Applied AI and clinical data engineering — from concept to production.
Representative sample projects — replace with your own project details, metrics, and links.
Objective: Detect data anomalies in EDC extracts before medical review.
Results: 62% fewer undetected discrepancies; deployed across 14 Phase III studies.
Objective: Metadata-driven automation of raw-to-SDTM dataset mapping and define.xml generation.
Results: Mapping cycle reduced from ~3 weeks to 4 days; Pinnacle 21 clean on first pass for 90% of domains.
Objective: Assign MedDRA preferred terms to adverse event verbatims with human-in-the-loop review.
Results: 94.3% top-1 accuracy; coding throughput up 5×. Basis of granted patent.
Objective: Real-time key risk indicator dashboard for risk-based monitoring across sites.
Results: Predicts site quality issues ~2 visits earlier; adopted by 3 sponsor teams.
Objective: Ingest and standardize high-frequency sensor data into CDISC-conformant structures.
Results: Processes 2M+ readings/day with automated QC gates and audit trails.
Objective: Forecast trial recruitment timelines and recommend site activation strategies.
Results: Median forecast error under 9%; patent pending.