Guided Notes

Predictive Analytics in Healthcare

Prediction vs explanation, building a readmission or no-show model, train/test splits, evaluation (confusion matrix, sensitivity, PPV, AUC), calibration, fairness and deployment.

What predictive models do

Slide 1 / 5

Predictive models estimate the probability of a future event for each patient: • readmission within 30 days, no-show, deterioration, sepsis, high cost next year, missed refills. They help target limited resources (care managers, reminder calls) to the people most likely to benefit. They don't explain causes — they rank risk.

← Python & pandas for Healthcare Data