Assessing the impact of climate and control interventions on spatio-temporal malaria dynamics using a stochastic metapopulation model

Despite intensive malaria control efforts, the lowlands of western Kenya continue to experience high malaria transmission. Spatial and temporal variations in climatic factors, interventions, parasite dispersal, and human travel, influence malaria incidence in moderate-to-high transmission areas. Additionally, population movement facilitates the importation of parasites from endemic to non-endemic areas, sustaining infections where local transmission would otherwise be unsustainable. The aim of this work was to develop a process-based stochastic metapopulation transmission model that accounts for key mechanisms of malaria dynamics, such as immunity, infectivity, and migration, while considering both the host and vector mobility. The model also incorporates and quantifies the effects of malaria interventions and climate variability at the local scale. Unlike existing models that often consider these drivers in isolation, our framework captures their joint influence within a single, mechanistic system. We show that, between 2008 and 2019, the developed metapopulation model accurately captured the effects of small-scale heterogeneity at the subpopulation level in western Kenya. Although demonstrated in a Kenyan context, the model is generalisable to other endemic regions and can support localized forecasting and intervention planning under future climate scenarios. Finaly, we assess its potential to forecast malaria incidence at the spatial-unit level, by integrating future climatic conditions with intervention scenarios.Author summary: Malaria remains a major public health challenge in the lowlands of western Kenya, where intense disease transmission persists despite ongoing control efforts. This persistence is driven by a complex inerplay of factors, including local climate and geography, population movements that spread the parasite between communities, and the varied impact of interventions like bed nets and indoor spraying. In this study, we developed a mathematical model to capture these diverse influences on malaria transmission. The model not only tracks parasite spread between humans and mosquitoes, but also accounts for communities interactions through travel and for the effects of local climate on mosquito development and disease risk. In addition, it incorporates key features of malaria dynamics, including human immunity, parasite transmissibility, and the impacts of control measures. We applied the model to data collected from 2008 to 2019 in western Kenya and found that it successfully reproduced complex, village-level variations in malaria cases. By taking into account climate trends and intervention coverage, the model can also forecast future malaria incidence under different scenarios.These findings underscore the importance of considering fine-scale heterogeneity and population movement when designing more effective, targeted strategies to reduce the malaria burden.