Based on multimodal data, we establish a dynamic deep learning model to conduct prognostic risk assessment for patients and recommend the 'most suitable' treatment/follow-up regimen after radical treatment, assisting clinicians in improving homogenized evaluation levels and achieving individualized precision therapy, thereby providing scientific evidence for the currently widely debated selection of postoperative adjuvant therapy in locally advanced head and neck squamous cell carcinoma patients.

This clinical study is an observational bidirectional study. A retrospective cohort is used to train the model for recurrent risk prediction and adjuvant treatment decision-making, while a prospective cohort serves as an external validation set to verify the model performance.

Inclusion Criteria:

• Age >= 18 years.
• Patients diagnosed and treated at Nanfang Hospital, Southern Medical University or its cooperative hospitals between January 2017 and December 2025.
• Patients who either seek medical treatment after June 2022 or sought medical treatment before that but have not lost contact with the hospital for more than 1 year are required to sign the informed consent form voluntarily.
• Newly diagnosed head and neck squamous cell carcinoma (HNSCC) confirmed by histopathology.
• Locally advanced disease without distant metastasis.
• Complete clinical data (including patient demographics, tumor characteristics, treatment details, and prognostic variables).
• Availability of MRI, CT, or PET-CT images before and after radical treatment.

Exclusion Criteria:

• History of or coexisting secondary malignancies.
• Lack of definitive radical treatment.
• Absence of baseline or follow-up imaging (MRI/CT/PET-CT).
• Investigator-determined unsuitability for study participation.

Patients diagnosed and treated at Nanfang Hospital, Southern Medical University or its cooperative hospitals between January 2017 and December 2025.