SUN-574 Proteomic Signatures and Machine-Learning Models Predict Electro-Acupuncture Responsiveness in Overweight and Obese Individuals

Abstract Disclosure: Z. Lyu: None. Z. Huang: None. L. Zhong: None. Background: Electro-acupuncture (EA) is a promising intervention for managing obesity and insulin resistance (IR). However, individual responses vary, highlighting the need for predictive models to guide personalized treatment strategies. Objective: This study aimed to identify serum proteomic signatures that predict EA responsiveness in overweight and obese individuals. Methods: A randomized double-blind sham-controlled clinical trial was conducted with 138 overweight and obese adults, of whom 71 received EA and 67 underwent sham acupuncture. Serum samples collected pre- and post-8-week treatment were analyzed using targeted proteomics for 688 inflammatory and metabolic proteins. Machine-learning algorithms, including MUVR and Lasso regression, were employed to develop predictive models based on baseline proteomic data. Results: Notable differences in protein expression were observed between groups post-treatment. In the EA group, MCP-4, CCL4, IL-12B, CCL25, and CXCL9 showed increased expression, while the sham group exhibited reduced levels of proteins including MCP-1, VEGFA, and IL-18. A predictive model integrating 6 significant markers (CCDC80, MCP-2, LILRA5, MCP-4, SLAMF1, IFN-gamma) achieved an AUC of 0.937, demonstrating high accuracy in identifying EA-sensitive patients. These markers suggest immune and inflammatory processes as critical determinants of responsiveness. Conclusion: Differential proteomic profiles between EA and sham groups underscore acupuncture’s complex immunomodulatory effects. Our predictive model offers a step forward in personalized obesity management, tailoring EA interventions to individual biological responses. Future work will validate these findings and explore underlying mechanisms, paving the way for precise and effective therapies for obesity and IR. Presentation: Sunday, July 13, 2025