JMIR Publications

ABSTRACT

Background:

Artificial intelligence (AI) chatbots are increasingly shaping health communication by mediating how patients, health professionals, researchers, and health institutions seek, interpret, produce, and act on health information. Existing reviews have largely focused on a single stakeholder group or clinical domain, leaving the broader question of who uses AI chatbots for what health purposes inadequately addressed. A multi-stakeholder synthesis is needed to understand where evidence is concentrated and where critical gaps remain.

Objective:

This systematic review aimed to map the empirical literature on AI chatbot use for health purposes across three stakeholder groups (the general public or patients, health professionals or researchers, and health institutions) and to characterize the purposes, evidence patterns, benefits, and risks associated with such use.

Methods:

Following PRISMA guidelines, we searched nine databases for peer-reviewed English-language empirical journal articles. Searches combined AI chatbot–related and health-related terms and covered records available through July 2025. After title, abstract, and full-text screening, 301 articles were retained for content coding. Articles were coded for publication year, country, sample size, method, chatbot modality, health topic, stakeholder group, and purposes of AI chatbot use, using a structured codebook.

Results:

The 301 included articles were published between 2023 and 2025. Studies were geographically concentrated in the United States and China and predominantly examined text-based interactions, with noncommunicable or chronic diseases and general health information as the most common health topics. Ten purposes of AI chatbot use were identified across stakeholder groups. Among the general public or patients (n = 152), purposes included (1) seeking health information, (2) symptom assessment and self-care management, (3) emotional support, and (4) preventive and transitional care support. Among health professionals or researchers (n = 189), purposes included (5) clinical decision support, (6) enhancing continuing education, (7) facilitating health research, (8) improving administrative efficiency, and (9) supporting patient interaction and doctor–patient communication. Among health institutions (n = 6), (10) public health management was the major purpose.

Conclusions:

This review provides a stakeholder–purpose mapping of AI chatbot use in health care, identifying ten purposes across three stakeholder groups. The evidence base is expanding rapidly but remains concentrated in text-based, lower-acuity, and information-oriented contexts, with AI chatbots used mainly to support interpretation, decision-making, communication, and workflow tasks. Reported benefits in access, scalability, personalization, low-barrier support, and administrative efficiency are accompanied by persistent concerns about accuracy, equity, transparency, and governance. Important gaps remain at the institutional level, in higher-stakes and longitudinal deployment contexts, in voice and multimodal modalities, and in empirical evaluation of equity and ethics. These findings offer a structured foundation for future research, design, and governance of AI chatbots in health care.