Transforming healthcare claims into actionable intervention strategies through risk stratification, clinical analytics, and cost concentration modeling.
Healthcare claims are not normally distributed.
A relatively small group of members consistently generates a disproportionate share of healthcare expenditure, yet traditional utilization reports often hide this pattern behind portfolio averages.
This project develops a risk stratification framework that combines claim utilization, financial severity, chronic disease burden, and clinical complexity into actionable intervention tiers for population health management.
Across 554 insured members, the framework identifies the members driving financial risk, recommends targeted intervention strategies, and estimates potential cost reduction through structured care management.
Instead of asking
"How much did we spend?"
this project asks
- Which members generate most healthcare expenditure?
- Which clinical conditions predict future cost?
- Which providers should be reviewed?
- Which members require immediate intervention?
- How should intervention resources be prioritized?
- What financial impact can be expected?
| Metric | Result |
|---|---|
| Population | 554 Members |
| Total Claims | 2,064 Claims |
| Annual Spend | Rp 4.34 Billion |
| High Risk Population | 19% |
| Spend Concentration | 53.2% |
| High Cost Population | 37% generate 62.5% of spend |
| Estimated Annual Saving | Rp 520M+ |
| Estimated ROI | 3 : 1 |
The framework demonstrates that targeted intervention is substantially more effective than portfolio-wide programs.
Key operational recommendations include
- Dedicated case management for Tier 1 members
- Chronic disease outreach for high-risk members
- Provider network optimization
- Clinical referral management
- Precision-based intervention prioritization
- Longitudinal population monitoring
Claims Data
│
▼
Data Cleaning
│
▼
Feature Engineering
│
▼
Composite Risk Scoring
│
▼
Clinical Classification
│
▼
Risk Tier Segmentation
│
▼
Provider Analysis
│
▼
Intervention Roadmap
The composite score combines
- Annual Claims Paid
- Claim Frequency
- Inpatient Admission
- Charlson Comorbidity Index
- Chronic Disease Flags
Risk Score
↓
Tier 1 — High Risk
Tier 2 — High Utilizer
Tier 3 — Shock Claim
Tier 4 — Standard Population
The framework incorporates
- ICD-10 diagnosis mapping
- Charlson Comorbidity Index (CCI)
- Chronic disease identification
- High-cost inpatient utilization
- Benefit utilization analysis
Provider evaluation considers
- Claim volume
- Total paid amount
- Cost concentration
- Geographic utilization
- Referral optimization opportunities
Classification performance is evaluated through
- Precision
- Recall
- F1 Score
- Threshold Sensitivity
Rather than maximizing accuracy alone, the operating threshold balances intervention capacity against missed high-risk members.
| Category | Technology |
|---|---|
| Analysis | Python |
| Data Processing | Pandas |
| Statistics | NumPy |
| Visualization | Plotly |
| Mapping | Leaflet |
| Frontend | HTML / CSS / JavaScript |
| Dataset | JSON |
Healthcare-Risk-Stratification/
│
├── Showcase/
│ ├── Risk_Stratification.pdf
│ ├── Presenting Risk Startification.html
│ ├── carousel_presentation.html
│ └── data.js
│
├── images/
│ ├── overview.png
│ ├── risk-matrix.png
│ ├── diagnosis.png
│ └── provider-network.png
│
└── README.md
The proposed intervention strategy is expected to achieve
- 12–18% reduction in annual healthcare expenditure
- Rp 520M+ projected annual savings
- Estimated intervention ROI of 3:1
- Improved prioritization of clinical outreach
- Reduced unnecessary inpatient utilization
- Better provider negotiation opportunities
This project represents a portfolio case study.
Current limitations include
- Single employer population
- One-year observation period
- No pharmacy claims
- No laboratory measurements
- No mortality outcomes
- No real-time claim ingestion
Planned extensions include
- Dynamic Risk Score
- Time-to-event (Survival Analysis)
- Readmission Prediction
- Provider Quality Index
- Explainable AI (SHAP)
- Automated Risk Monitoring
- Real-time Dashboard
- Care Management Tracking
Interactive Demonstration
👉 Presenting Risk Stratification
Executive Presentation
Healthcare Analytics · Data Analytics · Insurance Intelligence
Data creates value only when it changes decisions. This project demonstrates how healthcare analytics can translate claims data into measurable operational action.