Linmei Zhang Enzhao Zhu Jiaying Shi Xiao Wu Shaokang Cao Sining Huang Zisheng Ai Jiansheng Su ^*

• Tongji University, Shanghai, China

The conventional treatment for locally advanced head and neck squamous cell carcinoma (LA-HNSCC) patients is surgery, and the efficacy of definitive chemoradiotherapy (CRT) remains controversial.Objective: To evaluate the ability of deep learning (DL) models to identify patients with LA-HNSCC who can achieve organ preservation through definite CRT and provide individualized adjuvant treatment recommendations for patients who are better suited for surgery.Methods: Five models were developed for treatment recommendation. Their performance was assessed by comparing the difference in overall survival between patients whose actual treatment met the model recommendations and those who did not. Inverse probability treatment weighting (IPTW) was used to reduce bias. The effect of characteristics on treatment plan selection was quantified through causal inference.Results: 7,376 patients with LA-HNSCC were enrolled. Balanced Individual Treatment Effect for Survival data (BITES) performed best in both CRT recommendation (IPTW-adjusted hazard ratio (HR): 0.84, 95% confidence interval (CI), 0.72-0.98) and adjuvant therapy recommendation (IPTW-adjusted HR: 0.77, 95% CI, 0.61-0.85), which was better than other models and the National Comprehensive Cancer Network guidelines (IPTW-adjusted HR: 0.87, 95% CI, 0.73-0.96).BITES can select the most suitable treatment option for an individual patient from the three most common treatment options. DL models allow for the establishment of a valid and reliable treatment recommendation system with quantitative evidence.