Internet-based mental state monitoring using patient’s textual data – Year two

Among all chronic diseases, mental health issues have the highest burden on health care systems. However, unlike other chronic diseases, like Diabetes or hypertension, no monitoring procedures exist to monitor patients’ mental health status to prevent relapse and crisis situations. It is therefore necessary to develop cheap, convenient and accessible monitoring systems that could be used outside clinical setting. Most mental health diseases demonstrate a range of physical and behavioral symptoms (e.g. change in tone, posture and use of words, aka psychomotor symptoms) that could be measured using smart devices prevalently used by patients. Recent Internet-based methods of care delivery (eg online psychotherapy) provide the opportunity to utilize such digital evaluations of behavior (behavioral phenotyping) for long-term and remote monitoring of mental health status. Our proposal is to process digital behavioral data generated by the patients in an online platform (i.e. text, voice and video feedback) using machine learning approaches to develop an algorithm to predict their mental status. Furthermore, using recent advancements of deep learning in natural language processing, we are going to generate more personalized therapy content for patient interactions to improve the quality of the care.

Faculty Supervisor:

Nazanin Alavai

Student:

Amirhossein Shirazi

Partner:

OPTT

Discipline:

Psychology

Sector:

Health care and social assistance

University:

Queen's University