Existing Model
- IPF model
Local Model "installation"
Epic APP
Lets worry about this a bit later. Lets focus on making a SAAS model dployed for our ZCOR models firts
PreSAAS ZCoR Applioation
We want to make sure we can task the developer correctky. Lets make sure our current code is a pure python application that can run in a different place other than midway
Also we need to address the following points:
- what is the "rawest" data input we can have (maybe from Epic data pull?)
- make an example that reads data for a single patient and returns
- zcor score
- decision (at a tunable value of specificity or LR+)
- some shap values
- have option for calibration (we would need a small dataset for this, this is dersirable, but can we do without?)
- test what happens to our performance if we do not calibrate? **
