To evaluate the structural communication advantages of peers with medical background in social networks and their intervention effects on college students' knowledge, attitudes, and practices (KAP) levels in the prevention and control of human immunodeficiency virus (HIV), so as to provide theoretical support and practical pathways for structured health communication.

From April to June in 2025, 367 undergraduates were selected by stratified cluster random sampling in Nankai University. A multimodal strategy was employed to construct students' social networks. Fifty peer nodes were selected through a four-step screening process: betweenness centrality ≥0.3, eigenvector centrality ≥P₉₀, serving as student leader for at least one year, being mentioned ≥2 times in classroom interactions and information exchanges, with a total of five rounds, each lasting five days. Peer leaders daily pushed five Micromessage (knowledge points, interactive topics, or videos) through WeChat/QQ groups. KAP scale scores for AIDS prevention and treatment were measured before and after the intervention, and intervention effectiveness and predictive factors were analyzed using paired t-tests and Logistic regression analysis.

After the intervention, the total KAP score for AIDS prevention and treatment increased from (65.86±6.58) to (71.27±4.90) (t=-12.84, P < 0.01). The intermediary centrality (0.51±0.09) in the medical group was significantly higher than the non-medical group (0.34±0.12) (t=3.18, P < 0.01), with an average daily peak transmission rate of 8.71 people, and the network modularity increased from 0.39 to 0.47 (ΔQ=0.09, P < 0.01). The interaction between structural advantages and medical background significantly predicted KAP improvement (OR=4.21, 95%CI=1.89-9.38, P < 0.05).

Professional peers have significant structural communication advantages in HIV prevention and treatment interventions in colleges and universities. It is recommended to optimize sub-node screening strategies by combining network structure and professional background.