NLP meets Psychotherapy: from Estimating Depression Severity to Estimating the Client’s Well-Being

Language plays a crucial role in psychotherapy and the related processes. Linguistic expression not only reveals emotional states, but also provides insight into the immediate cognitions of the client. In this talk, I will present two projects related to the use of NLP for estimating the client’s depression severity and well-being. In the first project, we examine emotional coherence between the subjective experience of emotions and emotion expression in therapy and whether this coherence is associated with clients’ well-being. We present an end-to-end approach where we use emotion predictions from our transformer-based emotion recognition model to study emotional coherence and its diagnostic potential in psychotherapy research. In the second project, we propose an efficient semantic pipeline to study depression severity in individuals based on their social media posts. We represent each individual by a set of selected sentences and use them as evidence for predicting the individual’s symptom severity. For that, we explore different aggregation methods to answer one of four Beck Depression Inventory (BDI) options per symptom. We conclude by a short overview of our further ongoing and planned projects in NLP for Mental Health.

Post Talk Link:  Click Here 

Passcode: U0dg7Mi+