Harsh2396/trelegy-copay-fraud-detection-v2

Trelegy Ellipta — Copay Fraud Detection with Isolation Forest

Product: Trelegy Ellipta (GSK) — Fluticasone furoate / Umeclidinium / Vilanterol
Method: Hybrid Rules + Isolation Forest + SHAP Explainability
Use Case: Detecting copay card fraud in pharmaceutical patient assistance programs

🏗️ Project Overview

This project implements an end-to-end copay fraud detection system for Trelegy Ellipta, GSK's blockbuster triple-combination inhaler (~$8B+ annual revenue). The system uses a hybrid approach combining hard business rules with unsupervised anomaly detection (Isolation Forest) and SHAP-based explainability.

Why Trelegy?

• High-value drug (~$600-700/month retail) → high copay card value per claim
• Chronic daily use → monthly refills create recurring fraud opportunities
• Fixed 30-day supply (exactly 30 blisters per inhaler) → easy to detect refill anomalies
• Multiple assistance programs → card stacking opportunities
• Even 2% fraud leakage on Trelegy = ~$160M/year in losses

📊 Fraud Types Detected

#	Fraud Type	Description	Detection Method
1	Early Refill Abuse	Refilling before 30-day supply runs out (< 23 days)	Rules + IF
2	Pharmacy Hopping	Filling at multiple pharmacies to stack copay cards	IF (behavioral)
3	Ghost Fills	Pharmacy bills but doesn't dispense the inhaler	IF (pharmacy patterns)
4	Prescriber Collusion	Single prescriber generating abnormal volume; wrong specialty	Rules + IF
5	Strength-Switch	Alternating between NDCs (100/62.5/25 ↔ 200/62.5/25) without clinical reason	Rules + IF
6	Government Insurance Abuse	Medicare/Medicaid patient using copay card (violates program terms)	Rules

🏛️ Architecture

Raw Copay Claims
      │
      ▼
┌─────────────────┐
│ Phase 1: Rules  │ ← 10 Trelegy-specific hard rules
│ (Hard Flags)    │   (early refill, impossible qty, govt insurance, etc.)
└────────┬────────┘
         ▼
┌─────────────────┐
│ Phase 2: IF     │ ← Isolation Forest trained on rule-clean data
│ (Anomaly Detect)│   (catches unknown/novel fraud patterns)
└────────┬────────┘
         ▼
┌─────────────────┐
│ Phase 3: Score  │ ← Combined priority = 50% IF + 30% rule severity + 20% rule flag
│ (Priority Rank) │   Risk tiers: Low / Medium / High / Critical
└────────┬────────┘
         ▼
┌─────────────────┐
│ Phase 4: SHAP   │ ← TreeExplainer → "Why was this claim flagged?"
│ (Explainability)│   Top features driving each anomaly score
└────────┬────────┘
         ▼
┌─────────────────┐
│ Phase 5: Eval   │ ← AUPRC, AUROC, F1, Precision@K, detection by fraud type
│ (Reporting)     │   + 5 publication-quality visualizations
└─────────────────┘

📁 Project Structure

trelegy_copay_fraud/
├── README.md                          # This file
├── run_all.py                         # 🚀 Master script — run this
├── generate_synthetic_data.py         # Synthetic Trelegy copay claims generator
├── feature_engineering.py             # Feature engineering pipeline (48 features)
├── fraud_detection_pipeline.py        # Full detection pipeline (rules + IF + SHAP + eval)
├── requirements.txt                   # Python dependencies
├── data/                              # Generated synthetic data
│   ├── trelegy_copay_claims.csv       # Main claims dataset
│   ├── patient_master.csv             # Patient demographics
│   ├── pharmacy_master.csv            # Pharmacy reference
│   └── prescriber_master.csv          # Prescriber reference
└── results/                           # Pipeline outputs
    ├── investigation_queue_top500.csv # Top 500 flagged claims for review
    ├── scored_claims_full.csv         # All claims with scores & risk tiers
    ├── feature_importance_shap.csv    # SHAP feature importance ranking
    ├── metrics.json                   # Evaluation metrics
    ├── 01_evaluation_metrics.png      # ROC, PR curve, confusion matrix
    ├── 02_shap_summary.png            # SHAP beeswarm plot
    ├── 03_shap_bar_importance.png     # SHAP bar chart
    ├── 04_detection_by_fraud_type.png # Detection rates by fraud type
    ├── 05_risk_tier_distribution.png  # Risk tier analysis
    └── model/
        ├── isolation_forest_model.pkl # Trained IF model
        ├── scaler.pkl                 # StandardScaler
        ├── encoder.pkl                # OrdinalEncoder
        └── feature_names.pkl          # Feature name list

🚀 Quick Start

# 1. Install dependencies
pip install -r requirements.txt

# 2. Run the full pipeline
python run_all.py

This will:

1. Generate ~50,000 synthetic Trelegy copay claims (with ~3% injected fraud)
2. Engineer 48 features (temporal, behavioral, rolling windows, Trelegy-specific)
3. Apply 10 hard business rules
4. Train Isolation Forest on rule-clean data
5. Compute SHAP explanations for top features
6. Generate evaluation metrics and 5 visualizations
7. Output an investigation queue (top 500 highest-priority claims)

📈 Features (48 Total)

Temporal Features

• days_between_fills — Days since last fill (normal: 28-33 for Trelegy)
• early_refill_flag — Binary: fill before day 23
• days_since_first_fill — Patient tenure in program
• claim_month, claim_dow — Seasonality

Rolling Window Aggregates (7d/30d/90d)

• patient_fill_count_{7,30,90}d — Fill velocity per patient
• patient_copay_spend_{7,30,90}d — Copay card spend per patient
• patient_total_claim_{7,30,90}d — Total claim amount per patient
• pharmacy_claim_count_{30,90}d — Volume per pharmacy
• prescriber_claim_count_{30,90}d — Volume per prescriber

Patient Behavioral

• unique_pharmacies_overall — Number of distinct pharmacies used
• unique_programs_per_patient — Card stacking indicator
• total_fills_per_patient — Lifetime fill count
• avg_days_between_fills — Average refill gap
• std_days_between_fills — Refill pattern consistency
• max_fills_any_30d — Peak fill velocity

Pharmacy & Prescriber

• pharmacy_claims_per_patient_ratio — Ghost fill indicator
• prescriber_specialty_valid — Specialty match flag
• prescriber_total_claims — Volume anomaly

Trelegy-Specific

• ndc_switch_flag — Strength switching (0173-0893 ↔ 0173-0887)
• govt_insurance_flag — Medicare/Medicaid with copay card
• cross_state_fill — Patient state ≠ pharmacy state
• new_patient_burst — >1 fill in first 7 days of enrollment

📐 Evaluation Metrics

Per literature best practices (arXiv:2312.13896, arXiv:2208.11904):

• AUPRC — Primary metric (best for imbalanced fraud data)
• Precision@K — Operational metric (investigator queue efficiency)
• F1-Score — Balance of precision and recall
• AUROC — Supplementary (can be misleading at <1% fraud rate)

🔧 Trelegy Product Details

Attribute	Value
Brand	Trelegy Ellipta
Manufacturer	GSK (GlaxoSmithKline)
Active Ingredients	Fluticasone furoate / Umeclidinium / Vilanterol
Indications	COPD (maintenance), Asthma (adults ≥18)
NDC (COPD/Asthma)	0173-0893-14 (100/62.5/25 mcg)
NDC (Asthma)	0173-0887-14 (200/62.5/25 mcg)
Days Supply	30 days (30 blisters, 1 inhalation/day)
Retail Price	~$600-700/month

📚 References

• Liu, Ting, Zhou — "Isolation Forest" (IEEE ICDM 2008) — Original algorithm
• Hariri et al. — "Extended Isolation Forest" (2019) — High-dimensional improvement
• Thimonier et al. — arXiv:2312.13896 — IF vs LightGBM fraud benchmark
• Amazon FDB — arXiv:2208.14417 — Feature engineering best practices
• ZS Associates — Pharma Copay Fraud Case Study
• KPMG — Fighting Copay Fraud (PDF)

⚖️ Disclaimer

This project uses synthetic data only. No real patient, pharmacy, or prescriber data is included. The synthetic data is designed to be realistic for development and testing purposes but does not represent actual GSK copay program transactions.

📄 License

MIT License