Work

Comparing DeepSeek-V4 and human clinicians in self-management education for patients with diabetic foot ulcer

Year: 2026

Type: article

Abstract: Diabetic foot ulcer (DFU) is a prevalent global condition. Improving the self-management ability of patients with DFU can effectively reduce the incidence of ulcers. Large language models (LLMs) have shown potential to enhance self-management abilities. However, the applicability of LLMs in the professional context of DFU has yet to be verified. To evaluate and compare the performance of LLMs against experienced clinicians in generating accurate, safe patient education materials about the self-management of patients with DFU. This study evaluated self-management education materials generated in response to 30 DFU-related questions. Data collection was divided into two parts. (1) Education materials written by human clinicians. (2) Education materials generated by DeepSeek-V4. The evaluation was divided into three parts. (1) Assessment of understandability and actionability using PEMAT-P, (2) evaluation of readability through six established formulas, and (3) experts rating of quality, accuracy, comprehensiveness and humanistic care using four Likert 5-point scales. The information generated by DeepSeek-V4 and human clinicians showed statistically significant differences in terms of readability, quality, comprehensiveness, and humanistic care ( P < 0.05). Compared with human clinicians, DeepSeek-V4 showed significantly higher quality and comprehensiveness, and comparable understandability, actionability, accuracy and readability compared with clinicians. In terms of readability and humanistic care, DeepSeek-V4 performs relatively poorly. LLMs, particularly DeepSeek-V4, showed good performance in generating education materials for patients with DFU, which can assist human clinicians in providing health education.

Source: Scientific Reports

DOI: 10.1038/s41598-026-70588-1

Authors: Shangxue Sun, Xinyin Zhang, Yangyang Li, Fei Peng

Institution: Second Military Medical University

Language: English

Topic: Artificial Intelligence in Healthcare and Education

Subfield: Health Informatics

Field: Medicine

Domain: Health Sciences

SDG: Good health and well-being

Open Access status: gold

Estimated APC paid: $2,690