Predicting Mental Health Conditions Across Countries Using Lifestyle Factors: A Machine Learning Approach

Mental health disorders are a global concern, influenced by modifiable lifestyle factors such as exercise, sleep, diet, stress, and digital behavior. This study analyzes a dataset of 3,000 individuals across seven countries to predict mental health conditions using machine learning. A new “Free Time per Day” feature was engineered to capture worklife balance. We evaluated four models-logistic regression, random forest, gradient boosting, and neural network-on accuracy, F1-score, ROC-AUC, and other metrics. Gradient boosting performed best with$\sim 92 {\%}$accuracy and AUC of 0.95. Top predictors included stress level, subjective happiness, physical activity, and free time. Our results demonstrate the effectiveness of combining lifestyle data and machine learning for early mental health risk identification. The findings support data-driven strategies for preventive care and highlight the utility of lifestyle-based screening tools for clinical and public health use.