Readmission Classifier: Dataset Audit, Cost and Fairness

MediumClassical ML

Reported in interviews at OpenAI

Problem statement

How would you audit discharge data and build a readmission classifier for deployment? The file contains patient identifiers, dates, demographics, insurance, prior admissions, stay details, labs, medications and follow-up visits, with missing values and duplicate admissions. Predictions happen at discharge, readmissions are uncommon, and a small care team calls the highest-risk patients.

Audit leakage, build an evaluation pipeline and calculate the insurance-group recall gap.

Show hint

For every column, ask whether its value is known at the moment the model scores a patient.

Relevant roles: ML Engineer, Data Scientist and Applied Scientist.

Related free problems