i.
AI in drug discovery
Model risk and reproducibility for AI in target validation, lead optimisation, and assay-design support. The reasoning trace is part of the lab notebook, not a sidecar.
- Reproducibility frameworks for discovery AI
- Knowledge-graph retrieval with citation discipline
- Reasoning-trace as primary record
ii.
Clinical-trial AI governance
Bias, consent, audit. Trial-data hygiene with row-level lineage. Every classification reversible on an auditor's clock.
- Trial-data hygiene with row-level lineage
- Bias monitoring across population strata
- Consent flow built into the system
iii.
Pharmacovigilance AI
Adverse-event signal detection, AI-driven causality assessment, ICSR triage with reversible classification. Built so a regulator can reconstruct the AI's contribution to any decision, in writing, at any later date.
- ICSR triage with reversible classification
- Signal detection across MedDRA hierarchies
- Audit-ready provenance ledgers
iv.
FDA / CE compliance
AI-enabled tools for regulatory drafting, CMC submissions, and software-as-a-medical-device. Export-ready governance from prototype, classification mapping aligned to the strictest market.
- CMC submission copilots, agency-aligned
- Software-as-a-medical-device classification
- Deficiency-letter response automation