Dengue Cases Prediction and Analysis using LSTM and Region-specific Data

Dengue Fever presents a substantial global health challenge, necessitating the development of accurate predictive models for effective disease management. This research paper conducts a comprehensive analysis of a Long ShortTerm Memory (LSTM) neural network-based model for forecasting Dengue cases in the Philippines. The study includes data preprocessing, model development using TensorFlow and Keras, and thorough evaluation using metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), Mean Absolute Percentage Error (MAPE), and Rsquared (R2). The dataset, “Dengue Cases in the Philippines,” provides a detailed historical record of Dengue fever cases and fatalities across various regions, enabling a thorough analysis of temporal and spatial dynamics. Data preprocessing involves label encoding, normalization, and careful data splitting to ensure accurate model training and testing. The LSTM neural network architecture is optimized through hyperparameter tuning, including learning rate, batch size, and epochs, to improve prediction accuracy. Model performance is rigorously evaluated, with an MAE of 913, indicating areas for further improvement. While the model shows promise, there is room for refinement. This study contributes to the field by exploring the feasibility of LSTM models for Dengue prediction, with insights that may inform future research in disease forecasting and public health strategies. Limitations, ethical considerations, and recommendations for further work are discussed.