Prediction of seasonal infectious diseases based on hybrid machine learning approach

Abstract

The effective management of seasonal dengue fever and other viral diseases fever such as malaria, pneumonia, and typhoid fever requires the early deployment of control measures. Predicting disease outbreaks accurately can aid in gaining control of epidemic seasons. In this research, a machine learning-based prediction model is suggested for predicting seasonality diseases. The model uses real-time data collected from different regions around Madurai district between 2019 and 2020, with 29 features including illness such as dengue, malaria, pneumonia, typhoid, kala-azar, Japanese encephalitis, measles, and normal fever and cold infections. The proposed model is a hybrid approach that includes feature selection using the Antlion Optimization Algorithm (ALO) and classification using Random Forest (RF) integrated with the XG-Boost technique. The suggested model’s efficiency is examined by accuracy, precision, recall, specificity, and f1-score as performance metrics, and compared with other models such as ACO-ANN, PSO-RF, WO-RF, and ANOVA-SVM. The suggested framework attained a high level of precision of 96.17%, a precision of 93.95%, a recall of 95.86%, a specificity of 93.23%, and an f1-score of 96.22%. Based on the comparison, the suggested model surpassed the eficiency of the alternative methods compared models in terms of all the parameters.