Abstract:

【Abstract】 Objective　We conducted a study to investigate the efficacy of an AI clinical reasoning training system, based on a real case database and the DeepSeek large language model in the standardized residency training for ultrasound medicine.Methods　A total of 42 ultrasound residents trained between December 2023 and June 2025 were selected as study subjects and randomly assigned to an observation group (n=21) and a control group (n=21). The control group received traditional conventional training, while the observation group supplemented conventional training with the AI clinical reasoning training system for three months. After the training, a standardized ultrasound clinical reasoning ability assessment was conducted to compare the scores between the two groups, analyzing multiple dimensions including medical history taking integration, image interpretation, differential diagnosis, and diagnostic accuracy. A questionnaire survey was used to assess trainees' satisfaction with the teaching outcomes.Results　The assessment results showed that the total score of clinical reasoning ability in the observation group (91.14±2.20) was significantly higher than that in the control group (84.62±3.02) (P<0.001). Multidimensional analysis revealed that the observation group significantly outperformed the control group in history taking integration, comprehensiveness of differential diagnosis, and diagnostic accuracy (all P<0.001). However, no significant difference was found in ultrasound image recognition and interpretation between the two groups (P>0.05). The questionnaire survey indicated that the observation group's satisfaction reached 100% across all five aspects: improving theory-practice integration, enhancing self-directed learning ability, and strengthening clinical reasoning, which was significantly higher than that of the control group (all P<0.05). Conclusion　In standardized residency training for ultrasound medicine, the supplementary use of the AI clinical reasoning training system can effectively enhance residents' clinical reasoning ability and diagnostic proficiency, with significant effects particularly in history taking integration, differential diagnosis, and diagnostic accuracy, while also achieving higher teaching satisfaction. This study provides empirical evidence for the application of artificial intelligence technology in the systematic cultivation of clinical reasoning in ultrasound medicine education.