In text-based therapy sessions, clients share their emotions with their therapist. Monitoring the mental health of their clients is a crucial task for the therapists. The project aims to predict the mental health status of the client by analyzing client-therapist interactions using emotion-cause extraction. Emotion-cause extraction in text not only serves as the fundamental technique for discovering and analyzing emotional cues in conversations. The project leverages natural language processing (NLP) to offer therapists valuable insights into their client’s mental health helping them to customize treatment plans and support strategies more effectively. The proposed methodology utilizes LSTM, a type of recurrent neural network that can handle long time-series data. This approach improves the accuracy and depth of the emotional cues, providing a comprehensive understanding of the client’s mental status.