Study Setup
Configure your database, study period, methodology, and data model.
Future: i2b2, FHIR, PCORnet
How to structure patient rows
Generated script type
Study Definition
Define who enters the cohort, what outcome to predict, and optional exclusions or confounders. Each row is one piece of clinical evidence.
Cohort Entry
Who enters the study? Add one or more evidence rows (diagnosis, lab, drug, procedure).
Outcome
What event to predict in the follow-up window?
Exclusions (optional)
Patients matching any exclusion row are removed from the study.
Confounders (optional)
Risk factors tracked as binary columns in the output dataset.
Time Windows & Censoring
Define baseline and outcome window lengths.
How far back to look for patient history (e.g. 90 = 3 months, 365 = 1 year)
How far forward to look for outcome events (e.g. 180 = 6 months, 365 = 1 year)
Covariates (Features)
Select which features to include in your ML-ready dataset.
Review & Generate
Review your configuration. The self-check panel updates automatically.
Configuration Summary
Downloads a timestamped zip with study.sql, README.md, run.py, and optional artifacts