A Geostatistical and Machine Learning Approach to Modelling the Effects of Water Stress, Climate Change, and Sanitation Infrastructure on the Spatial Distribution of Infectious Diseases

Abstract

Infectious disease is becoming an increasingly important environmental issue, particularly in rapidly urbanising LMICs that lack the infrastructure needed to manage these challenges. It is important to map their spatial and temporal interactions for prevention and adaptation. We discuss novel advances in geostatistics and machine learning (ML) designed to map the spatial structure of these factors that influence the spatial pattern of infectious diseases; this includes core geostatistical methods such as kriging, variogram modelling, spatial regression, and autocorrelation analysis, as well as ML models such as random forests, convolutional neural networks (CNNs), and long short-term memory (LSTM) networks. The advantage of multimethod geostatistical-ML approaches over monolithic ones is that they achieve greater accuracy, interpretability, and uncertainty management. Case studies from 2020–2025 demonstrate that remote sensing, hydrologic and infrastructure data can be used to augment cholera, malaria and dengue models. Some of the challenges are data quality assurance, data interpretability, scalability, and privacy issues related to health data. Future priorities should focus on explainable AI, federated learning, and climate-health digital twins, which will help create resilient, secure, and future-proof models applicable globally. Finally, the integration of geostatistics and machine learning is a promising interdisciplinary approach to disease prediction and bolstering population resilience in a rapidly evolving world.