How databricks-industry-solutions/medicare-risk-adjustment works
This repository is a Databricks solution accelerator for extracting Medicare HCC evidence from clinical notes and calculating patient RAF risk scores. Its principal workflow infers diagnosis codes and demographics, filters and organizes HCC evidence, combines it with patient status, and writes risk-score outputs. The sampled implementation is concentrated in one notebook; README describes the broader risk-adjustment purpose, while omitted notebook sections leave some intermediate wiring uncertain.
Main entry points: 01-medicare-risk-adjustment.py Notebook workflow
Components 15
Clinical NLP
- Notebook workflow
01-medicare-risk-adjustment.py - Disease extraction
01-medicare-risk-adjustment.py - ICD-10 resolution
01-medicare-risk-adjustment.py - HCC status
01-medicare-risk-adjustment.py
Patient Inference
- Age inference
01-medicare-risk-adjustment.py - Gender inference
01-medicare-risk-adjustment.py - Patient status
01-medicare-risk-adjustment.py - Patient records
01-medicare-risk-adjustment.py
Risk Adjustment
- RAF calculation
01-medicare-risk-adjustment.py - Risk profile
01-medicare-risk-adjustment.py
Data Outputs
- Delta tables
- HCC status output
01-medicare-risk-adjustment.py - Patient risk output
01-medicare-risk-adjustment.py
Other components
- Clinical notes
- Clinical NLP models
Connections 19
- Clinical notes → Notebook workflow: provides notes
- Notebook workflow → Disease extraction: extracts diseases
- Disease extraction → ICD-10 resolution: resolves codes
- Notebook workflow → Clinical NLP models: uses pretrained models
- ICD-10 resolution → HCC status: derives HCC status
- Notebook workflow → Age inference: infers age
- Notebook workflow → Gender inference: infers gender
- Notebook workflow → Patient status: creates mock status
- Notebook workflow → Patient records: groups patient data
- HCC status → Delta tables: writes silver data
- Delta tables → HCC status output: stores HCC status
- Notebook workflow → RAF calculation: calculates scores
- Age inference → RAF calculation: supplies age
- Gender inference → RAF calculation: supplies gender
- Patient status → RAF calculation: supplies status
- Patient records → RAF calculation: supplies diagnoses
- RAF calculation → Risk profile: produces risk scores
- Risk profile → Delta tables: writes gold data
- Delta tables → Patient risk output: stores risk scores