Weekly Formulary Monitoring Pipeline — Change Detection + Access Trajectory
A payer moved our drug from Tier 2 to Tier 3 on a Tuesday. The brand team found out the following Monday.
In those 6 days every physician who tried to prescribe hit a higher copay barrier. Some patients abandoned at the pharmacy. Those dispensed units are permanently lost.
Without weekly monitoring you are always 4 to 8 weeks behind.
- Ingests weekly formulary snapshots across 8 payers
- Detects every change — tier moves, PA additions, step therapy, quantity limits
- Classifies each change by type, direction, and severity
- Generates an actionable alert queue with specific actions required
- Tracks access trajectory using linear regression on access scores
- Exports change log, trajectory summary, and weekly scores to CSV
| Parameter | Value | Notes |
|---|---|---|
| Payers tracked | 8 | Commercial, Part D, Managed Medicaid |
| Simulation horizon | 12 weeks | Post-launch monitoring window |
| Trajectory window | 8 weeks | Rolling regression window |
| IMPROVING threshold | slope above 0.05 | Sustained upward trend |
| DETERIORATING threshold | slope below -0.05 | Sustained downward trend |
Total changes detected: 4 CRITICAL alerts: 0 HIGH alerts: 3 Deteriorating payers: 1 Most critical: Molina Week 8 — Tier 3 to Non-formulary
| Change | Direction | Severity |
|---|---|---|
| Tier 2 to Non-formulary | DETERIORATION | CRITICAL |
| Tier 1 to Tier 3 | DETERIORATION | CRITICAL |
| Tier 2 to Tier 3 | DETERIORATION | HIGH |
| PA added | DETERIORATION | HIGH |
| Step therapy 2+ agents | DETERIORATION | CRITICAL |
| Tier 3 to Tier 2 | IMPROVEMENT | MEDIUM |
| PA removed | IMPROVEMENT | HIGH |
git clone https://github.com/sannapa2016/formulary-access-tracker.git
cd formulary-access-tracker
pip install -r requirements.txt
pip install -e .
python main.pyformulary-access-tracker/ ├── src/ │ ├── init.py Makes src a Python package │ ├── formulary.py FormularyRecord dataclass and baseline payers │ ├── change_detector.py Change detection engine and severity classification │ └── trajectory.py 12-week simulation and access trajectory analysis ├── outputs/ CSV results generated on run ├── main.py Single entry point └── requirements.txt numpy, pandas
Siva Annapareddy Founder and AVP, Amrak Pharma Analytics 18 years in pharma commercial analytics
Project 5 of 36 — open-source pharma analytics portfolio