An AI-Powered Mental Health Monitoring Application: A Research on Mental Health Classification.

Abstract

The project discussed here deals with the conditions called Anxiety, Depression, Stress, and Suicidal tendencies, leading in a youth-oriented culture, under the broad area of Mental Health detection. A study is presented next that considers the labeled dataset comprising around 53,000 sentences describing seven states of minds, namely, Anxiety, Normal, Depression, Suicidal, Stress, Bipolar, and Personality Disorder. The different machine learning and deep learning models used include Roberta, GRU, LSTM, Logistic Regression, and Random Forest. Of these, the Roberta model produced a maximum accuracy of 81% and demonstrated an excellent performance in capturing contextual relationships within text. Further, this final model was then integrated with a chat-based application with the aim of real-time monitoring of states within user networks. This gives insight into good mental health and can, therefore, be helpful for intervention and support. The system provides a promising solution to address mental health concerns. This research contributes significantly to the field of mental health support, aiming to promote emotional well-being and foster healthier communities through AI-driven solutions.