Investigating Data-Driven Climate-Adaptive Optimization of P. vivax Malaria Interventions in Seoul, the Republic of Korea

Plasmodium vivax malaria control requires addressing unique challenges such as latent hypnozoite reservoirs, relapse-driven persistence, and strong climatic modulation of transmission. This study introduces a novel, integrative modeling-optimization framework that couples a relapse- and climate-aware transmission model with structural identifiablity analysis and metaheuristic optimization to design adaptive intervention strategies. Using surveillance data from Seoul, Korea, we calibrated key parameters through identifiabilityguided estimation and optimized 48 monthly decision variables representing both pharmaceutical (tafenoquine substitution) and non-pharmaceutical interventions under resource constraints. The Improved Multi-Operator Differential Evolution (IMODE) algorithm efficiently navigated the high-dimensional, non-convex decision space, yielding climate-adaptive intervention schedules that reduced relapses by 80% and total infections by 16.6% relative to primaquineonly baselines. Economic analysis confirmed substantial cost-benefit (IBCR = 6.04), even under limited tafenoquine stockpiles. This work provides one of the first demonstrations of a structurally identifiable, climate-sensitive malaria model directly coupled to global evolutionary optimization, bridging mechanistic modeling with operational decision-making. The framework is broadly transferable to other vector-borne or climate-sensitive diseases, supporting data-informed elimination strategies in the face of environmental and logistical uncertainty. HighlightsO_LIClimate-adaptive P. vivax model with identifiability-guided optimization C_LIO_LIRobust parameter calibration of relapse dynamics using Seoul data C_LIO_LIIMODE optimizes 48 monthly interventions under resource constraints C_LIO_LI80% relapse reduction and strong cost-benefit (IBCR=6.04) achieved C_LIO_LITransferable framework for climate-sensitive disease elimination planning C_LI