Hybrid Deep Learning Optimization for Predicting the Spread of Emerging Infectious Diseases

The accurate prediction of emerging infectious disease spread is essential for timely public health interventions, yet traditional models often fall short in capturing the complex spatial-temporal dynamics involved. This study introduces a Hybrid Deep Learning Optimization (HDLO) framework that integrates Convolutional Neural Networks (CNNs) for spatial feature extraction, Long Short-Term Memory (LSTM) networks for temporal sequence modeling, and Particle Swarm Optimization (PSO) for hyperparameter tuning. The framework processes heterogeneous data—including epidemiological records, environmental indicators, and mobility patterns—through preprocessing steps such as normalization, handling missing values, and class balancing using SMOTE. The CNN and LSTM outputs are fused and optimized by PSO to enhance predictive performance while maintaining computational efficiency. Evaluated on real-world datasets, including COVID-19 case trends, the HDLO achieved superior results with lower Mean Absolute Error (2.41), Root Mean Squared Error (3.85), and a higher F1-Score (0.91) compared to SEIR and standalone deep learning models. Visualization tools such as spatial heatmaps and trend graphs aid interpretability, and uncertainty intervals enhance decision-making. The HDLO framework demonstrates strong potential for dynamic, realtime forecasting of infectious diseases, offering a accurate, solution that sets a new benchmark for predictive modelling in epidemiology.