Dengue fever prediction using LSTM and integrated temporal-spatial attention: a case study of Malaysia

This paper presents an evaluation of Long Short-Term Memory (LSTM) models with integrated temporal-spatial attention mechanisms for prediction dengue fever in Malaysia. The performance of the models was assessed using Root Mean Square Error (RMSE) as the evaluation metric and compared with other LSTM variations and benchmark methods. The results indicate that the Spatial–Temporal Stacked LSTM (ST-SLSTM) model outperformed all other models, including the Spatial–Temporal LSTM (ST-LSTM) model, with a minimum RMSE of 1.94 for a lookback period of 5. Comparatively, the RMSE values of the LSTM and benchmark techniques like RBFSVM, DT, S-ANN, and D-ANN were higher, ranging from 4.37 to 5.58. These findings demonstrate the superiority of the proposed LSTM models with integrated attention mechanisms in dengue fever prediction. The attention mechanisms effectively capture temporal and spatial patterns, leading to enhanced predictive performance. The implications of this research are significant, offering potential improvements in resource allocation and timely interventions for managing and controlling dengue fever outbreaks. Future research directions include cross-validation in different geographical regions, incorporation of additional data sources, optimization of computational efficiency, and exploration of applications in other healthcare domains.

Text 124120.pdf - Published Version Restricted to Repository staff only Download (2MB) Official URL or Download Paper: https://link.springer.com/article/10.1007/s41324-0...

Actions (login required)

View Item