Clinical Trial Patient Matching: A Real-Time, Common Data Model and Artificial Intelligence-Driven System for...

Area of Science:

• Oncology
• Medical Informatics
• Clinical Trial Management

Background:

• Cancer clinical trial enrollment is critically low (5%-7%) due to manual patient-trial matching bottlenecks.
• Existing AI/ML systems lack data standardization and cross-health system compatibility.

Purpose of the Study:

• Develop and validate a semiautomated Clinical Trial Patient Matching (CTPM) tool.
• Enhance the efficiency and scalability of clinical trial recruitment.

Main Methods:

• Created a hybrid rules-based and natural language processing (NLP) pipeline.
• Standardized electronic health record (EHR) data to the Observational Medical Outcomes Partnership (OMOP) common data model.
• Validated CTPM against manual chart review for a colorectal cancer trial, then implemented across 29 trials.

Main Results:

• CTPM achieved 94% retrospective and 88% prospective accuracy for a CRC trial, with 100% sensitivity.
• Reduced chart review workload 10-fold and screening time by 41%.
• Screened 98,348 patients across 29 trials, identifying 825 eligible candidates and facilitating 117 enrollments.

Conclusions:

• AI/NLP tool improves clinical trial recruitment efficiency by focusing research teams on qualified candidates.
• OMOP-based framework enables scalability across health systems.
• Addresses enrollment challenges, potentially increasing patient access to novel cancer therapies.