Methodology

Source → record → tier.

Every DrugIQ record passes through the same five stages: ingest from an authoritative source, normalize to the canonical schema, reconcile against other sources, assign a trust tier, and publish to the API. This page documents each stage, the conflict-resolution rules, the tier qualifying criteria, and our auditability commitments.

The pipeline

Ingest

Pull from authoritative sources at their published cadence (CDSCO MIS weekly, FDA monthly, NPPA on-publication, FAERS quarterly). Raw payload archived with timestamp and hash for audit replay.

Normalize

Map each source's native schema to the DrugIQ canonical schema. Resolve entity to canonical drug ID via deterministic match (brand + molecule + strength + formulation) with AI-assisted fallback for ambiguous cases.

Reconcile

Merge incoming source against existing record using precedence rules (below). Flag conflicts for review. Where sources disagree on a fact, both values retained with source attribution; canonical value selected per precedence.

Tier

Assign A1, A2, or A3 based on the qualifying criteria (below). Tier may move up (more sources confirm, expert review completes) or down (source retraction, conflict detected). Every tier change is logged.

Publish

Updated record available via API, web, and exports. Tier badge and source attribution travel with the record everywhere it is rendered or returned.

Conflict resolution precedence

Statutory pricing conflict

Most recent NPPA notification > NLEM 2022 baseline > DPCO 2013 ceiling > manufacturer MRP.

Identity / composition conflict

CDSCO licence record is canonical for Indian-marketed identity. Global cross-reference (FDA, EMA, PMDA) shown alongside when available; never overrides CDSCO.

Clinical guidance conflict

CPIC (drug-gene) > FDA label > ICMR > regional guidelines. Conflicts surfaced with both sources cited; never silently resolved.

Safety signal

Never overridden. All safety signals (FAERS, EudraVigilance, VigiBase) shown with source, signal score, and report volume. Aggregation is additive, not selective.

How a record earns its tier

All A2 criteria met, plus: a domain expert has reviewed and signed off on the record within the last 90 days. Expert review covers identity, composition, regulatory status, clinical claims, and source attribution.

All A3 criteria met, plus: at least one additional authoritative source confirms the canonical facts (identity, composition, regulatory status). AI consistency check passes across all sources. No unresolved conflicts.

At least one authoritative source. Record normalised to canonical schema. AI consistency check passes within the single source. No conflicting data from any other source examined.

Schema versioning

DrugIQ canonical schema follows semver.

Minor and patch releases are additive and non-breaking. Major releases (breaking changes) are announced 90 days ahead with migration guidance. Full schema changelog public.

Auditability

Every record change traceable to a source.

Raw source payloads archived with timestamp and hash. Tier changes logged. Audit log queryable via API or exported to your SIEM at the Enterprise tier.

Source → record → tier.

DrugIQ is in private preview.