HUMAN-CENTERED AI IN MENTAL HEALTH: ETHICAL AND TECHNICAL CHALLENGES IN DIGITAL THERAPEUTICS

Abstract

The increase in the number of mental illnesses across the world and the absence of treatment using traditional medicine led to the development of digital solutions for mental health based on AI. However, while recognizing the benefits of digital technologies for mental health, one cannot ignore that their efficiency will depend on the adoption of human-centered solutions. For this reason, the purpose of the current paper is to examine the potential contribution of Human-Centered AI towards enhancing safety, usability, and efficiency of modern applications for mental health care. In particular, the investigation will identify some major challenges associated with the development of AI technologies for digital health, including heterogeneity of data, model bias, inadequate explainability, and adaptability issues. Moreover, some key ethical issues related to the implementation of AI technology in mental health care will be considered. On the basis of the results of the literature review, a theoretical framework based on human-centered AI will be outlined in the study. Specifically, the approach will be described through the continuous feedback from users and machine learning procedures. Human-in-the-loop decision making and customization depending on particular situations could help solve some of the issues faced currently due to advancements in technology. Also, the article tackles some of the compromises that exist between AI and humans.