From Classroom to Ward: An Artificial Intelligent Pathway for Cultivating Medical Humanistic Care Competence Based on Care Ethics (Preprint)

BACKGROUND Currently, the vast majority of medical colleges and universities in China offer medical humanities curricula, with communication skills in medicine being a core component. This field depends on the integrated development of foundational humanities theories—such as psychology and ethics—and clinical disciplines. At Fujian Medical University, the communication skills curriculum comprises both theoretical instruction and extracurricular education.However, several interrelated challenges persist in traditional extracurricular practices. Theoretical instruction often lacks integration with authentic clinical contexts, making it difficult for students to translate abstract ethical principles into the ability to recognize and respond to patients’ genuine needs within complex power dynamics.Moreover, pedagogical approaches that treat empathy as a teachable skill—combined with a temporal gap between coursework and clinical practice—tend to foster emotionally scripted performances under pressure, impeding the integration of care ethics into students’ professional identities.This disconnect between ethical cognition and caring behavior manifests as three practical dilemmas:a lack of authentic "encounter" where students memorize abstract principles without perceiving patients' emotional cues in real situations; a scarcity of deep "focus" where empathy becomes procedural formality rather than genuine emotional engagement; and a rupture of experiential "continuity" between classroom learning and clinical practice, creating a significant knowledge-action gap.Traditional curricula also tend to overlook the emotional labor inherent in clinical communication, leaving students unprepared for the affective demands of real-world practice. To address these challenges, this study draws on care ethics as its theoretical foundation. Unlike traditional ethics that emphasize universal principles, care ethics foregrounds the relational and emotional dimensions of moral life. Originating in Gilligan’s critique of justice-focused ethics and systematically developed by Noddings into a framework for moral education, this approach posits that cultivating caring beings requires the establishment of reciprocal caring encounters. Building on this foundation, we deconstruct medical humanistic care competence into three interrelated dimensions:relational cognition—understanding physician-patient power asymmetry，emotional response—developing contextualized empathic abilities;and behavioral transformation—translating internal states into observable clinical skills. In addition, we introduce emotional labor as a key analytical lens. Defined as the management of feeling to create publicly observable displays,emotional labor research reveals that empathic responses are not merely spontaneous expressions but rather professionally cultivated performances requiring sustained cognitive and emotional effort—and may lead to burnout in the absence of adequate organizational support.Research has shown that embedding emotional labor theory into simulation and reflective practice helps students recognize, normalize, and regulate these experiences.Collectively, the foregoing theoretical frameworks inform our pedagogical design, emphasizing sustainable emotion regulation and the internalization of a professional caregiver identity. OBJECTIVE We aim to design an artificial intelligent pathway and adopt more positive pedagogical approaches to foster students’relational cognition, genuine emotional engagement, and the internalization of caring behaviors into professional identity. METHODS Participants At Fujian Medical University in China, a communication skills for medicine curriculum was offered as an compulsory undergraduate curriculum to both students. Overall, 150 medical students from the classes of Grade 2021 took the course in 2024, and 76 of them were willing to participate in the traditional extracurricular education; these students were used as the control group. An artificial intelligent closed-loop pathway was offered in 2025 as an upgrade to the extracurricular education. 136 medical students from the classes of 2022 took the course, and 100 of them were willing to participate in the artificial intelligent training project. The two groups of students received the same theoretical teaching content, and the teachers used the same teaching skills to complete the theoretical teaching task. In accordance with the ethical review exemption, informed consent was waived; however, all participants were informed of the study’s purpose and their right to withdraw at any time without academic penalty. Oral consent was obtained for any interview-related data, in line with institutional guidelines. Study steps Traditional training predominantly rely on instructor-led role-playing and structured case analysis, in which standardized clinical scenarios—including predefined physician–patient roles and medical conditions—are enacted, followed by systematic evaluation of communication behaviors and their impact on diagnostic reasoning and therapeutic outcomes.The artificial intelligent closed-loop pathway was composed of three phases of practical application. Phase A concerned classroom-based high-fidelity simulation, in which students engaged in dialogue with intelligent patient agents capable of simulating diverse emotional states and providing real-time feedback. Phase B involved structured practice integration during clinical clerkships, wherein students documented specific caring behaviours using a "Humanistic Care Record Form". Phase C comprised postclass activities, which included two steps: first, data-driven reflective advancement, where students submitted reflective reports within 24 hours for NLP analysis and received personalized learning resources; second, authentic experiential learning, which contained not only volunteer work in long-term care facilities but also immersive patient role-play and full-process patient accompaniment in hospitals. The entire procedure was conducted by the medical students themselves with instruction from physicians across various clinical specialties. Of note,this project was distinguished by a multidisciplinary collaborative approach wherein clinical courses offered concurrently with the Communication skills for medicine" curriculum—such as Surgery and Internal Medicine—were jointly planned and collaboratively delivered by interprofessional faculty teams. This design was intended to ensure that clinical instruction is infused with humanistic values, while humanities-based teaching is grounded in authentic clinical practice, thereby fostering a mutually reinforcing integration of technical competence and caring sensibility throughout students' learning experiences.Figure2 illustrates the overall study procedures of the artificial intelligent closed-loop pathway. The detailed implementation process of the proposed pathway is exemplified by the teaching case of "An elderly individual living alone who experienced a fall," as described below. After the project was completed, the participation rate for the artificial intelligence-driven training was calculated and compared with that of the traditional extracurricular education. The passing rate of the students in the communication skills for medicine curriculum with the artificial intelligence-driven training was also compared with that of the students in the traditional curriculum.Qualitative data were collected via reflective reports from all 100 participants and analyzed thematically to complement the quantitative findings. Teaching case selection The teaching case "An elderly individual living alone who experienced a fall" was selected as a representative example that reflects the multifaceted challenges of humanistic care in contemporary medical practice. This case was chosen for three reasons. First, it reflects the demographic realities of aging societies, where elderly patients living alone constitute a growing and vulnerable population requiring comprehensive humanistic care. Second, the case encompasses the full spectrum of clinical care—preoperative assessment, surgical intervention, and postoperative rehabilitation—allowing students to practice humanistic competencies across the entire care continuum. Third, the clinical complexity of an elderly fall patient inherently demands multidisciplinary collaboration, involving not only surgeons and anesthesiologists but also geriatric specialists, rehabilitation therapists, social workers, and discharge planners, thereby providing an ideal context for students to learn how humanistic care must be coordinated across diverse clinical specialties. This case thus serves as a paradigmatic teaching tool that integrates the core competency dimensions—relational cognition (understanding the patient's social isolation and vulnerability), emotional response (addressing preoperative anxiety and fear), and behavioral transformation (implementing sustained caring practices throughout the perioperative period)—within a clinically authentic and socially relevant context. Phase A: Classroom-based High-Fidelity Simulation This initial phase establishes a foundational, low-stakes environment for students to encounter and begin navigating the complexities of the patient-provider relationship. The simulation employs a non-proprietary, general-purpose chatbot, accessible via a mature open-source API. Rather than building a bespoke system, the research team configured this readily available technology by programming its role settings with detailed patient case parameters—including demographic characteristics, medical history, psychological profile, and communication style. Unlike