One in four of us will experience a serious mental health disorder each year. Early prognosis and diagnosis can markedly reduce the impact of these disorders, and in certain cases save people lives. Our work in the Insight Centre | Data Science Institute along with collaboration from Chalmers University of Technology in Sweden focuses on early diagnosis through dialog systems. Employing novel data science methods to diagnosis and predict disorders at the earliest stages of occurrence.

Passive prediction is a new method of diagnosing mental health disorders. It is based on the concept that the way in which people talk and write can be analysed to extract certain information. In this study, we have designed a chatbot that has a normal conversation with a person. During the conversation we monitor certain aspect of the written text and use machine learning methods to predict the possible presence of depression and anxiety. The first aim of the study is to test if our machine learning methods are effective. The second aim of the study is to test if the chatbot is a effective way to interact with people.