Malaria remains a significant public health challenge in sub-Saharan Africa, with Rwanda experiencing a dramatic increase in malaria cases despite substantial funding for control measures. This study investigates the impact of climate change on malaria transmission in Rwanda, using a case study approach and focusing on the spatiotemporal variability of malaria incidence. The research seeks to understand how changes in temperature and precipitation may alter the risk of malaria transmission, particularly in highland regions that currently have low incidence. This is crucial because existing malaria models often rely on linear growth assumptions, which may not accurately reflect the exponential population growth in Africa. The study's significance lies in its potential to inform evidence-based policies and control measures to mitigate future malaria risks under a warming climate, contributing to a broader understanding of climate change's impact on global malaria transmission patterns. The study's focus on Rwanda is particularly relevant due to its unique climate conditions, rapid population growth, and extensive experience with malaria control efforts, making it a valuable case study for other malaria-endemic countries. The increasing incidence of malaria in Rwanda after 2011, despite relatively stable funding for malaria control, indicates a need to explore other contributing factors, such as climate change, warranting this in-depth investigation.

Previous research has developed mathematical models to simulate mosquito populations and malaria transmission dynamics, driven by factors like precipitation and temperature. These models typically incorporate time-lag effects to account for the influence of climatic variables on mosquito life cycles. Forecasted changes in global malaria transmission patterns consistently highlight altered seasonal and spatial characteristics of malaria risk under various climate change scenarios. However, existing models often rely on linear growth assumptions, potentially underestimating the effect of exponential population growth, particularly in Africa, where the population is projected to experience significant growth. Existing research also points to the fundamental role of altitude in temperature variation and its impact on mosquito development and malaria incidence. Therefore, incorporating human population and elevation data into malaria projections is expected to enhance future assessments.

This study used a Random Forest Model (RFM) ensemble learning method to estimate the effects of climate change on malaria incidence in Rwanda between 2010 and 2015. Monthly malaria case data (2010-2015) were obtained from the Rwandan Health Management Information System (HMIS), and meteorological data (Tmin, Tmax, rainfall) for 30 districts were collected from the Rwanda Meteorological Agency. The RFM incorporated climate variables (current month and 1-3-month lags), geographic location, month of year, elevation, and total human population. Model accuracy was assessed using R-squared, RMSE, and MRE. Future malaria risk was projected under two Shared Socioeconomic Pathways (SSP2-4.5 and SSP5-8.5) using eight General Circulation Models (GCMs) from CMIP6. Climate data were downscaled using bilinear interpolation and compared changes in Tmax, Tmin, rainfall, and malaria incidence for 2030-2035, 2050-2055, and 2090-2095. Future population projections were integrated from SSP2 and SSP5 data. Statistical analyses, including linear regression, Pearson's correlation, and trend analysis, were performed. Partial dependence plots analyzed the complex spatiotemporal variability between factors and malaria incidence. The study's use of a 10-fold cross-validation technique and variable importance analysis adds robustness to the RFM results.

Spatial analysis of malaria incidence (2010-2015) revealed clustering in eastern and southern Rwanda (low elevations), while western and northern regions (higher elevations) experienced low incidence. This pattern highlights temperature as a primary driver of incidence, as opposed to precipitation. The RFM demonstrated high accuracy in predicting malaria incidence (R² = 0.79, RMSE = 0.64, MRE = 24.74%). Key determinants were human population and geographic factors, followed by elevation and temperature; rainfall showed a lagged effect, with a two-month lag showing the strongest influence. Future projections (SSP2-4.5 and SSP5-8.5) under both scenarios predicted significant increases in Tmax and Tmin, creating more favorable conditions for malaria transmission. The historical bimodal pattern of malaria incidence is projected to shift towards a single peak, particularly between February-April and August-October, potentially weakening seasonal fluctuations from January to June. Spatially, malaria incidence is predicted to increase in highland areas of western and northern Rwanda, with significant increases projected in districts like Muhanga, Nyamagabe, and Rubavu. The peak malaria season is projected to shift to earlier in the year at higher elevations (Western Rwanda) and later in the year at lower elevations (Southeastern Rwanda). The duration of malaria transmission is predicted to lengthen, with some districts potentially becoming year-round endemic areas.

The findings strongly suggest a significant influence of climate change on malaria transmission in Rwanda, particularly in previously low-transmission highland regions. Increased temperature and precipitation create more suitable conditions for mosquito breeding and Plasmodium development at higher altitudes. The projected shift toward a single-peak malaria transmission season weakens the seasonal predictability of outbreaks, complicating control efforts. The study's results corroborate previous findings showing climate change exacerbation of malaria transmission in East African highlands. The identification of specific districts experiencing significant increases in malaria incidence provides crucial information for targeted interventions. These findings are not only relevant to Rwanda but can also inform malaria control strategies in other regions with similar climatic and topographic conditions. However, it's crucial to consider that factors such as public health infrastructure, drug resistance, and population dynamics also significantly influence malaria transmission dynamics. The results suggest a need for a shift in malaria control strategies, moving from a focus on specific seasons to year-round measures.

This study provides compelling evidence of the increasing malaria transmission risk in Rwanda's highlands due to climate change. The projected increases in temperature and precipitation, coupled with the shifts in seasonal patterns and spatial distribution of malaria incidence, highlight the urgent need for adaptive measures. Future research should incorporate data on mosquito resistance and malaria interventions to refine predictive accuracy and provide more robust recommendations for malaria elimination. The development of global malaria trend projection models using comprehensive district-level data is a critical next step.

This study has some limitations. First, to enhance accuracy, it's important to integrate quantifiable indicators related to mosquito resistance and malaria interventions into the model. Second, including longitudinal data on insecticide resistance, malaria funding allocation, and ITN coverage would improve the model's predictive capabilities. Finally, building comprehensive global malaria trend models requires extensive district-level health and climate data, which are currently limited in many African countries. These limitations highlight avenues for future research.