Primary care general practitioners encounter significant challenges in diagnosing and managing skin diseases, highlighting the urgent need for artificial intelligence (AI) assisted systems. Although AI has the potential to improve diagnostic and treatment efficiency, research on its application in primary care settings remains limited.

To investigate the effectiveness and impact of an AI-assisted system in supporting primary care general practitioners with the diagnosis and management of skin diseases.

From December 2022 to March 2024, 19 general practitioners from community health centers in Hangzhou were voluntarily recruited for this study. They were randomly divided into two groups: an AI group with 10 physicians and a control group with 9 physicians. During this period, these physicians treated a total of 90 patients with skin diseases: 50 in the AI group and 40 in the control group. Physicians in the AI group utilized the Ruifu AI-assisted system for diagnosing and managing dermatological diseases, whereas those in the control group followed standard treatment protocols without AI assistance. Both groups compiled patients' medical records, auxiliary examination reports, and photographs of skin lesions during consultations. Two skin disease experts were invited to conduct remote consultations to evaluate the diagnostic accuracy of the two groups. On the first day (1 d) and the fourteenth day (14 d) of treatment, patients underwent assessments using the Dermatology Life Quality Index (DLQI), and satisfaction surveys were conducted separately for patients in the AI and control groups. A questionnaire survey was administered to doctors in the AI group to assess their experience with the Ruifu AI-assisted system.

No significant differences were observed in gender, age, or education level among patients in the AI and control groups (P>0.05), nor among physicians in terms of gender, age, education, and professional titles (P>0.05). The AI group's general practitioners achieved higher diagnostic accuracy for skin diseases than those in the control group (64.0% vs 37.5%, P=0.012). Fourteen days post-treatment, improvements in the DLQI scores were observed in both the AI and control groups, with significant differences (P<0.05), and the improvement in the AI group was more significant (P<0.05). The satisfaction level of the AI group was higher than that of the control group (P=0.024), and there was a positive correlation between the 14 d DLQI score and patient satisfaction in the AI group (r_s=0.471, 95%CI=0.186-0.683, P=0.002), the correlation between the improvement in DLQI score and patient satisfaction was even more significant (r_s=0.816, 95%CI=0.676-0.899, P<0.001). The results of the questionnaire survey revealed that a majority of physicians demonstrated a positive attitude towards their use of the AI-assisted system, acknowledging its practical value in several areas: diagnosis selection (70.0%), auxiliary diagnosis (80.0%), treatment recommendations (60.0%), and the provision of professional knowledge (90.0%). Remarkably, 90.0% of the physicians indicated their intention to continue utilizing the AI-assisted system.

In the primary care setting, the application of AI-assisted systems has enhanced the diagnostic accuracy of general practitioners in identifying skin diseases, improves the quality of life for patients, and increases patient satisfaction. The majority of general practitioners report positive experiences with the use of AI-assisted systems.