TRIALMAP: AN END-TO-END SOLUTION FOR MODERN CLINICAL TRIALS DESIGN WITH RWD AND AI
Author(s)
Yong Chen1, Bingyu Zhang, M.S.2, Yiwen Lu, B.S.2, Hua Xu, PhD3, David Asch, MD4;
1University of Pennsylvania, Professor, Philadelphia, PA, USA, 2University of Pennsylvania, Philadelphia, PA, USA, 3Yale University, New Haven, CT, USA, 4University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, USA
1University of Pennsylvania, Professor, Philadelphia, PA, USA, 2University of Pennsylvania, Philadelphia, PA, USA, 3Yale University, New Haven, CT, USA, 4University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, USA
OBJECTIVES: Eligibility criteria play a central role in clinical trial design, determining who can participate and how well results translate to real-world practice. Yet these decisions are often based on expert opinion or precedent, rather than systematic evaluation. We present TrialMap, a data-driven framework that leverages real-world data and artificial intelligence to evaluate and optimize eligibility strategies across competing priorities.
METHODS: TrialMap encodes eligibility criteria into machine-readable rules and systematically generates alternative eligibility pathways by relaxing selected criteria. For each pathway, it conducts debiased target trial emulation with negative control calibration and evaluates performance across six dimensions: efficacy, safety, feasibility, validity, generalizability, and efficiency. Multiplicity is addressed through Pareto front identification and SUCRA-based stability assessment.
RESULTS: Applied to 15 first-line oncology trials using Flatiron Health EHR data, TrialMap showed that original criteria retained only 18-43% of real-world patients. Multiple admissible pathways emerged, revealing substantial variation in trade-offs across objectives. Some relaxed designs preserved treatment effect estimates while expanding feasibility and generalizability.
CONCLUSIONS: TrialMap enables transparent, data-informed eligibility design, supporting more inclusive, reliable, and fit-for-purpose trials across diseases and trial phases.
METHODS: TrialMap encodes eligibility criteria into machine-readable rules and systematically generates alternative eligibility pathways by relaxing selected criteria. For each pathway, it conducts debiased target trial emulation with negative control calibration and evaluates performance across six dimensions: efficacy, safety, feasibility, validity, generalizability, and efficiency. Multiplicity is addressed through Pareto front identification and SUCRA-based stability assessment.
RESULTS: Applied to 15 first-line oncology trials using Flatiron Health EHR data, TrialMap showed that original criteria retained only 18-43% of real-world patients. Multiple admissible pathways emerged, revealing substantial variation in trade-offs across objectives. Some relaxed designs preserved treatment effect estimates while expanding feasibility and generalizability.
CONCLUSIONS: TrialMap enables transparent, data-informed eligibility design, supporting more inclusive, reliable, and fit-for-purpose trials across diseases and trial phases.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
RWD59
Topic
Real World Data & Information Systems
Disease
SDC: Cardiovascular Disorders (including MI, Stroke, Circulatory), SDC: Diabetes/Endocrine/Metabolic Disorders (including obesity), SDC: Oncology