Driving Healthcare Operations with Data Science
The document discusses using data science to drive healthcare operations. It describes using models to close gaps in patient care by predicting which diabetic patients will develop complications in the next 6 months based on demographic data, medical history, medications and lab tests. The challenges are class imbalance, with few patients historically developing complications, and missing lab data. Gradient boosting decision trees are able to handle these issues better than logistic regression. Testing shows the model can identify high-risk patients to call with a 24% precision and 66% recall. A trial using the model to select patients for home visits found more complications than random selection, showing the approach can improve outcomes.
Driving Healthcare Operations with Data Science
- 1. Driving Healthcare Operations with Data Science
- 2. "literally a health insurance company"
- 3. "Operations" Clinical Operations ● Close member "gaps in care" ○ Not taking their meds ○ Not seeing their doctors ○ Not getting tested ● Document conditions Insurance Operations ● Approve / deny claims ● Approve / deny authorizations ● Catch fraud
- 4. "Operations" Clinical Operations ● Close member "gaps in care" ○ Not taking their meds ○ Not seeing their doctors ○ Not getting tested ● Document conditions Insurance Operations ● Approve / deny claims ● Approve / deny authorizations ● Catch fraud
- 5. E.g.
- 6. Talk to them about consequences of not doing so? Knock on the doors of most non-adherent members? Ask members politely? Use different messages for rich and poor members?
- 7. Enter
- 9. Data Science What should we do? For whom? Did it work? Enter
- 10. Case Study: Whom to Call for Home Visits?
- 12. Can we predict which of our diabetic members will have complications in the next 6 months?
- 13. Time Observation Interval Prediction Interval
- 14. Time Observation Interval Prediction Interval Demographic info, lab tests, medications, other diagnoses Diagnosed with diabetes complications?
- 15. Features Labels Member Age Hypertension hba1c CP001 65 Yes 6.5 CP002 77 No 8.3 CP002 84 Yes 7.4 Diagnosed with Complication in 6-month Interval Yes No Yes
- 16. Challenge: High Class Imbalance ● Historically, only 8% of diabetic members have been diagnosed with complications over a 6-month period. ● Easy to get "high" accuracy, but hard to get decent precision/recall tradeoff.
- 17. Approach: High Class Imbalance ● Evaluate using area under ROC curve. ● Empirically, tree ensemble models appear to handle the imbalance better than logistic regression.
- 18. Challenge: Missing Data ● Glycated hemoglobin clearly an important feature… but we only have measurements for ~60% of members. ● Whether we have a measurement correlates with both: ○ Diabetes complications. ○ How well a model trained without the lab measurement performs.
- 19. Approach: Missing Data ● Simply hardcode all missing values to something outside the measurement range. ○ In our case, 0.0. ● This way, tree models can split on "have a measurement" vs. "don't have a measurement".
- 20. Final Model: Gradient Boosting Tree Ensemble
- 22. Most Predictive Features Glycated Hemoglobin Age Hypertension Takes Insulin
- 24. Did it work?
- 25. Do we catch more complications if we make calls using the model?
- 26. Control Group Treatment Group
- 27. Control Group Treatment Group Call Group (Chosen at Random) Call Group (Chosen by Model)
- 28. Found Complications Didn't Find Complications Control Group 8 92 Treatment Group 24 76 FAKE RESULTS
- 29. Found Complications Didn't Find Complications Control Group 8 92 Treatment Group 24 76 FAKE RESULTS Chi-Squared Test