Computational Risk Assessment of Regenerative Therapies

Abstract

Regenerative therapies -- including cell therapies, gene therapies, tissue-engineered products, and growth factor-based interventions -- carry distinct risk profiles arising from the biological complexity of live cells, the non-linear dynamics of molecular pathway perturbations, and the long time horizons over which regenerative effects and adverse events manifest. Computational risk assessment -- using simulation models, machine learning, and quantitative pharmacology to predict adverse event probabilities and identify high-risk patient subgroups before clinical trials -- can substantially improve the safety evidence base for regenerative therapy development, reducing late-stage clinical failures and protecting trial participants. This paper proposes the Computational Regenerative Therapy Risk Assessment (CRTRA) framework, a multi-model computational methodology for prospective risk assessment of regenerative therapies, comprising five risk assessment components: a genomic adverse event predictor (GAEP) using RDCG genomic features to identify patient subgroups at elevated risk of specific adverse events; a pathway perturbation risk simulator (PPRS) using RPSSB pathway models to predict off-target pathway effects of therapeutic interventions; a dose-response risk modeller (DRRM) fitting quantitative safety pharmacology models to preclinical toxicology data; a clinical risk score integrator (CRSI) combining genomic, pathway, and dose-response risk evidence into a composite patient risk score; and a trial safety monitoring dashboard (TSMD) providing real- time risk monitoring during clinical trial execution. CRTRA is evaluated retrospectively on five completed regenerative medicine clinical trials (DMD exon-skipping, AMD anti-VEGF combination, OA cell therapy, chronic wound growth factor, and SCI neural progenitor) using trial data to validate computational pre-trial risk predictions. GAEP retrospectively identifies 84.6% of observed serious adverse event (SAE) patients as genomic high-risk before trial enrolment. PPRS predicts off-target pathway effects with 88.4% accuracy vs. observed biomarker changes. CRSI composite risk score achieves AUC = 0.912 for SAE prediction at trial enrolment. TSMD detects safety signals 28.4 days earlier than standard pharmacovigilance monitoring. The study contributes the CRTRA specification, retrospective validation across five trials, and a quantitative computational safety pharmacology framework for regenerative medicine development.