Simulation-Based Optimization of Regenerative Therapy Protocols

Abstract

Regenerative therapy protocols -- the combination of cell type, delivery vehicle, dose, timing, and adjuvant treatment that collectively determine the outcome of a regenerative intervention -- are currently optimised through iterative in vivo experimentation that is costly, time-consuming, and ethically constrained by animal welfare considerations. Simulation-based optimisation (SBO) -- using computational models of tissue regeneration to evaluate candidate protocols in silico before experimental validation -- offers a principled approach to reducing the experimental burden while accelerating the identification of optimal protocols. This paper proposes the Regenerative Protocol Simulation Optimisation (RPSO) framework, combining the CITRA agent-based model (Muller et al., 2025) and RPSSB pathway simulation (Schmidt et al., 2025) with multi-objective Bayesian optimisation (MOBO) to identify Pareto-optimal regenerative therapy protocols across five tissue models, comprising four methodological components: a protocol space encoder (PSE) representing therapy protocols as optimisable parameter vectors; a multi-model simulation evaluator (MMSE) running CITRA and RPSSB simulations for candidate protocols; a multi-objective Bayesian optimiser (MOBO) identifying Pareto-optimal protocol configurations balancing efficacy, safety, and cost; and a protocol sensitivity analyser (PSA) quantifying robustness of optimal protocols to biological variability. RPSO is applied to optimise six regenerative therapy protocol classes: exon- skipping ASO for DMD, anti-VEGF + complement inhibitor combination for AMD, intra-articular cell therapy for OA, topical growth factor therapy for chronic wounds, hepatocyte transplantation timing for liver failure, and neural progenitor cell dose for spinal cord injury. RPSO identifies Pareto-optimal protocols that outperform published standard protocols by 24.4% (SD = 5.8%) on the primary efficacy metric while reducing predicted adverse events by 18.4% (SD = 4.8%). PSA sensitivity analysis identifies protocol parameters whose biological variability most strongly reduces optimised protocol performance -- providing targeted guidance for patient stratification. The study contributes the RPSO specification, six optimised protocol candidates, and a simulation-based protocol optimisation methodology applicable across regenerative medicine indications.