Exploring Strategies for Personalized Radiation Therapy: Part III Identifying genetic determinants for Radiation Response with Meta Learning

2025-08-11

AI Summary

Existing radiobiological sensitivity prediction models (e.g., RSI) rely on fixed gene weights, limiting their ability to capture tumor heterogeneity and cross-cancer dynamic gene expression patterns. To address this, we propose a meta-learning–based single-sample radio-sensitivity prediction framework trained on cell-line gene expression data and optimized for the SF2 survival fraction metric. Our method enables rapid, individualized modeling by introducing a sample-adaptive gene weighting mechanism that explicitly encodes context-dependent gene–gene interactions and expression-level modulation—thereby overcoming the restrictive static-weight assumption. Experimental results demonstrate significantly improved generalization performance on highly heterogeneous subtypes (e.g., adenocarcinoma, large-cell carcinoma), superior cross-cancer prediction accuracy over state-of-the-art models, and the discovery of cancer-type–specific patterns in radiation-associated gene effects.

Technology Category

• Machine Learning: Learning Preferences or Rankings
• Search and Optimization: Metareasoning and Metaheuristics
• Game Theory and Economic Paradigms: Adversarial Learning

Application Category

• User Modeling, Personalization and Recommendation: Metrics for user behavior and evaluating success
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Radiation response in cancer is shaped by complex, patient specific biology, yet current treatment strategies often rely on uniform dose prescriptions without accounting for tumor heterogeneity. In this study, we introduce a meta learning framework for one-shot prediction of radiosensitivity measured by SF2 using cell line level gene expression data. Unlike the widely used Radiosensitivity Index RSI a rank-based linear model trained on a fixed 10-gene signature, our proposed meta-learned model allows the importance of each gene to vary by sample through fine tuning. This flexibility addresses key limitations of static models like RSI, which assume uniform gene contributions across tumor types and discard expression magnitude and gene gene interactions. Our results show that meta learning offers robust generalization to unseen samples and performs well in tumor subgroups with high radiosensitivity variability, such as adenocarcinoma and large cell carcinoma. By learning transferable structure across tasks while preserving sample specific adaptability, our approach enables rapid adaptation to individual samples, improving predictive accuracy across diverse tumor subtypes while uncovering context dependent patterns of gene influence that may inform personalized therapy.

Problem

• Identifying genetic determinants for personalized radiation therapy response
• Overcoming limitations of static models in predicting radiosensitivity
• Improving predictive accuracy across diverse tumor subtypes

Innovation

• Meta learning for one-shot radiosensitivity prediction
• Sample-specific gene importance via fine tuning
• Robust generalization across diverse tumor subtypes