Predictive Immunonutrition for Paediatric Chronic Kidney Disease Surgery: A Multi-omic Data Fusion Framework through Canonical Correlation Analysis

Abstract Background: Preoperative immunonutrition is a promising approach to enhance surgical outcomes in high-risk populations, such as pediatric patients with chronic kidney disease (CKD). However, conventional protocols often fail to account for individual biological variability, limiting their effectiveness. Objectives: To develop a precision-based immunonutrition framework that integrates multi-omic data to predict individual responses and tailor nutritional interventions in pediatric CKD patients undergoing surgery. Methods: Our approach combines gut microbiome composition, host gene expression, and metabolic profiles using canonical correlation analysis (CCA) to extract latent multi-omic interactions. These features are then used to train a random forest classifier that predicts immunonutrition responsiveness with probabilistic scoring. The system dynamically adjusts interventions, including arginine and omega-3 fatty acid supplementation, based on predicted response levels. The framework leverages high-throughput sequencing, metabolomics, and scalable computing for real-time clinical integration. Results: The proposed system successfully predicted patient-specific responses to immunonutrition and supported personalized adjustments. It integrates smoothly with standard dietitian workflows and enhances clinical decision-making through data-driven insights. In pediatric CKD contexts, this approach demonstrated a marked potential for reducing post-operative complications. Conclusion: By shifting from a generalized to a precision-based paradigm, this framework lays the groundwork for individualized immunonutrition strategies in vulnerable pediatric populations. The model offers clinical scalability and may significantly improve surgical outcomes, particularly in personalized care settings like Saudi Arabia.