thread Reply:* Send me Google link, I"m heads down on quick tasks
*Thread Reply:* INSERT INTO 'tort-intake-professionals.tip_prod_application.io_lr_case_types_revenue_rates'
(id,
lr_case_types_id,
io_casetype_billable,
io_casetype_billable_rate,
io_agent_commission_rate,
io_call_time_minute_rate,
io_call_time_flat_fee)
VALUE
('1','1650', 'true','122.00','60.00','2.00','0.00')
INSERT INTO tortintakeprofessionals.tipprodapplication.iolrcasetypesrevenuerates (id, lrcasetypesid, iocasetypebillable, iocasetypebillablerate, ioagentcommissionrate, iocalltimeminuterate, iocalltimeflatfee) VALUES ('1', '1650', TRUE, '122.00', '60.00', '2.00', '0.00');
-- Create a new table CREATE TABLE tipprodapplication.iolrcasetypesrevenuerates ( id INTEGER, lrcasetypesid INTEGER, iocasetypebillable BOOLEAN, iocasetypebillablerate NUMERIC, ioagentcommissionrate NUMERIC, iocalltimeminuterate NUMERIC, iocalltimeflatfee NUMERIC );
-- Insert data from existing table and add values for the new columns INSERT INTO tipprodapplication.iolrcasetypesrevenuerates ( lrcasetypesid, iocasetypebillable, iocasetypebillablerate, ioagentcommissionrate, iocalltimeminuterate, iocalltimeflatfee ) SELECT ** FROM tortintakeprofessionals.lrcasetypes.lrcasetypes;
CREATE OR REPLACE TABLE tort-intake-professionals.tipprodapplication.iolrcasetypesrevenuerates AS SELECT **, id as lrcasetypesid, NULL AS iocasetypebillable, NULL AS iocasetypebillablerate, NULL AS ioagentcommissionrate, NULL AS iocalltimeminuterate, NULL AS iocalltimeflatfee FROM lrcasetypes.lrcasetypes;
They are funny lol they are sending over the updated contact with the 13 cents cuz I asked for 20K+ api calls if they kept it at 15
What is your mobile number they need it for invoice
I could, I did move it, I am in the SQL hole digging my way out on this copper crap, people been blowing me up all day on it.
just needed your opninion on something real quick
this api is so annyoing. i think i cant do this hititng the api so i have to try downloading the jsons and reading it that way. its crazy how 1000 api calls act like its to much
Not as robust, the "Fivetran to Snowflake" will be better in the future.
this si really cool ai sofware for meeting notes integrates with google cloud for task and certain things and slack as well
Not sure. But we pay for FireFli, let me know if you want to be invited
update still in a blocker for the api calls. wokring with there support
Great. I plan to pivot into our AI dashboard next week and really get into the weeds of pushing Tortintellignce.ai .
gotcha! thats good with me support just got back to me let me try this and see if it owrks
figners crossed i have the 11 thsounds + cases identified for pzifer since 2017+ and then i am goign ot be readying over 100k documents so hopefully i can find some informaiton maybe i have to improve the search function of text idk but im hoping ot see something
let me know if you have some free time this morning
I do during our tech meeting @James Scott, just been crazy morning, as Mondays...
Ok that’s fine I have some updates too on what we want to don
so the cost of the addition month i did was 7,074.47
I got the invoice, it was 54,000 API calls, can you give me the stats that we talked about on that, how many calls per case, which each call was pulling and what percentage of Pfizer case data?
Morning, just spoke to Cam on phone. Told him we spent $7000-ish for API overage fees as incremental data modeling fees but that the $120K per year will be a MUST when we scale the data and product. He said "cool", so time back instead of meeting @James Scott 👏:skintone4:
Good morning ! I am under the weather today I won’t be in attendance for the meeting however I am running the code now to get the case ids and there associated drug so once that done the main focus can be on the dashboard since everything is tied together
@James Scott, I pushed to tomorrow. I may becoming under the weather, so hope we both beat it but thanks for the udpate.
SELECT ** FROM ai-projects-406720.drug_model.unicourt_case_documents_drug_counts LIMIT 1000
now i can relate this back to a ocunt of open and or closed cases that contain those drug names and use that as a metric in the dataset
I don’t think we will be charging much this month
what channel would i use for the tort dashboard?
for me and Cam only use #C067XCXTURL and for entire dev team use #C065YE3Q9UJ 😎
And Keep Luke busy on building a sexy dashboard for AI project in Studio. Once he is done, I'll probably make him move it to Looker Enterprise.
yea luke was working on it right? wanted to have a session with him
@James Scott, yeah, you'll need to ride him to complete it. His daily routines of data analyst work do hijack him so please send out a recurring meeting for that for at least twice a week so we can keep on pace with showing Cameron.
I will be a couple minutes late finishing up this house inspection
like access? admin access would be great but as far as services i need to set up bedrock and i think thats about it
@James Scott
Console sign-in URL
https://960735831404.signin.aws.amazon.com/console User name
jamesscott@shield-legal.com Console password
Sq7HK9c'
Hey I have. Call with my realtor at 230 I can give updates on the ai meeting at 3 about my progress and the dashboard
whats the link to the data svhema u created i just see it as an imagine
i have to edit it and add in table names for luke
We good, wasn’t sure if we used anything for HF?
no we ahvent used hugging face for anything !
*Thread Reply:* https://www.everlaw.com/blog/ediscovery-software/what-is-ediscovery-software/
for the llm i created a 100 dolalrs budget shouldnt need it but just wanted to let u know
I just submitted to 3 legal analytics AI platforms for research, and a few e-discovery / e-legal-assistant tools.
think ima revamp the metric of the finding of the drugs to the llm did some checks and where the origanal model is saying found the llm is saying it not
Also I’m pretty sick today still I saw u set up some meetings do u need me on them
No, lets punt the "webinar instead of demo" on Case Text and will reschedule. What about the AI dashboard meeting? I want to make sure you and Luke are moving forward, should you push to Friday?
Yes we can push that to tomorrow too I have a sinusbintectionbandncokminncokd just came back from the doctors
Ok, i'll get the e-discovery platform reveal to demo after qualifcation call today later
You can skip, they are just qualifiying us for the demo to be seutp
Told him he needs to spend hours learning the data, fields and try to be resourceful then after that, have q&a with you and me , per our chat
Ask Shellie Mooney, I actually don’t know our holiday schedule but if not, take some liberty cause we should
Just don't mention tortelligence.ai and we are building our own tools, your role is to give intelligence on our media strategy
thats a nice feature for multople projects
Ask them what type of AI they are using on those models
Hey, got a demo coming up for a product Lex Machina that uses data like UniCourt to just create "legal reserach" for attorneys. Massive help in classifying UniCourt data once we see the demo. But they do not have the breadth of what UniCourt has:
Let’s see it I saw it ! I am going to be at Disney but I can deff listen in
I’ll take screenshots for you, it just gonna show you how to use some AI to extract an auto classify with some supervision all sorts of objects out of data to organize a better in scale.
i will be able to join and listen in and asses
saw your email i was going to ask how was the meeting
Want you to see how the classify court data and not sure they have heard of "UniCourt" either, so shhhhhhhhh next Monday. I'm around tomorrow and off Wed to Sunday for vacation.
If we use this taxonomy on our UniCourt data, we’ll have a way to make the data understandable and digestible for users
UniCourt has more than them right? All 50 states and federal
hey sorry id idnt get these notifications and yes they do i think unicourt is more robust but the interface extraction with this product seems a lot better
i have a call se tup with unicourt now
is there also any other ai/ml related task we can do? somethign quick and simple that would improve htings
How is the MDL baseline versus the prediction MDL list going?
That is what we need to get a dashboard done by me for Cameron.
i am wokring on it for the api i should have some clarification questions for my call with unicourt on friday then add it to the dashboard
And then I want to have some AI to identify the "document types", "parties", "complaints", "motions, "etc."
Did u say something else it said u did I think but yup I am on it !
I am going to lean in, we are soon cutting ties with Luke, but keep that 100% confidential.
If you can get into Looker Enterprise and build this MDL Historical List that lets us "judge our" Impact Ranking against known mass torts, then create your preditive MDL Futures List based upon this Impact Ranking Projection, that plus the stats on the cases, deaths and all of that will be good to show Cameron.
Then, could you write python cloud functions to build Shield pipelines in Fivetran to BigQuery?
*Thread Reply:* This would need to be talked about more but should be easy to do
*Thread Reply:* For sure, I am discussing with Nick and Cameron on Monday but this would be easy and super helpful to get us out of a rut.
That is non AI project but would help me and Cameron speed up a few things, these other analysts and new hires in AWS land have been not working out. Love your thoughts on helping there...cause once we get about 4 to 5 pipes built, then we can finally get the AI going on the Marketing Strategy for Cameron where you and Nick can do that together along with continue to iterate on the Drug Mass Torts model and court data stats and content analyses functions.
Let’s get on a call about it tomorrow and discuss more !!
We will do that in our AI meeting on the calendar for an hour
Please add estimated delivery dates on the subtasks:
https://tortintakeprofessionals.monday.com/boards/6304263960/pulses/6305306724
*Thread Reply:* @James Scott, please add status and dates to Monday! 🙂
*Thread Reply:* can you link me to the project? its not in the ai one
*Thread Reply:* https://tortintakeprofessionals.monday.com/boards/6304263960/pulses/6305306724
*Thread Reply:* i believe id id this correctly go check
Yes, i won't know until later this week the final wind-down plan, so shhhhhhh. 🙂
ok thats why i messaged seperatly, i wanted to use him to teach me something on looker real quick
from the email, there is a lot of calls, i was trying to like guage this so there are 1100 mdl cases, but the sub cases of those 1100 are a lot, i got to 3 million before i stopped and asked like tis this corerct which rob there ai data engineer and arichert of it all said hes not suprised, which through the switch in approaches we are now focusing in on the treasure drove of data they actually have because our focuses is norrow to mdl and only those mdl ids
i am running jsut pfizer now getting the subcases
for just 69 mdl records lol or litgations
That’s great numbers. When we provides your dash, we have to convince Cameron to let us get the Snowflake ❄️ license
Mhmm! To jus focus on Pfizer for the dashboard and that the other 19 companies
So hopefully it’s not to many but that would be great to even read the model through and see what days we can pool with the drug names I’m sure with the narrowed and direct focus we can specifically say these drugs weee in these cases
Me too on the Shield Financials + Copper + Leadspedia 127 column DBT data model View. 🎯
I'm on MSFT teams with LexisNexis for their AI products Co-counsel
I am in a meeting with one of my friends to get this looker dashboard graphs done I had a question that is needed for developing he’s an expert
do we have looker entperise ? and do we have looker support
let me know what you think of it sos far
go to the brand x drug sheeet in tehre
Looking good @James Scott, question. Will you show a column called "Impact Ranking", a heuristic based on 100 (most negative impact) and down? That "Tortelligence Impact Ranking" becomes a true showcase of our prediction model. And in a table report for that, would love to have the "drugs" first, then companies" second, having the drug anchor the ranking, it will allow me to look at the "Existing MDLs by Impact Ranking" so we can "validate" the model a bit based upon what we know about the market. Then the Predicitive Impack Ranking list will be more "human trustable".
*Thread Reply:* Yes I thought I responded
*Thread Reply:* Some clarifying questions, impact ranking ? Is that our current ranking algorithm
*Thread Reply:* Maybe a call would be good to explain
*Thread Reply:* @James Scott , yes. Having our AI/ML as a heuristic that productizes it.
*Thread Reply:* Ok let’s do in a little bit
*Thread Reply:* later afternoon if thts ok so i cna finsih this up over the weekend
*Thread Reply:* Sure. I’ll be busy from 1pm to 2pm CT
*Thread Reply:* im ready whenevr i dont know what 1 pm ct is on west coast time
*Thread Reply:* @James Scott , I’m strung up on like four things. Will be a while or we can talk early Monday. Let me know.
*Thread Reply:* i am here jsut let me know or we can talk otmorrow
@James Scott, and noting the manufacturer names have same companies with different labels, can we clean those names up so we can group by "single unique manufacturer"?
yes we can clean those up i havent done that yet and yes i can !
let me know if u wanna have a quick call today
(2) Upgrade to 25k calls/month. This would be a new 12 month contract. • $41,088 - $18,392 = Payment due upon contract execution would be $22,696. Overage rate would be $0.128.
so i might need to chance the prediction its not producing the best results
@James Scott, show me a sample of the historical and predicted lists?
this is the stats for number of cases
Mean Squared Error (MSE): 310804.95210385695
Mean Absolute Error (MAE): 181.16032413460687
Root Mean Squared Error (RMSE): 557.4988359663695
R-squared (R2): 0.9707355661245721
Mean Squared Error (MSE): 85.36494334930144
Mean Absolute Error (MAE): 7.0130719768170415
Root Mean Squared Error (RMSE): 9.239315090919968
R-squared (R2): 0.9889710395461634
i can do a resdiual grpah but i thinkt his is like pretty good actually 7 mae
this is a graph that shows how close the predicted were to actual the closer to 0 the better
i am adding it to the graphs and adding the map graphs in there today
It keeps kicking me out but that’s fine with me the dates
I am adding in the maps and stuff shortly but I have a mock of everything else in there
Awesome. I’ll take a good long look tomorrow. What times you free?
take a look !! https://lookerstudio.google.com/s/uxwa33c1LDY
I am in the tropical storm here in Florida my WiFi have been in and out all day
I am in London at tiktok with Andy, Nick and a new Social Media person
It’s great, ping Cameron and ask him to confirm when in September he’ll be in Vegas for me please in the #C067XCXTURL channel
@James Scott, can you list out all of the ML we are using in our prediction model below for me please. Have a meeting with investors and need to get prepared.
Linear regression algorithm, and the ranking algorithm. If you want we can explain what these are more like if we are using a deep neural network for regression
I can switch to a DNN for the regression or future prediction of rankings, and all ranking algorithms is a weight maxnorm algorithim
Deep Neural Network, so we are using Deep Learning @James Scott?
yes, i can switch our current algorhtim for forecasting the predction to a deep neural network
its not a long fix, it mihgt make improvements to the forecasting, which does need improvements
Is it something you’re actively working on to do? Before the September meeting?
Hey didn’t see this yes it can def be done by then
DM Cameron and Brian Thiessen for expense reimbursement procedure for airfare
Just a Marc introduction, he's our investor. No risk to us, just aid to us, IF, we used them. Maybe ask them about their depth of AI experience?
Why Ryan I am still feeling terrible can we postpone to Monday morning
How you coming on catch up from last week? I can push this to tomorrow, got a mapping coordinates request from the investor, that's a few hours task.
good morning, you were saying something yesterday about the dicussion with the team, what should i say about the meeting for the trail run
To talk about the subscription with unicourt ! You said reach out to them to discuss the document chat bot I wanted to hear the results of the meeting to tell the support team for access to the trial run to search the documents for the drugs
do you have a looker template for the nice tables you were tlaking about
or point me where to look for one to apply
In Looker Studio, make a Data Source that is the SQL for your two tables, then just use their Table object or Pivot Table object. Here is one that is a good example of a table you can Edit. Be careful, Sales Ops uses this dashboard every day.
https://lookerstudio.google.com/reporting/a36366d9-97f6-48ac-b7d9-246a18bbbe99/page/p_kujrmk8d2c
And if we get this dashboard done and need to wait a until next week for our UniCourt meeting, I do have a Fivetran Cloud Function for some data pipelines I need done that aren't getting done by the team due to Cameron putting Integration atop the priority list. I'll talk to Nick McFadden about that and let you know. 🙂
Approximate Mass Tort Pharma Injury Settlement Values
Death = 1,000,000 Severe = 250,000 Minor = 75,000
take a look at the dashboard, i added everything
let me know when you take a look and what more we can add, i was thinking some color coding of the report titles
I will, just got a busy morning. I can take a quick look now.
The new columns look good. Add those to the 2nd tab / page here
I will create pivot table versions without the years tomorrow so we can see how those look, grouping on drug name, sorting by impact ranking descending and rolling up metrics years combined.
sounds good oh ye ai forget to touch the second page let me look into that too
let me try the color coding of the titles
did you see unicourt email? i am not sure what cameron needs to be on the call but we already pushed this back aweek
So make sure in our dashboard list prototype to have cases, deaths and those columns be summed metrics, once we inquire with Unicourt about how to get the "match of the drug" and the "MDL Child Case count" for industry stats. Anything else @James Scott we need to cover?
Oh, the Beta of their Gen AI product to accomplish these possibly.
i am confused as to the first part you said what do you mean like using fake dat afor our meeting or?
sorry lol like fake data for our meeting
So make sure in our dashboard list prototype to have cases, deaths and those columns be summed metrics
Oh. No, I'll show you what I mean below. So that as we explain to Unicourt we need to make sure we have the count done right, then a pivot on the "Drug Product category" synthesized column might show more sensical "summed totals" in the case count?
I created 2 pivot tables, once you have validated the API method for MDL cases, let me know. We can spend an hour going over the AI dashboard. I have some educational questions on the data to ask to make sure I understand it all. But the 2 new pivot pages are ready.
arent they sending us some stuff as well, want to make sure not to dplicate work
Can you validate the API method just so we have that nailed down from our "current service model", just so we have that in our back pocket with Cam
yes of course, i am wokring on adding the lead case in there now
good morning, are you sending them the list to test this week or me? wanted to get that list and data in the dashboard so we can have something concrete to show cameron that is there
awesome! so they have to sign the nda before we can send that list of drugs? i can get the drugs names now
Yes, please send that list over for me. I will send that to them while I wait the NDA. It's not involving our AI model.
here you go this is just the unique name of everything you can sort a-z for the 10 more specific ones we need. and once we have that and if they find that, wthen we can filter our dashboard as an example for those 10 names
@James Scott, here are the 12 I will send them from your list:
CYMBALTA
FARXIGA
LIPITOR
LYRICA
MOUNJARO
PAXLOVID
PLAVIX
RINVOQ
TRUVADA
VIAGRA
ZYRTE
hey i wont be able to make the meeting today, i have to head to the hospital my grandma got a big rip in her intestines
@James Scott, sorry to hear, giving my best wishes.
When you get back can you make sure all data panels work on this original dashboard? https://lookerstudio.google.com/u/0/reporting/69008bc1-14ec-420d-9780-4635d5550f08/page/AqWxD
ok and i saw the email unicourt sent too
should we send them the case number of some asicated stuff i been goolging to try and find mdl stuff but i have case numbers
Ok awesome ima just give them the web link as well
i found this site that pulls up all cases with like the drug name in it
maybe we shouldnt be using unicort or something like this
this site migh tbe better than unicourt
Hmmm. It’s a Thomson Reuters product. Not sure its source data but we could learn more from them. Did you reply to Jasmine yet?
i have the list ready to give them i was going to reply this morning i just sent some links to the cases, however i contacts that site to see if i can filter on mdl cases i have the case numbers not mdl numbers
Ok. Send UniCourt what we need and we can get case text demo maybe?
can you send me the hotal reservation details
i wouldnt be ablt to stay there sunday night i fly in
Call up and add a night to your reservation at Cosmo direct or try one night at Aria or Bellagio, it is walking distance to Cosmo.
ok let me take a look and are u joining hte meeting
7180 Pollock Drive Las Vegas, NV 89119 2nd Floor Check In with Reception (Glass door with our logos)
I have a call with one of the other services tomorrow to see if we can find what we need in time form the meeting but do you think we are good to go or need more stuff
Stall them one day until after Unicourt convo tomorrow and lets see
@James Scott, going to push our AI dash meeting, Marc just loaded me up with ad hoc tasks stealing a few hours of my day
another caveat is the generic name brand drug
they found lipatpr 3000 cases associated with it, the caveat is that not all these drugs are int here buy there drug name, it could be there generic name and or there substance name so from our list they found cases with lipator but not the other ones
do we care if they are active or inactive
Lets get all cases, and then we can ask Cameron if they want cases active or inactive or both? (probably both)
And the explanation makes sense, these drugs could use slang, substance, drug brand, drug lab names, etc.
So using the MDL Dockets and Parent & Child Case number is the superior method
we are going to look at 2015 forward mdl cases active and non active
good morning run, take a look at that email, there is a lot of mdl since 2015 we were looking at as well as some pertaining to some drugs, we need to pick which ones we want them to give us counts for i believe
no like which drugs do we want them to get the member count cases for
That what I meant, with our early model using FDA data mostly, would our Top 10 not work?
i think we are looking for the ones with open and or closed mdl, form the list of drugs we went over, i dont think thats top 10 list but those are the ones we are focusing on getting member count casses for currently
Then should we use the Historical list and use any that have the MDL?
• BIKTARVY • CYMBALTA • FARXIGA • LIPITOR • LYRICA • MOUNJARO • PAXLOVID • PLAVIX • RINVOQ • TRUVADA • VIAGRA • ZYRTEC
this is the drug list you sent over we are wokring on, some of these have been found from the mdl name
I picked ones that I recognized for that list form our Historical List.
yes thats the same list we are working on as a poc
they sent an email clarigiying what they are going to pull
i jsut made a table for the data count records, ima add it in the the board as a hey you can see how its going to go if we filted for these drugs this is the mdl dates, there member counts, there xyz
i think we might need a quick working session cuz some of these arent neccesairly counts and i cant add them in the pivot
mdlfileddatemdlcloseddatemdlopenclosedtotalmdlmembercasecountmonthsasmdl
brandname mdlfileddate mdlcloseddate mdlopenclosed totalmdlmembercasecount monthsas_mdl
I did 30 minutes ago, I had to takeoff for some errands
Looks great. So the date they provided for those counts, is that not in the API? Or they’re having to put those in manually into the API?
That is from pacer I don’t think so it might be
Dashboard enhancement been moved to 2?
@James Scott, I have finished my design and clean up to ensure all pages look the same whether pivot table or table style. We can discuss the UniCourt / Pacer data in our call at 12PM.
@James Scott CHANTIX is active MDL, please update database
i tried its not that easy updating that one buut i might not make it in time for that one record unless i can hardvode the value in looker
And ChatGPT4o read this entire PDF and let me ask about matching drug names, pretty cool feature we need native in our future app! 🙂
is we can have acess to chatgpt in the azure dataframe we can use that as our ocr took to read pdfs, its very similar to the current state of things in aws though too
he'll be on at 00:45:00 after, pushing up meeting in calendar
notes from meeting - ask unicourt is in the api already that would have settlement date, import warning label for these drugs and compare against adverse events side effects, (fail to warn metric), bigger thing is viability of that defendant defendant information, us based, have to have money from manufacture, issue that is not warned against, index injuries and scrape against whats not warned on this
data on drugs side effects warnings adverse events not warned for, where are these people living that are reporting it, demographics on people from adverse advents? maybe hard to get what might be viable is this information, last 20 cases that was resolved, similar drugs by warning labels, define similair, last 2 months what are the injuries that arent reported on the drug
not always under failure to warn, design defect, product recall(also triggered by fda), manfucature mistake, dangerous product design
also another algorithm we can introduce is a reccomendation of lawsuit and why
*Thread Reply:* @James Scott, totally. Just saw this and we read each others minds on that call.
Legal Matters Viability Index or something.
*Thread Reply:* That should be beside impact ranking I believe we should probably get a visualization going cuz tbsf would help me extract the data in the best way to fit inside what we wanna do
*Thread Reply:* Agreed. Hey, see Rob from Unicourts response. Can you get his new data into our system so we can flag that IBRANCE and others are MDLs, etc.
*Thread Reply:* He won't have them done until Monday. I'll reply in.
*Thread Reply:* Monday, we’ll meet and get our mind map diagram updated.
add in maybe pivot tbale link to documents
@James Scott love that idea, or LLM on those docs 🙂
Updated and we cna
yea i some great ideas for this might need to white board it but sounds good!
Look forward to it, awesome. Once you start layering in the new data, I'll get to building more prettiness.
The *FDA Drug API* (openFDA) provides access to a wealth of information regarding approved drugs, including *labeling information. However, warning labels* required for manufacturers, such as *boxed warnings* or *precautions, are included in the structured product labeling (SPL)*, which is part of the labeling information available through the API.
Key Information Found in the FDA Drug API:
- *Indications and Usage*: How the drug should be used.
- *Dosage and Administration*: Instructions on proper dosage.
- *Warnings and Precautions*: This section includes any warnings related to the drug, such as serious side effects, contraindications, and safety concerns.
- *Boxed Warnings*: The most serious warnings, often required by the FDA for drugs with significant risk.
- *Adverse Reactions*: Potential side effects and interactions with other substances.
- *Drug Interactions*: Information about how the drug may interact with other drugs.
How to Access Warning Labels:
- *Boxed Warnings* and other *safety precautions* are included in the data returned from the *FDA's Drug Label endpoint* in the openFDA API. This structured labeling contains the detailed product safety information that manufacturers are required to provide.
Example of a Warning Label in the API:
By querying the */drug/label endpoint*, you can retrieve sections like:
- *"boxed_warning"*
- *"warnings"*
- *"precautions"*
This provides the detailed text of the warnings the manufacturer is required to include.
Conclusion:
Yes, the FDA Drug API includes the warning labels required by the manufacturer, as part of the structured product labeling (SPL), which is available in the API under various sections such as "boxed_warning" or "warnings and precautions."
For more information, you can check out the FDA openAPI documentation.
@James Scott So our Drug Summary would be:
Drug Brand Name: Drug Generic Name: Manufacturer Name: Drug Description: Indications and Usage: How the drug should be used. Dosage and Administration: Instructions on proper dosage. Warnings and Precautions: This section includes any warnings related to the drug, such as serious side effects, contraindications, and safety concerns. Boxed Warnings: The most serious warnings, often required by the FDA for drugs with significant risk. Adverse Reactions: Potential side effects and interactions with other substances. Drug Interactions: Information about how the drug may interact with other drugs. Substances: Active Substances:
Taco stand https://search.app/rCAkRsg6y6gcmiFd7
are you going to the conference next week
You should, you will meet our many clients, see the full client to cash flow.
We have so much data to get, pipe and model it's going to be an amazing 2025 and beyond!
yea man im super excited i kinda want get this poc finished asap to share to the clients and cam the full view and see the thoughts of it all
Me too, we can hustle this week and next on all of it and see where we land. In Mondays meeting, lets have everyone drop off at end except you, me and Nick. We can talk James and Zack.
saw you cancled the meeting, we can have a small quick cathcup later]
Hey I stepped out can we push the call back for later ? I got the times mixed from from cst to est
Sure, lets do in the AM and keep it on-point and if you can have the data for Manufacturer, Drugs (with MDLs Docket number metadata) datasets put into DataStudio so we can get those pages / panes added, that will keep me progressing as you work on Legal Matters.
quick quesiton, when doing the summary of a partificular case file, what are some major points that should be in there?
Great question.
The parties, the plaintiffs complaint and motions. Any defendant responses. Any judgments from the courts. Case status
in a 4-5 sentence summary? i think the complaint is important in a summary but i can see what i can do i think with the link to the respective document if they wanted more informaiton they can just open it up with the next click.
these arent all the same file type, once we extract all the same ones like the compaint file then these answers would be more streamline
hey quick quesiton do we have anyone that oculd comb this for me? i need this but an example of there complain file and drugs or mdls that were classified as this as an example
maybe i can get it from chatgpt but this would be the recommendaiton algorthim completed i just need examples
to further enhance it from the complaint file
What do you need exactly from someone with legal knowledge?
well not necessarily legal knowledge, when a case is filed right i dont know which member or lead case they say we want to sue because of those meanings, i need a couple examples maybe 1-2 of each that gives the complaint type of failure to warn or manufacuring defects or xyz
like label it saying this was filed because of failure to warn
so we can say espeically with the non mdl cases, 80 percent of them are falure to warn or devepetive marking or xyz
Ah! You need an actual MDL that had a "failure to warn" ruling?
yes, because i can say to the llm extract the reason as to why this drug was filled and its because of xyz and then when cam or whoever views the dashboard metrics they can see that 60 percent is failure to warn, 10 is contimination etc etc thats a very important piece i think they would need to know along with the summary so they dont have to read everything its already there for them
i just wanted to give some examples like 1 or 2
Failure to Warn MDLS
- Vioxx (Merck) (MDL 1657) – a painkiller linked to heart attacks and strokes.
- Zantac (Ranitidine) MDL 2924
- Ozempic (MDL 3094)
- Nexium/Prilosec (Proton Pump Inhibitors) - MDL 2789: This MDL involves lawsuits related to the use of proton pump inhibitors (PPIs), such as Nexium and Prilosec, which are alleged to have caused kidney injuries. Plaintiffs claim the manufacturers did not adequately warn about the risks of chronic kidney disease and other complications(Maryland Personal Injury Lawyers)
- Xarelto (Rivaroxaban) - MDL 2592: This MDL consolidated lawsuits regarding the blood-thinning medication Xarelto. Plaintiffs alleged that the manufacturers failed to warn about the risk of uncontrollable bleeding, as Xarelto lacks an antidote to reverse its effects during emergency situations (Casetext - CoCounsel)
- Talcum Powder (Johnson & Johnson) - MDL 2738: Thousands of lawsuits were consolidated in this MDL, claiming that the use of Johnson & Johnson's talcum powder products caused ovarian cancer due to the presence of asbestos in the product. Plaintiffs argue that the company failed to provide adequate warnings (Casetext - CoCounsel)
I am aware of Zantac, Ozempic, Prilosec, Xarelto, Talc but used ChatGPT to get their descriptors.
James, morning. I am going to push our AI dashboard meeting to Monday, being in Vegas and helping the call center clean up their mess so the Finance team can send out accurate invoices that match our dashboards. This has been blowing up my week. Let me know your status through yesterday on the items from last week. 🙂
i did wanna catch up with u for a couple minutes finished both llms and deep diving in the mdl fata more, i have in our google staroages links to all the oducments and to provide on the dashboard as well
Did you see what he said ? llm won’t be able to do until larger subscription
Send him the list of Docket Numbers I gave you, then lets use all of the ones that are MDLs for the basis of our "demo data".
Once you tell me we have the Predicted MDLs list, we can triple check those drugs for MDLs to make sure we don't have false information in our algos
Sounds good and I sent them already the case numbers and stuff all sent over
You sent him the new MDL Docket #s I looked up in the sheet?
@James Scott, going to push our AI meeting to later in the afternoon. I have to model out call center pricing increases for Marc and Cam. 🙂
Ok I am getting things all in tables for me
@James Scott, please work on the dashboards. I got yet another "urgent investor request" and this is two days in a row of it blowing up my calendar
2 things im waiting on rob for, the pdf and updated drug model is all done, that can be added to the looker dashboard so people can click and see the drug labels as it is on the bottle with there ranking or updated ranking, the second thing is the data. i am waiting on rob to help me with some of the code cuz its not returning any information or any documents but other than that the first part but the details just need a day or 2 to get ficed then we will ahve everything all the llm models we need are done and need to be ran through the new pdfs once they are downloaded for summary and recommendaiton
Great. Did you get any data sources and panes added to dashboard of what we do have? Manufacturer financials, Drug Summary, etc?
Financially should already be in there, drug summary is a good field I can tweak the llm to add in to take the pdfs off I can add that
Yeah, add as much of the "story board" data for what we have into the DataStudio, we can organize, format and make pretty later.
Gengraf (cyclosporine) capsules are an immunosuppressant medication used to prevent organ rejection in kidney, liver, and heart transplant patients, as well as to treat severe rheumatoid arthritis and severe plaque psoriasis. The active ingredient cyclosporine works by inhibiting T-cell activation and proliferation. Common side effects include kidney problems, high blood pressure, tremors, and excessive hair growth. Gengraf capsules have increased bioavailability compared to Sandimmune cyclosporine capsules, so they are not interchangeable without physician supervision.
this is the summary aextractiong from the pdf files we have of drugs and will be included on the dashboard, running through all of them now
this is going ot be added in the ranking so its the drug name, pdf link summary and the new ranking metrics
running the llms, there will be an additional cost in aws account, shouldnt be to much though
I'll be on 10 minutes or so after start time, heading to cafe to work next session
Just parked, heading up the elevator and will be on shortly
joiing the case text demo ima ask about apis in python
i asked the question but basically its a legal llm that u can use to ask questions add databases to search and use your legal information, so i see this working perfectling with our litigation use cases and prove responses to how we can provide or what type of litigation we can use
Does this work in conjunction with UniCourt or in replace of?
depending on the api answer if i can query it could replace
I did I got an answer ima work with support cuz it wasn’t clear yes they work with it but it’s through an llm ? Idk how to leverage it
the demo i had would be perfect for the recommendaiton tool, since there llm is focused on mitigation and law we can create a database of our drug pdfs text etc etc feed it to that llm and then have a reccomdation catered toweards our date i think for like 250 bucks a month
Send me the dates you can be in LV in late December
I am flying back home tomorrow I can let u know prob in the next day or so! I talked with casetext they should be giving me a demo this week
Ok. Safe travels and send me job description tomorrow if you can.
feels good to be back on my main machine wokring
Send me that Data Engineer job description ASAP, meeting with Cam and Shellie shortly.
It doesn’t show on my slack for some reason
need a quick call with you asap when you get on
hey super quick question, how are we reimbursted for the tickets
James, DM @Brian Thiessen and ask, I'm not sure if he changed the Expense Reimbursement process.
And WOW, what a game last night. Holy Crap. Cam took the over at 53, talk about a gray hair inducing game.
i didnt add it in there yet in looker! and you mean the ravens game?>!?!
yea it was a good game! i been talking trash to my friends in cincinatti all day lol wish i woulda went to the game
Excuse my hangriness, I just needed to take a break and eat
ahahha no ur fine i understand lol sorry for the late response ima work on the map extracting the court location cuz seems like the non mdl cases are filed locally or that data is different than unicourt which is great for us so we can see where the people are
added the graph chart looks kinda good need to make it fancy the pages etup
Nice. I'll work on it some before we meet up later this week.
Is there anything ai ml related that we need to add or stats based we have recommendations all our summary’s trying to think of anything that will spice it up Rutter
Can you get a full bio/demo profile of the plaintiffs, and law firms reps and judges in case information broken out as separate fields? Maybe a lawsuit_actors table or something?
The who, what, where, when and how of these MDL-potential cases.
FYI, asked him to have her send her resume.
Sounds good I was just texting with Cory as well
Currently me and unicoirt is working through why the case id of the document form the court is different than there cuz if i can pull there information from u court the persona profile is easier idk any sites that let u search or an api to search people information maybe pacer since it’s court information I’m not to sure
Are we forwarded candidates and information cuz somone just contacted us regarding a role or is that the recruiter
@James Scott , she is a little direct for sure, but she is the mentor of James Turner. I think she might be worth a conversation for your data engineer position?
i can talk to her for sure and set something up, its hard looking at these resumes sometimes they dont usually show 2 things, 1. how was value added and 2. a true grasps of knowledge, and as a full stack software engineer im sure she can do some of the gcp stuff for sure, but its very python/data intensive in which i dont think we are looking for somone to come in and have a significant ramp up time
If you think she’s a good candidate, then we need Cameron to interview her next, then me. If you like her, we’ll loop in Shellie.
@James Scott, pushing our work session up a bit, need to get some lunch.
Grabbing Thai, be on in 10 minutes or so after start time
take a look at the dashboard this morning i fixed everything
Looking good !! Ahaha yes this is def it for the visualization peace
It’s looking more and more like a product 🥲
I need to spend a couple days going over the metrics and numbers
We need to get the understanding of what we want to show present with this dashboard some of the I would say hyperlinks reduces what we need to show if somone wants to know more information about the case than they can use the hyperlink, is the goal just to. Show or have them download the datasets as well
yeah, we need to check my graybox values and make sure they are correct
@James Scott Happy Thanksgiving. Thankful for you, the work you have done and the teamwork to get this project to this point. Have a great holiday!
you 2 ryan bneen meaning to catch up wtih you!!! happy holidays to you as well
good idea, so like other services for the dashboard and tiers, i was thinking we have liek all the information present through weblink so as an example when we serach for xelanz they wanna know whos on the case judges etc etc, and if they wanan see that they can click the link and great but we could charge extra for how all these other products if they want the backend tables to download the information
like the case url link has all the informaation a person needs to know about the case, but we can cahrge a higher tier subcription if they want to download that information in a table format, we can either use case text to display or unicourt
@James Scott, love that idea. Do we have the "Actors' Summaries" in DB fields to add to the dashboard?
And today, I talk to Cameron and will confirm the holiday schedule so we can book travel too for AI demo and holiday event.
Plaintiff - Age Range, Gender, Home City, Occupation, etc and Attorneys of Record (Whose on the Complaint) Defendant - Company, cleaned up to match our Manufacturers financials, FDA incidents and Attorneys of Record (whos on the Compalint)
I know it's in the summary PDF links, but if we can have fields summarizing this, then our dashboard without a URL click will tell us the "When, What, Where, Who, How" and you could even put the Plaintiff summary as a combined demographic-profile listing age-brackets (18 - 24, 25 - 34, 35 - 44, 45 - 54, 55 - 64, 60+, etc.)
this is deff harder than it sounds i have gotten started with this, but we can talk about it more for the update, i would like to show cam the current version of this this week, so we can gauge if we are on the tright track and if there is anything else needed to be added before the big meeting.
Sounds good, we can discuss in our weekly session.
any lawyers you can ask readily that what kind of document would contain plaintiff demographics?
Plaintiff Fact Sheets would have most of them, let me send you an example of some
You don’t have to join tomorrow that’s in case you had time
Do you want me to join? Or after you grill her, you can send her to Cameron to ensure she is a culture fit, a good "safety net" if she interviews one way and goes rogue. Cameron requires it too. then send her to me for the last interview. That way, we can draw her out, see how serious she is....once you are done, and want to continue, DM me, Cam, Shellie Mooney and let us know. Then we can loop her into interviews with Shellie for Cam and myself.
Sounds good with me ! And I don’t need you too
we never did get that pacer account did we?
He didn’t respond, do you want to at sign tag him and remind him
Let’s do it, I have a credit card I can use as well
Huddle me, I have a few minutes while doing update SQL for a revenue adjustment for Cameron.
*Thread Reply:* @James Scott
Shield Expense Card
4197460006979356
05/28 CVV 831
Cameron Rentch
5170 Badura Avenue
Las Vegas NV 89118
do we have a need for grpahic designers
my sister is head of grapahic design at the naacp and is looking to move
Oh, interesting. You can inquire with Chris Stevens, his team are contractors including himself and he could let us know how the workload versus resourcing looks, want me to introduce you? I've known Chris for 10 years and brought him to Shield.
can you put lead or senior in the job title for these positions
https://app.fireflies.ai/view/Data-Engineer-Discussion::jzd4kyof0yZtyiYa
Demographic Profile Template:
Personal Information: • Name: • Date of Birth: • Gender: (Male, Female, Non-Binary, Prefer not to say) • Marital Status: (Single, Married, Divorced, Separated, Widowed) Contact Information: Email Address and Phone Number.
Geographic Location: Country, State/Province, City/Town, and Zip Code.
Education: • Highest Level of Education Completed: (High School Diploma, Associate Degree, Bachelor's Degree, Master's Degree, Doctoral Degree) • Field of Study: Employment: • Current Employment Status: (Employed Full-time, Employed Part-time, Self-Employed, Unemployed, Retired) • Occupation: • Industry: • Annual Household Income: (Select range) Ethnicity and Race: • Ethnicity: (Please select all that apply, if applicable) • Race: (Please select all that apply, if applicable) Lifestyle: • Household Size: • Number of Children Under 18: • Primary Language Spoken at Home: • Do you have internet access at home?: (Yes, No) Optional Questions (depending on research focus): • Religious Affiliation: • Political Affiliation: • Health Status: • Hobbies and Interests: • Media Consumption Habits: (TV channels, online platforms, etc.) Important Considerations: • Adapt the template to fit your specific research needs: Depending on your study, you may need to add or remove categories and questions. • Consider cultural sensitivity: Be mindful of potentially sensitive topics and ensure your language is inclusive. • Clearly define response options: Provide clear and comprehensive answer choices for each question. • Ensure confidentiality: Clearly state how participant data will be protected and anonymized.
@James Scott, now that we have PACER access, can you use the MDL #s and use PACER to build the Parent Case & Member Cases for "existing MDLs" which will ensure our "predicted MDLs" list is accurate with no existing MDLs drugs on it.
the attorney profile is complete, we wont have them for every case, because it is highly dependant on finding the case id, however i did some reasearch and was able to find most of them, there will be or a need of manual effort for this if we want to match everything but i have i would say maybe 70 percent of them added
@James Scott, perfect. New table or additional column added to existing dataset?
Meaning, have we already added .drugmodel.unicourtmdl_attorney to our dashboard, if not, what is the FK back to the other tables? The Dashboard will have to have this added to the BLENDED / JOINED dataset too to ensure it works. Can you add it?
*Thread Reply:* the join for all of this would be brand_name
*Thread Reply:* or if we are talking about the specifc case
*Thread Reply:* @James Scott, can you add it into the Dashboard as Data Source and then JOIN in the Blended Data Source?
who do i talk about for reimbursement again? for the previous tickets
can we make the meeting early monday morning, i have to go out and pick up my sister from her appointment since she went under and cant drive
@James Scott, yes. And Marc and Cam said Tuesday, 17th, 3PM for AI demo. Book flight now for arriving Tuesday, leaving whenever you need. I'll go Tuesday, stay Wednesday and leave Thursday morning.
@James Scott , booking you from Monday the 16th to Wednesday check out. Let me j in now if different days are better.
did u resecdehule the google meeting for managers at 830?
yea iw as like its 11 here and the meeting was at 830 am ahah
And I'm am still having issues, so I blocked other people ability to change it, something weird was going on
looking at the flow of this dashboard i dont think the attorny profile kinda fits in this design, the attorny profile as at an individual case loookup level, while our dashboard main page is at the drug level, there can be a seperate page for those kind of specific which is done, except the bio profile.
the bio profile adding in synthetic or fake data i dont think would be accurate in this, at most i can grab is the sex of the person but other than that is just guessing, unless we specifically state this is infered froom the case then adding in whatever information is ok in that regard, but as a main or selling point i dont think it would be accurate
Can the instructions and labels also tell us who it’s for and not who for maybe to help synthesize it?
i think i can use a llm or the same llm add a new field and have a model build a profile on who would take this drug age range things like that
COMBING THROUGH the data, a lot of the information your looking for for patient igraphy we can pull from fda?
i added in a bio profile i can build about the people, there age weight male female and dieases but can you do what you did to link it to the fitlers
Those bios are linked via Drug Name you said earlier to the Blended Data?
Can you dump those Views into the page, I'll make pretty.
ok take a look i thikn you just need to make it pretty or the formatting but the data is there
awesome, will do soon. Doing the daily numbers crunch.
ok let me know when its done so i can check
Once we show Cam, I’ll get that stuff pretty, it’s been 3 days of fire drills from Cameron
Cameron said he was joining , you make it?
AI Dashboard Progress
Tuesday, Dec 10 · 12:30–1 PM
Google Meet joining info
Video call link: https://meet.google.com/nqd-xfwp-rnj
Or dial:
any more takeways from the prsentation with cam i think we are spot on for the most part
and let me knoiw when u have time for a quick call
Yeah, he was disappointed but I relayed your quandary
We'll do it like we did with Cam. I'll open, you give the play by play, I'll be color man. Cameron will be aligned with us.
Sounds good with me !!! Yes I wanted to be there I haven’t got a hold of him I messed him but no response I guess he was upset
Reping him, he never responds on first attempts for anything
@James Scott, check dashboard, new chrome from Cameron's logo.
and row 7 is our Drug Ranking is NULL, and a few of the drugs (#2,#3, etc.) do not have any 2023 data. Could you look to fix that?
And @James Scott, which of the 10 have a AI Ligitation Claim reco other than "Failure To Warn"?
NULL one is Oxbryta, put the name back in it, it's okay to show it "on the list" since this data is six months old now. It will give them some validation since we are doing Oxbryta.
there are different classifications of drug litgation
rinvoq is misleading marking maybe similar to deceptive might need to be adjusted
Leave it for now. I made them all 3 have conditioned formatting
i am trying to remove the null brand name but it doesnt seem to be wokring
trying to remove the nulls but its not wokring i filter it out in the query and the filer idk where its ocmuing from
I think you just removed oxbryta values from that one no
When you selected in the dashboard, it shows you that it’s that drug
0 is or nun is also a value, your count is pointing to case ur link so how many cases it has, but filted on it i do see data
which tells me its not being joined properly to the othyer tables or it doesnt exist in the other tables
i am going to exclude it for now until i can find the error in case i dont before the meeting
good job on the visuilizations i have nothing to add in there i wonder if we can incporate this in a webpage on our site, more traffic to our website would deff increase reveunue and income, idk how cam had it visioning
Well ask. We will rebuild this in Looker enterprise once we rig up the pipes and create pre-aggregate data flow.
I was trying to remove it from the dataset looker is being weird. If u look at the dashboard when we filter for that drug it removes the null, null is coming up in place of that name
Keep it and just update BigQuery Drug Name and Manufacturers values not to be NULL
If they are not NULL, we'll take a look at the DataStudio config, not sure why we'd have that issue unless a filter is sitting out.
i fixed it was actually simple, the problem was that looker was pulling from the caseurllink which didnt ahve that value because it didnt exist
Excellent, can can you clean up th Candidate Cases for the "bogus candidate cases rows".
It was good. It was 16 of us, all upper management at Summit Club, saw Mark Wahlberg's mansion
I'll call you on way to airport, we have a provisional patent meeting soon with our patent attorney.
Great news! Ok give me a call whenever
@James Scott, I'll call you before our patent call, been juggling stuff for past two day! Ready for the holiday break
thats on monday did u mean call today still?
Add "Case Research' and "Litigation Viability Rating" features.
@James Scott, also, add Integration Portal, discuss that it will connect into their Case Management software for learning about their specialities, precedence, private client profiles to further learn how to identify opportunities for litigations.
updated!! and i been a little sick from schirstimas this lignering cold is annoying
Feel better. All good. You can send it out to David and everyone on the email chain please.
does david like google docs? i am not sure lol
Just got off the phone with Corey is start date is the 13tg
Yea I will help him get started lay of the land as far as our techs tech goes
Send me a shield legal user name for you both, will setup repository
we havent had any unicourt usage for the last couple of months
Get casetext on the phone, lets find out about how we can use them instead of Unicourt, or if we need both at smaller license levels. Can you setup a CaseText demo meeting and play dumb, no bot-automation-adminttance of course.
same plan as yesterday give the intro and i can take the graphs?
And after you finish each graph, just ask me if I want to add anything.
i will ahve slack up in me other window, i can share screen for the demo
we did better on that one i am sure there will be much more ot come
Yes cam said they were talking a lot in the room so I guess he was able to handle the questions
from my end, i am going to update the dashboard and model with more realistic and accurate model predictions so in the next week or so we can show better results
yea josh is refreshing the data actually already done on his end, hes building out the scripts so they run automatically, went back to 2004 like they talked about,
Nice, using GCP BigQuery Python Notebook pipes or Cloud Function Python Notebook?
i think its docker and the big query workflows python notebokes
I had 12 hour day for Michigan Pollution price fix from a Cam oversight, fun
eww that sounds gross lol but yes im excited about the models so when we have our next presentation hopefully we can say these are actually prediction numbers and metrics that need to be tested
van you give me admin access in monday so i can get a to do loist or josh started
i know ur vacationing so whenever you have time!
For clarification we have been setting uk the pipelines in big query work flows, are u saying it needs to be in gcf ? Or five grab?
I’m cool with it being in the GCP system, I just want to make sure that Z can other people can support you guys on orchestration as you guys move onto other engineering and science science
Gotcha! Ok we can have a catch up call if u want since it’s been his first week or so2 with us
For sure. My Tuesdays have been bonkers, but I don’t get back from Vegas and back in my cockpit till Thursday. So if you wanna set something up for Thursday or Friday, that would work best.
That’s fine with me ! Are u still vacationing or are you working today
also, anyone good with monday? i need a tutorial so i can lay out josh task for him
we predicted dupixent was going to be number 1 for 2024 and it was indeed number 1 for 2024 lol
*Thread Reply:* Nice! Is the data refreshed in our MVP dash?
not in the dash, so its going to take a couple weeks to update the dash, the whole pipeline is getting productionalized so the new code scripts refinement, models and analysis, i had to help jos on the data engineer piece to run my model, and with the new models im making sure everything makes sense and are backed up, so when you and cam see the dashboard now its more realistic numbers to make analysis from as well as the pipelines is automatic in a sense of running hands off without intervention from anyone
as an example when you look at the dash and look at xeljanx and see in the anlytics dashboard seriosu injuries 203 and 2024 is seperated by millions, thats a model error, you wont see that anymore
so as im completed the scripts to run in sequence in google worflows, jos is productionlizing the code and automating it in github, and mapping the process in miro,
and combining all the diffrent data sources
Awesomeness, let me know when it’s ready. At that time, we can work on assigning a dashboard builder to start in Looker Enterprise. Zek, Brian or someone else under Nick.
i will ! let me know if you need anyhting in the meantime
also, i will keep the monday board updated with task and everything so you can see the progress.
Friendly reminder to get the sales call going with case text please
Did you still wanna have our catch up today or tomorrow
Ok I will put something in your calendar early or later ?
i think we have 5tran during that time
whats the aws account information again
yes i need to prod some of the llm models so they arent running in notebooks but i can call them through code for the prod data engineering
Console sign-in URL
https://960735831404.signin.aws.amazon.com/console User name
jamesscott@shield-legal.com Console password
nXK{V8-u Hide
I'll start and hand the table description for you.
I won’t be in dbt training I have An appointment at that time
any of the times work for you with casetext
about engineering i have a de that i used to help build my pipelines for my stock application been working with him for a month and would love to have him have a call with us
and probably less expensive than other people for sure
Have him send in his resume to me and Nick McFadden.
ok, i can have him send his resume i will get it for you
diseasae state, harard and odds ration
PubMed - Odds Ratio Metric at 1.8 or higher PubMed - Disease State
clincial studies report similar to canidate case details link
reference if any of the adverse events are in the studies
In PubMed, disease state refers to a patient's condition or the progression of a disease. Disease state can be used to evaluate treatment effectiveness, identify patients at risk, and improve diagnostics.
Disease state examples in PubMed: • Disease State Index (DSI) • A value that measures how similar a patient's data is to diagnosed cases of mild cognitive impairment (MCI). The DSI is used to identify patients who are progressing to Alzheimer's disease. • Disease State Fingerprint (DSF) • A visualization that summarizes patient data and shows how it contributes to a patient's disease state. The DSF is used to improve Alzheimer's diagnostics. • Disease State Management • A healthcare approach that integrates services from all areas of healthcare. Disease state management includes measuring indicators to evaluate treatment effectiveness. Disease state definition A disease state is an impairment of normal physiological function that can be caused by infection or stress. Disease states can produce characteristic symptoms and lead to illness or sickness.
ah you have your note taker going this is good we will need to and fda application umber should tie by back to studies
Good stuff, load these up in your Monday as new stories, and lets get PubMed on the phone for pricing @James Scott. Nice work.
that was a great call deff need that scrip
do you have time to meet 5-10 minutes befor eour meeting with cam
TBH, maybe we should cancel and wait on this, we have SOOOOOO many projects going, it's insane. Maybe wait 45 days and we meet. T'houghts @James Scott?
Yes that’s fine with me I thought it was going to assist you, if it’s not than we can cancel and put it on the backlog I can put it in our Monday
I could use your skills in B.I. before we unlease Media Strategy & Automation analytical services. Cameron JUST added ALL Leadspedia functionality into the HTFU project which was already a SOW for a Case Management Software, it's insane. We will build some Lead Management + Case Management + Communications Management + Integration Hub + god know what else. So having a new project would just crush everyone right now.
Would you be willing to help me get some basic JOIN sql ready for SL, Zek can be your hands, but I need help getting the pre-financial tables built in BigQuery, then I will train Zek on how to implement into dev, staging and production in DBT with your SQL?
If so, I can add you to the SL BI Priorities meeting coming up and get you ramped up.
Yea let’s take the time at 230 to talk about it !
@James Scott, kill the Cam meeting, and I added you to invite.
*Thread Reply:* @James Scott, just quick question on this data
We can use James Turner and TIP Legal Services as a way to get Cocounsel subscription and bot it like casetext if we have to with them probably not understanding Jason and pulling that, what a bad sales person.
yes we can do that, but it seems like casetext could be depreciated at anytime as well, so its hard to say
the current process works in the mean time
as an example we arent necessarily hitting the data that much anymore, the update frequency can be set to once a week or once a month, and we only would extract the most up to date cases we have a big historical list now
So yeah, we will need a cocounsel bot if UniCourt LLM isn’t ready.
yea i think so too, but even then im not convinved on cocounsel as well
What are some of the big law firms we work with
sent u a table with zeke, wanted to know if there was more informaiton needed than that, thats what james and nick put together idk if u seen it
I'll definitely take a look today after the morning meetings, thank you.
Crap, will you do it please. I’ve been on the road all morning, had to go take care of something at the lake.
no problem ! i thought i saw it on my calendar
Maybe double check, I might’ve done it quick and I’ve been on the road for the past four hours
Morning, in todays meeting. Boil your questions down to short and sweet questions for Abe. Just so they are not open ended. Then hopefully you get the answerws. I'll be quiet after the intros niceties.
https://pmc.ncbi.nlm.nih.gov/articles/PMC3380258/
added these 2 links as knowledges bases for llm understanding of what is a clincial trial and what are icd codes, the results are above, i am still tweaking it but getting good extraction results as well as relating the injury back to icd10 for a numbers comparison
this has been my test document or trial https://pubmed.ncbi.nlm.nih.gov/40015592/
yea there is a lot too account for/serach/add for the llm so adding it all in there but the foundaiton for it all is done
So we can kill this contract with UniCourt and ask them to demo their AI LLM API tool when ready, we may be interested. Aligned?
I was just about to say the same thing lol and yes I’m aligned
To join the video meeting, click this link: https://meet.google.com/cxw-gmpn-xxy
Otherwise, to join by phone, dial
we have a github account right we are switching our coding to vscode and i wanan use copilot
I will be a couple minutes late to the bi meeting
*Thread Reply:* @James Scott, Looker E style easily can do this
i wanna change the dashboard to look something like this
whens the meeting with cam which one is that
If they can show us the LLM product and it lets it do for us what we need, checking for the lack of cases then possibly if you haven’t gotten co-council account set up on the center side. I can help with that.
i can get that set up i think its fairly simple
@James Scott, just tell him possibly in the future. For now, we are keeping scope aligned with the partners
we can use this one dashboard for all our initatives
like if u look on the blue panel we can change the images to drugs food enviromental
yes i just need the template let me see if it can be switched over
Data studio and Looker Enterprise are completely separate products
did some research the looker studio dashboards can be applied to looker enterprise
Programmatically? Or just build them the way you want them.
That is a very good question let me look
want to use the company card for this and plug in my data to it https://lookerstudio.google.com/reporting/f1837c75-6dd4-4355-9966-ca1785d14302/page/4OrWB
4197460006979356
05/28 831
Cameron Rentch
5170 Badura Avenue
Las Vegas NV 89118
this one is going to look 100 time cooler
hiccup is he didnt copy my orginal code correctly so we are having to go back and understand where the discrepencies is
i forget how u did it but https://lookerstudio.google.com/reporting/3b963f3c-12e5-4e39-b940-10116892dab9/page/p_ijm0rntamd/edit
can u restore the previous version of ur dashbaord
i see it, i just dont have the permisions maybe to restore it, when i look on aother versions i do maybe cuz u created it im not sure
@James Scott, made you Owner role of the dash, so you can choose which version you want to revert.
yup restored it, lets have a quick walk through working session in the next day or so
*Thread Reply:* Can we do Monday? I know MMTP conference JUST wrapping up in Las Vegas, so Cam, Andy and crew are all playing catch up.
*Thread Reply:* thats fien with me, are u there with them
*Thread Reply:* I was supposed to be, but I left home early and I’m in Arkansas with my girlfriend
*Thread Reply:* I’ll be on the road tomorrow, today. I’m just fixing up some new dashboards and checking some null row days.
*Thread Reply:* the new girl or the old one u got so many nowadays lol
*Thread Reply:* She’s really a throw back woman. Wants to take care of her man.
*Thread Reply:* show me her when u get a chance congrats on that one ! i need to get like u when i grow up ahha
*Thread Reply:* Lol, you got plenty of time, my young friend
*Thread Reply:* u go man!! im bout to start taking some lessons from you lol
i have an doctors appoint at 230, i do wanna get with u today or tomorrow, i need help making the dashboar dpretty but other than that all the data is mostly in there ready to be used
on the card, i am going to need to put on an ai embeddings tool for one our models
Sounds good, what are you using the embedding tool for, to understand its prediction?
for this kind or project, we are storing the drug trial information in a vecstore so i dont have to constnatly run llms against the whole data, its kinda similar to a sql database and the embedding is turning the text to numbers for the llm to understand
Yep, that makes sense. And people use the inverse to turn LLM deep learning embeds into usable information?
if we wanted to develop our own llm, the embeddings layer created from this would be used in training
@James Scott, send me Editor mode to my medialab.agency email (that is same as shield-legal.com) and make me editor, I'll spend some time messing with some prettiness and visuals.
Hmmmm....maybe change the owner to me, see if that works?
I want to use GitHub code spaces for the coding infrastructure it’s in vs code which would help with things
What’s the python ide for the other teams
Ask @Nicholas McFadden in a DM ( he wont see this)
I can join I was on vacation I told ha
wanna have a talk about the dashboard with ya at some point me and nick are talking today about jose
hows the week been going see you on pto too lol
I took my boys to the mountains, you’ve been in Puerto Rico, did they lose power countrywide
No I am right next to the hospital so it’s the same power grid which is prioritized but a lot of the island did !
And I just went to the mall. Let me say, wow
hows the mountains what you do up there
We did hiking, bonfire on her land, lots of drinks and relaxing
Is that the lady u were talking to? And aww the kiddos !!
@James Scott, should we setup the meeting with Josh already?
I am trying to have him finish out this eeek with an end to end run to test if his pipeline works
What’s the goal here? To let him know covertly that we’ll talk about this next week?
how its just not checking stuff like none of this is that hard
if i sent the invite earlier i kinda expressed my unsatifaction but when u joined i didnt wanna slam into him again so i kept it light
We can push back the ai meeting from our talk yesterday with Shellie in transition to move things over to nick
James, Marc Barnazel was growling at me yesterday to get the updated algorithm changes document to David Velasco the attorney for the patent. Could you try to get that to me by end of the week, and send it out to the email chain with all four of us on it
did we ever get enterprise chatgpt or no
also, more so than that, i think im fine, i want to use the card fot github codespaces its just gituhub but able to run things so its easier to log and code me and nick was talking about it as an ide for the teams
Yes I been re running all the code since Josh left and doing it myself I am on step 8/10 8 is running now
Ok. Does Abe know, should I reply to email or did you ?
yes he said a couple days is fine i am expecting to have this out by tomorrow - friday, so then i am going to work on the dashboard with all the correct data, i might do the dashboard in a python app, i been working on it, its jhjust more conttrol and felibility on how exactly i want the data and things, it should look good
@James Scott It says I need to sign in, do you want to invite me to it?
Does the drug list include drugs not in FDA list but in clinical trials?
This is how we will share with team members going forward and they can access the pace
In that top list, are all those drugs required to be on the FDA approved drug list? Or some of them listed are only clinical trial drugs not on FDA approved list?
i have not made that comparative analyis at all, all i know and i assume that those drugs are what has been sent to the fda, i assume they are all aprpoved but i have no idea
@James Scott , Abe wanted the names of drugs that are just in clinical trials, so maybe we look to ensure that our model has those in there +ones that are approved on the FDA list
so the rank needs to be adjusted to only show drugs that are in clinical trials?
and once i have that list, then i rank?
I think the rank should have both, and maybe add a column that says approved or trials so Abe knows.
i see what your saying, so include a filter for drugs in trials, but keep it at everything
i just need to add in a filter to the table to say if its in trials yes or no
Exactly, Abe wants to be able to see the drugs that have those ratios in the trials only, or other clients may want some cases. And already in the approved list.
give me about till later in the day your really going to like these updates
James, I got slammed with validating Ed and Zek attempts at finally after six months of handholding, to close TIP and SL April 2025 end of month. Cameron is traveling to Germany this week. So tomorrow in our meeting, lets do the deep dive. I'll take a look before and let you know if I have any questions ahead of time. Could you get the attorney the updated patent documentation in th meantime?
yes yes sorry been wokring on this but yes
All good, I just know when Cam comes back he'll do a full status update request, so best to have this week be a freebie and get all stuff moving.
if you can look at the report, im uploading and updating all the backend data now
so we can search ffor that record specifically
take at al ook of the dashbord thats what ima be presenting
Not bad ! Just have to populate it with more data but I’m glad he sees the promise of it all with the flexibility now
I think it's definitely on the right track. His understanding of it was good, the use cases were endless and his input is valuable on some extra features.
I know lol I am really excited about tin
Now I gotta get cam to approve the c2c stuff lol drop him a hint that the presentation and eveyebrjnf is awesome
Idk when but cam and mark wanna see it
Once you have it ready with those changes so it’s “go time”. That work?
Yes it’s just adding in the data and stuff that’s all good but we can still show it to mark and cam this week
I’m PTO Wednesday to Monday this week, aim for next Wednesday
I’m on the plane, but can you send David the attorney the updated document for the patent before our meeting tomorrow?
we also going to show marc the new dashboard?
awesome should be great i am updating more data now and i think they are gong to like it, and the patent is also tomorrow too
1. Claim 1: This claim is fully supported by the current system and aligns with the core functionality of our impact prediction and ranking platform.
2. Partial Support with Exception • Claim 2: This claim is mostly supported; however, we would note the following exception: > The bullet point > “Generating ranked lists of drugs based on multi-objective optimization, wherein prediction confidence is quantified using an ensemble modeling approach that combines Bayesian networks and decision trees.” > is not currently implemented in this form. Our existing models do not employ this specific ensemble configuration or multi-objective optimization technique. 3 & 4 Clinical Trial Focus Update We are no longer reviewing case information at this stage and have shifted focus to clinical trial analysis. Please replace the original Claim 3 & Claim 4 with the following revised version:
Replacement Claim – Clinical Trial and Litigation Platform A multi-modal clinical trial analysis and litigation recommendation platform, comprising: • An API-driven integration module configured to securely ingest data from: • Clinical trial registries and FDA databases, • Electronic health records (EHRs), and • Legal case management systems related to pharmaceutical litigation. • A natural language processing (NLP) engine that extracts and summarizes clinical trial documentation and regulatory filings into structured data points, using a transformer-based model optimized for biomedical and legal terminology. • A recommendation engine configured to: • Analyze key trial factors—including adverse event frequencies, patient demographics, study design, and trial outcomes—to assess litigation relevance, and • Generate jurisdiction-specific litigation strategies based on clinical findings and historical precedent. • A query engine capable of: • Supporting multi-lingual searches with automatic translation for international trial and litigation data, and • Delivering dynamically updated legal guidance informed by real-time changes in regulatory and case law environments. • A feedback loop mechanism that refines litigation recommendations using reinforcement learning techniques, adapting outputs based on user selections, legal outcomes, and trial result patterns over time.
@James Scott, this looks good. Please send and reply all so Marc & Cam see we sent it. Excellent my friend.
let me know if you ahve soem free time today
You can call me now on my mobile, I’m on the road
i need to use our card for billing for chat gpt openai cost
James, here you go. Sorry, still on PTO
4197460006979356
05/28 831
Cameron Rentch
5170 Badura Avenue
Las Vegas NV 89118
Yes! Been catching up. I never am sent message to Cameron, I’ll do it this week. How’s the chatbot dev? And can you receive David’s patent email and send him reply today?
i have the chatbot done, its just like conteplating what i want it to do, i been thinking its easier to develop it to assist in search for drugs, but not having it do the whole drug universe, as an example, you can ask it what drugs pertain to coughing, it iwll give you a response, but if you ask it give me the number 23 rank drug, i dont want it to do that cuz the user can just see it
It should summarize the financial stats on the company, drug, the trial KPIs, the market impact on consumers (adverse reactions) and really give an executive summary along with your "legal action" recommendation from how I envisioned it for a guy like "Abe". Yeah?
i think we should have a call today on direction on this today
*Thread Reply:* Yes, for a bit. Waiting on an HR meeting.
*Thread Reply:* ok let me know when ur done
@James Scott, making a few road map slides for Andy, send me the Streamlit dashboard screenshot unless new one is ready
ASAP please, trying to keep this slide exercise under 2 hours. :-0
@James Scott, what is the data source provider for the clinical trial data again, slipped memory (getting old LOL)
do u have time today also to show the other page
where u in the meeting i didnt think u wre
Oh I didn’t know u were ok I can do that that if u have time now we can meet
chatbot is now attached to the database and can query and run information from it
@James Scott , excellent. Can you get me a link and a user for that new platform? And I’ll tell cameron to join the Wednesday meeting so he can see it for next steps. I appreciate the big push to get this thing redeveloped
oh thats not set up, let me try and do that, i am working out bugs and kinks with it right now
Excellent. I'll call you later today for an update on team stuff.
when you get on, help me set this up https://cloud.google.com/looker/docs/studio/conversational-analytics-looker-studio
@James Scott will call when I’m online today, sorry for the late response
@James Scott , Looker Enterprise has Conversations already in Beta, much better to use depending upon your data source. What sucks about them is they just use LLM to make SQL, so we need to see how it improves.
and just have the dashboard as the front layer
are you the admin on our main github account
can you give me admin privelages to the github dustin is being annoying
i thought i was on ther ebut i have another serpate giuthub enterprise? idk i can switch what i have over to that one
Dustin is annoying, sure. Send me your user handle
I'll catch him in Vegas and plan it for Tuesday or Wednesday.
Or I can bring you on the team meeting on Tuesday with Cam, Marc and Joe. That probably works best.
fyi product looks great ready to ship honestly i think i have to get with dustin on scale but now im just adding in and correcting some of the data but its great
@James Scott , excellent. Get me setup to play with it on our call tomorrow
did you want chris tackling this or did u want my input as well
I want you to review his code with him, make sure it does what it supposed to do...Let him know to include you in all meetings with CC and him.
in cams message, what does diagnosis - circulate mean
he means, send over screenshots of your software UI for Dupxient, he means "send me information from MediLens on this drug and circulate to this DM in Slack".