Advancing Precision Oncology Decision-Making with Deep Learning: Integrating Genomic, Clinical, and Imaging Data

Precision oncology seeks to personalize cancer treatment by leveraging diverse data sources, yet integrating genomic, clinical, and imaging data remains challenging. This narrative review evaluates the role of deep learning in enhancing precision oncology decision-making through multimodal data integration. Seven key studies were analyzed, revealing that deep learning models, such as convolutional and deep neural networks, achieve accuracies of ۹۹.۹% and AUC values up to ۹۷% in tasks including diagnosis, subtyping, drug response prediction, and survival analysis. Integration strategies like early, intermediate, late, and decision fusion enable automated biomarker discovery and personalized treatment planning. However, barriers such as limited external validation, model interpretability, data heterogeneity, and lack of standardization hinder clinical adoption. Comparison with broader literature underscores the need for robust validation and explainable AI to bridge the gap between research and practice. Future directions include developing standardized frameworks, interpretable models, and large-scale, multicenter validation to fully realize deep learning's potential in transforming cancer care.

کلیدواژه ها:

Deep Learning ، Precision Oncology ، Genomic Data ، Clinical Data ، Medical Imaging