You cannot select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
42 lines
1.8 KiB
Plaintext
42 lines
1.8 KiB
Plaintext
How do I begin work on stuff
|
|
|
|
- next step is causal story. key points include
|
|
- we are trying to separate strategic and operational concerns. (why is this a difficult problem?)
|
|
- we can't trust what we are told
|
|
- terminations could be due to safety, strategic, or operational concerns.
|
|
- explaining confounding between
|
|
-population/market and market conditions.
|
|
- market conditions and enrollment.
|
|
- population/market and enrollment.
|
|
- describe other confounders
|
|
- safety and effectiveness
|
|
- duration <--> enrollment/termination
|
|
- Condition
|
|
- Decision to procede with Phase III trial
|
|
- How do I handle this?
|
|
- Introduce Do-Calculus
|
|
- DAG model
|
|
- What do I need to control for, in some form or other?
|
|
- Introduce Data
|
|
- Clinical Trial Progression
|
|
- AACT gives us information on
|
|
- terminated/completed status
|
|
- compound-indication pairs
|
|
- MeSH/RxNorm links
|
|
- Snapshots
|
|
- Market Conditions
|
|
- can't directly measure alternate treatments/standards of care.
|
|
- Can get measures of USP - formulary alternatives
|
|
- Can get number of generics or brand names with same drug.
|
|
- Population Sizes
|
|
- IHME Global Burden of Disease dataset. Best measure of impact of a given disease category.
|
|
- DALY's
|
|
- How much data do I have?
|
|
- Econometric model
|
|
- for a given state, what is the probability it will terminate?
|
|
- more accurately for my dist-diff analysis: for a given state, what is the distribution of the probabilities it will terminate?
|
|
- basic bernoulli-logistic model, linear in parameters.
|
|
- What are the specific things I am looking at?
|
|
- number of competing treatments.
|
|
- delaying close of enrollment.
|