Enhancing Water Quality Predictions with Transformers for Schistosomiasis Management

Abstract

Water Quality Prediction (WQP) is crucial for managing water resources and preventing waterborne diseases such as schistosomiasis, which is particularly severe in Africa and other endemic regions. This paper presents a novel approach to WQP using a transformer-based model designed to predict key water quality parameters essential for controlling schistosomiasis. Utilizing a dataset from the LOCAR project, which includes extensive hydrological and physicochemical data from various catchments, our model significantly outperforms LSTM model. Specifically, it achieves up to a 74.31% improvement in performance, as measured by Mean Squared Error (MSE). This advancement highlights the model's superior accuracy and its potential to enhance decision-making and management strategies for water quality and disease prevention in high-risk areas. This progress is crucial for improving public health and water resource management globally.