Back to Products & Projects
AI/MLProduct StrategyClinical DataMonetisation
Drug-Disease-Lab Cross-Sell Mapping Engine
- •Led the product scoping and delivery of a drug-disease-lab mapping database — structuring the clinical relationships between drug indications, disease conditions, and recommended diagnostic tests.
- •The engine powered Tata 1MG's cross-sell recommendation logic across pharma and diagnostics.
- •Collaborated with the medical affairs, diagnostics, and recommendations teams.
The Challenge
- No structured drug-disease-lab mapping existed internally — recommendation logic was based on rudimentary keyword matching.
- Clinical mapping required medical affairs validation.
- The mapping needed to be granular enough to distinguish between drugs used for the same condition with different diagnostic monitoring requirements.
The Approach
- Structured a three-entity mapping schema: Drug (indication-level) to Disease (ICD code-aligned) to Diagnostic Test (LOINC-aligned), enabling multi-directional clinical inference.
- Used AI-assisted extraction to generate initial mapping candidates from drug label indications and clinical guidelines.
- Built the mapping database in phases — prioritising the top 100 drugs by GMV contribution.
- Integrated the mapping output into the recommendation engine and diagnostics cross-sell modules.
Results
Drug-disease-lab mapping live across top GMV drug categories in Phase 1.
Cross-sell conversion from drug PDPs to diagnostics improved following recommendation engine integration.
Recommendation relevance score (internal metric) improved vs pre-mapping baseline.
Tech Stack
PythonLLM (AI-assisted extraction)LOINC databaseICD-10CMSRecommendation engine API