Enhancing Mental Health Treatment Prediction Using BXGBoost: A Hybrid Machine Learning Approach

Mental health disorders have become a significant global concern, affecting millions of individuals and straining healthcare systems. Accurate and timely prediction of appropriate treatment plans for mental health patients is crucial for improving clinical outcomes. Machine learning techniques, particularly ensemble methods, have demonstrated strong predictive capabilities in healthcare applications. This study proposes BXGBoost (Bagging XGBoost Classifier), an advanced ensemble learning approach that combines bagging and eXtreme Gradient Boosting (XGBoost) to enhance the prediction of optimal treatment for mental health patients. Unlike conventional machine learning models, BXGBoost leverages both the stability of bagging and the adaptive learning ability of boosting, mitigating overfitting while improving classification performance. The dataset consists of demographic information, medical history, treatment history, psychological test scores (e.g., PHQ-9 for depression, GAD-7 for anxiety), and patient-reported symptoms. The proposed BXGBoost model is evaluated against benchmark machine learning models, including traditional Decision Trees, K Nearest Neighbor, AdaBoost and standalone XGBoost. Performance metrics such as accuracy, precision, recall, F1-score, sensitivity, and specificity are computed to assess classification efficacy. Experimental results indicate that BXGBoost outperforms other models, achieving a classification accuracy of 82.47, demonstrating its superior ability to predict optimal mental health treatments.