Application of Machine Learning to identify factors affecting prevalence of Brucella: A Central India Study

India is an agricultural country, and a large portion of the population is regularly exposed to animal rearing, which increases the risk of zoonotic diseases. The Covid-19 pandemic is a recent and deadly example of zoonotic disease transmission, according to the World Health Organization. In the past three decades, multiple infectious diseases have been transmitted between animals and humans. One such bacterial infection is Brucellosis, caused by factors such as exposure to animals, consumption of raw milk, dairy products, and meat, along with certain lifestyle habits. In this study, a cross-sectional surveillance-based investigation was conducted to assess the prevalence of zoonotic infections in the Central Indian population. The study focused on three states: Maharashtra, Madhya Pradesh, and Chhattisgarh, where 7026 samples were recruited under strict inclusion and exclusion criteria. Data on participants’ lifestyle, dietary habits, and animal exposure were collected through structured questionnaires, followed by serum testing for Brucella antibodies using the Novatec IgM detection kit, with a sensitivity and specificity of over 95%. To analyse the dataset, four machine learning classifiers were employed: Random Forest, Logistic Regression, Support Vector Classifier (SVC), and Decision Tree. The classifiers were applied to the data both before and after handling data imbalance and encoding categorical variables. The performance of the models was evaluated using accuracy, precision, recall, and F1-score as metrics. Among the models, Random Forest achieved the highest accuracy of 99.11%, followed by SVC with 96.47% accuracy. The feature importance analysis revealed that animal exposure and dietary habits were the most significant predictors of Brucellosis risk. This study demonstrates the potential of machine learning techniques in predicting zoonotic infections such as Brucellosis by integrating lifestyle and clinical data. The results indicate that combining epidemiological data with advanced predictive models can improve early detection and intervention strategies for zoonotic diseases in high-risk populations.