Scholarlink

<p>An increasing number of individuals are using generative AI applications (e.g., ChatGPT) for health consultation, yet research on psychological downsides remains limited. Grounded in the Comprehensive Model of Information Seeking (CMIS), this study examines how users’ interactions with “Dr. AI” may trigger cyberchondria. Using data from 849 users, a hybrid approach combining structural equation modeling (SEM) and artificial neural networks (ANN) shows that individual factors (AI self-efficacy, self-assessed health, positive metacognition) and media characteristics (diagnostic capability, empathy) contribute to AI addiction, which exacerbates cyberchondria through information overload. Higher education increases risk. SEM and ANN yielded consistent findings but differed in the predictive strength of diagnostic capability and self-efficacy. This study extends the application of the CMIS framework to the context of generative AI, uncovering the complex interplay between technology use and mental health outcomes, and offering practical implications for the design and regulation of AI-powered health tools.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

How does “Dr. AI” trigger cyberchondria? An empirical study based on the CMIS framework using a hybrid SEM–ANN approach

  • 期刊论文

  • 发表日期：

    2025年12月26日

查看全文
  • Chenyu Gu,
  • Jiayue Zhang

摘要

An increasing number of individuals are using generative AI applications (e.g., ChatGPT) for health consultation, yet research on psychological downsides remains limited. Grounded in the Comprehensive Model of Information Seeking (CMIS), this study examines how users’ interactions with “Dr. AI” may trigger cyberchondria. Using data from 849 users, a hybrid approach combining structural equation modeling (SEM) and artificial neural networks (ANN) shows that individual factors (AI self-efficacy, self-assessed health, positive metacognition) and media characteristics (diagnostic capability, empathy) contribute to AI addiction, which exacerbates cyberchondria through information overload. Higher education increases risk. SEM and ANN yielded consistent findings but differed in the predictive strength of diagnostic capability and self-efficacy. This study extends the application of the CMIS framework to the context of generative AI, uncovering the complex interplay between technology use and mental health outcomes, and offering practical implications for the design and regulation of AI-powered health tools.