Multimodal Web Application to Infer Emotional Intelligence of Adolescent Counsellor

Abstract

There are only 0.3 psychiatrists and 0.047 psychologists per 100,000 people in India, compared to a country like the US, which has 29 psychologists per 100,000 people (according to WHO), thereby leading to lack of counselling services and mental health-care. Fortunately, researchers in India have found mental health interventions delivered by lay counsellors rather than specialists to be effective in treating and preventing mental health problems. However, choosing a lay counsellor from a pool of candidates becomes a very important but time-consuming and tedious task because of our deficits in evaluating emotional capabilities, implicit biases and facilitation skills in a resume and standard interview. In this paper, we present a highly scalable web application that can help in hiring emotionally intelligent lay-counselors. The backend framework measures several vital emotional intelligence features that are crucial in a prospective lay counsellor. The framework uses multi-modal data and provides a ranking of potential counsellors. The results and inferencing help establish the importance of each modality and gives insights on features that are key to identify the emotional skills. We compare the predicted rankings to those given by the interviewers (a clinical psychologist and a psychiatrist) and recognize the benefits of automation of the process as well as a need for a deeper analysis of interview questions, discriminative features and importance of multi-modality assessments.