A Web-Based Application for Real-Time Malaria Prediction using Environmental Variables

Malaria remains a persistent public health challenge in Zimbabwe, particularly in rural districts such as Mudzi in Mashonaland East Province, where seasonal transmission and limited healthcare access undermine conventional control measures. This study presents the development and deployment of MalariaDash, a web-based application designed for real-time forecasting of malaria case counts using satellite-derived environmental data. The system integrates remote sensing inputs such as land surface temperature, vegetation indices, elevation, and proximity to water bodies with a Random Forest regression model trained on twenty-one months of health facility data. Predictor variables were selected based on their statistical significance in earlier modelling efforts. Unlike existing studies that focus on retrospective risk mapping or static models, MalariaDash operationalises machine learning outputs into an interactive platform. The application dynamically retrieves environmental data through Sentinel Hub and weather APIs, enabling location-specific and time-specific malaria predictions. It is implemented using the Django web framework and provides a user interface where environmental conditions and predicted malaria incidence can be viewed for selected dates and areas. This research demonstrates the practical value of linking environmental intelligence with predictive analytics in a rural African context. By delivering spatially explicit and near-real-time forecasts, MalariaDash enables health authorities to adopt proactive, targeted interventions rather than reactive responses. The approach illustrates a novel integration of geospatial modelling, machine learning, and operational web deployment aimed at improving local malaria control strategies. Author SummaryIn many rural parts of Zimbabwe, malaria continues to affect thousands of people every year. Although we know that changes in temperature, rainfall, and vegetation can influence where and when malaria outbreaks occur, local health workers often lack tools to make use of this information. In this study, we developed a web-based tool called MalariaDash that helps predict malaria cases using freely available environmental data from satellites. By combining these environmental signals with past malaria records, we created a system that allows users to select a location on a map and get a real-time forecast of malaria risk. What makes this tool different is that it is not just a research model, but it is designed for practical use by local health teams. We tested it in Mudzi District, a rural area with a high malaria burden in Zimbabwe, and the system performed well in identifying high-risk times and places. Our hope is that this approach can help health workers respond earlier and plan better, especially in areas where resources are limited and time matters most.