General practice clinical reasoning is a core competency for general practitioners to deliver comprehensive, continuous, and coordinated medical services. Traditional training models are limited by insufficient typical cases, inadequate general practice-specific training, and uneven distribution of teaching faculty. Artificial intelligence provides a novel pathway for the intelligent transformation of general practice clinical reasoning education.

To develop and validate a systematic AI-enabled clinical reasoning training program tailored to the characteristics of general practice, and to evaluate its implementation effect and acceptance among teachers and trainees in general practice education.

From March to May 2025, a "Problem-Design-Implementation-Evaluation-Reflection" framework will be adopted. Firstly, a diagnostic questionnaire survey will be conducted among 122 general practice clinical instructors and 164 general practice resident physicians undergoing standardized training (hereinafter referred to as general practice residents). Based on this, an AI-integrated training program incorporating general practice characteristics will be designed, comprising four major modules: "General Practice Intelligent Case Deduction - Real-time Feedback and Assessment-Personalized Learning Pathways-General Practice Faculty Guidance." A parallel-group randomized controlled trial (RCT) was performed: 90 general practice residents were randomly assigned 1∶1 to the experimental group (n=45) and the control group (n=45). The experimental group received a 12-week AI-integrated training intervention, including an AI general practice clinical reasoning platform, general practice-specific cases, comorbidity management, family assessment, and community emergency response. The control group received traditional general practice clinical reasoning training (theoretical lectures, case discussions, community clerkships, etc.). Intervention effects were assessed before and after training using general practice theoretical knowledge tests, the General Practice Clinical Reasoning Scale (G-DRS), chronic disease management reasoning scores, referral decision-making ability scores, and a learning satisfaction questionnaire.

The current diagnostic results show that 93.4% (114/122) of teachers and 87.2% (143/164) of students have used AI tools. Both teachers and students hold a positive attitude towards AI empowering the cultivation of clinical thinking in general practice, but only 33.6% (41/122) believe that existing tools meet the needs for general practice thinking cultivation. Teachers and students primarily focus on issues such as the authenticity of general practice-specific cases [61.5% (75/122)], simulation of grassroots referral scenarios [54.9% (67/122)], the risk of clinical thinking rigidity due to over-reliance on AI technology [79.9% (131/164)], and feedback lacking a humanistic dimension [56.6% (69/122)]. After group training, the experimental group's G-DRS score (27.9±3.5), chronic disease management thinking score (86.2±8.3), and referral decision-making ability score (85.7±7.9) were all higher than those of the control group [(22.5±4.1), (80.1±9.6), and (79.3±8.8), respectively] (P<0.05). After the intervention, the increase in the G-DRS score was greater in the experimental group than in the control group (P<0.05). The popularity of each teaching module was, in descending order: intelligent case deduction in general practice [95.3% (41/43)], general practice faculty guidance and workshops [93.0% (40/43)], real-time feedback and evaluation [88.4% (38/43)], and personalized learning paths [79.1% (34/43)].

General practice teachers and residents are open to AI-enabled clinical reasoning training, with key concerns focusing on case authenticity, scenario simulation, and humanistic feedback. The AI-enabled training program can effectively improve the clinical reasoning ability of general practice trainees. It provides replicable and scalable empirical evidence for the intelligent development of general practice education in China.

全科临床思维是全科医生开展综合性、连续性、协调性医疗服务的核心能力。传统培养模式存在典型案例稀少、全科特色训练不足、师资分布不均等"短板"。人工智能（AI）技术为全科临床思维培养的智能化转型提供了新路径。

构建并验证一套贴合全科医学特色的AI赋能临床思维系统化培养方案，评估其在全科医学教育中的实施效果与师生接受度。

2025年3—5月，采用"问题-设计-实施-评估-反思"框架。首先对122名全科临床教师和164名全科专业住院医师规范化培训（以下简称为全科规培）医师开展现状诊断问卷调查；在此基础上设计融入全科特色的AI融合培养方案，包含"全科智能病例推演-实时反馈评估-个性化学习路径-全科师资引导"四大模块；采用平行组随机对照试验，将90名全科规培医师按1∶1随机分为试验组（n=45）和对照组（n=45），试验组接受12周AI融合培养方案（包含AI全科临床思维平台、全科特色案例、共病管理、家庭评估、社区应急等）干预，对照组接受传统全科临床思维培养方法（含理论授课、案例讨论、社区见习等）干预。培训前后通过全科医学理论知识考核、全科临床推理评分量表（G-DRS）、慢性病管理思维评分、转诊决策能力评分及学习满意度问卷评估干预效果。

现状诊断显示，93.4%（114/122）的教师和87.2%（143/164）的学员曾使用AI工具，师生均对AI赋能全科临床思维培养持积极态度，但仅33.6%（41/122）认为现有工具契合全科思维培养需求。师生重点关注全科特色案例真实性[61.5%（75/122）]、基层转诊场景模拟[54.9%（67/122）]、过度依赖AI技术导致临床思维僵化[79.9%（131/164）]、反馈缺少人文维度[56.6%（69/122）]等问题。分组培训后，试验组G-DRS评分（27.9±3.5）分、慢性病管理思维评分（86.2±8.3）分、转诊决策能力评分（85.7±7.9）分均高于对照组[分别为（22.5±4.1）、（80.1±9.6）、（79.3±8.8）分]（P<0.05）。干预后试验组的G-DRS评分提升高于对照组（P<0.05）。各教学模块的受欢迎度依次是：全科智能病例推演[95.3%（41/43）]、全科师资引导与研讨[93.0%（40/43）]、实时反馈与评估[88.4%（38/43）]、个性化学习路径[79.1%（34/43）]。

全科师生对AI赋能全科临床思维培养持开放态度，关注焦点为全科案例真实性、场景模拟度、人文反馈等。AI赋能全科临床思维培养方案能提升全科医学专业学生的全科临床思维能力，可为我国全科医学教育的智能化发展提供可复制、可拓展的实证依据。