AI for Mental Health

Given a sequence of utterances from the counseling dialogue, the task is to understand the dialogue-acts and response-acts. These acts are further exploited to generate effective response generation.

Given a sequence of counseling dialogue, the task is to generate a dialogue summary aka counseling note corresponding to each session.

Given a context of counseling dialogue, our focus is on exploiting reinforcement learning and domain knowledge to generate effective responses.

We work on analyzing counseling conversations in multiple counseling methods, including individual and peer counseling methods. The agenda is to understand users' behavior viz. intents, criticism, readability during discussions on symptoms, reasoning, gender, patient story, etc., on online mental health communities and virtual mental health assistants for effective counseling.