Public Trust in Artificial Intelligence Applications in Mental Health Care: Topic Modeling Analysis

Area of Science:

• Digital Health
• Artificial Intelligence in Healthcare
• Mental Health Technology

Background:

• Mental disorders (MDs) present significant global healthcare burdens.
• Mobile health (mHealth) apps, particularly those using artificial intelligence (AI), are emerging as potential solutions for mental healthcare (MHC).

Purpose of the Study:

• To investigate public trust in AI-powered mental health apps.
• To identify dominant topics and themes within user reviews of leading mental health apps using topic modeling (TM).

Main Methods:

• Searched Google Play for top mental health (MH) apps and extracted user reviews (Jan 2020–Apr 2022).
• Utilized Latent Dirichlet Allocation (LDA) TM with Python (spaCy) for data cleaning and topic generation.
• Employed pyLDAvis for multidimensional scaling to categorize topics into themes and conducted qualitative analysis.

Main Results:

• Analyzed 3931 reviews from 8 leading MH apps (e.g., Wysa, Youper, BetterHelp).
• Identified four dominant topics: cheering people up (27%), calming people down (26%), understanding the inner world (25%), and serving as a therapist alternative/complement (22%).
• Combined topics into three themes: dispelling negative emotions, understanding the inner world, and serving as a therapist alternative, indicating overall high public trust.

Conclusions:

• This study is the first to use TM on user reviews to assess public trust in AI for MHC.
• Findings reveal a high degree of public trust, with user feedback offering valuable insights for app improvement.
• Addressing user concerns can enhance AI mental health apps, aiding prevalent MD treatment and reducing healthcare system strain.