Integrating Multimodal AI to Predict Treatment Response and Refine Risk Stratification in Esophageal Cancer （Radiogenomics-Esophagus）

Built upon retrospective cohorts for model development and rigorously validated in prospective cohorts, the proposed AI predictive model integrates multimodal data (radiomics, pathomics, genomics, and multi-omics)-each reflecting distinct dimensions of tumor heterogeneity-to enable joint prediction of treatment response and clinical outcomes.

Inclusion criteria

• Histopathologically diagnosed esophageal cancer
• Complete baseline clinical data available (including demographic characteristics, ECOG performance score, TNM staging, etc.)
• No other primary malignant tumors
• Provision of informed consent
• Availability of pre-treatment CT imaging

Exclusion criteria

• Imaging data quality insufficient for analysis
• Presence of another primary malignant tumor
• Severe systemic disease