Spatio-Temporal Prediction of Leptospirosis in Kuantan Using Hydrometeorological Variables and Random Forest Machine Learning

Leptospirosis, a zoonotic disease prevalent in tropical regions with consistent rainfall, has been extensively studied using hydrometeorological data. This study focuses on developing a spatio-temporal model to predict leptospirosis in the Kuantan district, Pahang, known for its heavy rainfall and high disease incidence. Utilizing the random forest machine learning algorithm, we integrated hydrometeorological variables such as rainfall, streamflow, water level, relative humidity, and temperature across four model scenarios, lagging them from zero to 12 weeks at four-weeks intervals. Our models achieved an average testing accuracy of 73.4%, with sensitivity and specificity of 83.8% and 62.9%, respectively. Notably, we observed a minimal variation among the model scenarios, contrasting with previous studies where lag time improved the results. These findings underscore the potential of our models as a predictive tool for leptospirosis, enhancing spatial and temporal understanding in the Kuantan district. This improved insight can inform targeted disease prevention strategies, ultimately aiding in better management of leptospirosis outbreaks.