JMIR Publications

ABSTRACT

Background:

Artificial Intelligence (AI) technology has advanced rapidly, continuously expanding its applications across various scenarios in the healthcare sector. These applications span multiple domains, including image-assisted diagnosis, intelligent electronic medical record entry, nursing plan formulation, and intelligent decision support[1][2]. The introduction of AI tools in healthcare holds promise for enhancing service efficiency, optimizing clinical workflows, and alleviating the workload of medical staff, making it a crucial driver for high-quality development in the healthcare industry. However, current implementation outcomes indicate that AI applications in healthcare have not fully met expectations[3]. Limited acceptance and depth of utilization among some healthcare practitioners have resulted in a phenomenon where AI products are easy to introduce but difficult to implement effectively. Existing research primarily focuses on technical optimization of AI tools, with effect evaluations being relatively simple and fragmented. There is a lack of systematic, in-depth analysis of the challenges in applying AI technology within the medical field[4], and even fewer studies adopt a qualitative perspective[5]. Qualitative research primarily centers on uncovering subjects' authentic perceptions, subjective experiences, and underlying needs. Compared to quantitative surveys, it more accurately captures researchers' genuine thoughts, inner concerns, and value judgments when confronting AI applications. It delves into the psychological mechanisms and sociocultural factors driving these challenges，precisely the current research gap[6]. Existing qualitative studies have yet to comprehensively explore healthcare practitioners' challenges, practical needs, genuine perceptions, and psychological processes when applying AI[7], making it difficult to fully reveal the multifaceted causes and underlying logic of AI's application dilemmas in healthcare. With the continuous advancement of national policies promoting the AI+healthcare strategy, the integration of AI technology and the healthcare industry has deepened further. However, the challenges in its application have also become more pronounced[8]. Failure to promptly clarify the manifestations, underlying causes, and influencing factors of these challenges will severely constrain the sustainable and effective development of AI in healthcare, thereby hindering the achievement of national health strategy goals. Against this backdrop, it is imperative to systematically investigate the practical challenges of scaling AI technology in healthcare from a qualitative research perspective. This involves delving into the underlying factors: stakeholder perceptions, technological limitations, institutional gaps, and societal influences. Therefore, this study employs grounded theory to conduct semi-structured in-depth interviews with 23 healthcare practitioners. This aims to fill gaps in existing research, refine the research framework for AI applications in healthcare, and provide empirical support for relevant authorities to formulate supportive policies, regulatory standards, and ethical guidelines for AI healthcare applications. It also offers targeted recommendations for optimizing AI healthcare products and enhancing healthcare professionals' application capabilities, thereby promoting the standardized and sustainable development of AI healthcare. This contributes to the high-quality development of the healthcare industry and the advancement of Healthy China initiatives.

Objective:

To explore the practical challenges encountered in applying artificial intelligence within the healthcare sector, providing theoretical and practical insights for optimizing the AI healthcare application ecosystem and promoting human-machine collaboration.

Methods:

Employing grounded theory research methodology, 23 healthcare professionals with diverse roles, titles, and years of experience were selected for semi-structured interviews. Data were collected through open coding, axial coding, and selective coding to progressively refine core categories and theoretical models.

Results:

The interviews yielded five themes and 13 subthemes related to AI application challenges in healthcare. These challenges can be categorized into four dimensions: technical, organizational support, cognitive/emotional, and institutional. Significant differences in AI application challenges were observed across different groups of healthcare practitioners.

Conclusions:

The application of AI in healthcare is influenced by the interplay of multidimensional factors including technology, organization, cognition, and institutions. Differentiated and personalized solutions tailored to the actual needs of different groups are required to avoid a one-size-fits-all implementation approach. Clinical Trial: Ethical approval was obtained from Ethics Committee of the Second Affiliated Hospital of Guangxi Medical University 2025-KYC(0531).