The project aims to use Machine Learning techniques to predict depression and suicide risk among Nepalese individuals using social media activity. Using Naïve Bayes, Logistic Regression, and Support Vector Machine algorithms, a dataset of 2200 entries was analyzed. The Support Vector Machine achieved an accuracy of 95.45%. Early detection of suicidal and depressed rates was crucial for mitigating depression and reducing suicide rates. The study evaluates different Machine Learning methods sensitivity and accuracy for early and late detection.