Genomics models in radiotherapy: From mechanistic to machine learning.

Area of Science:

• Radiation biology
• Medical physics
• Genomics

Background:

• Machine learning (ML) is widely used for high-dimensional prediction in classification and regression.
• ML applications in medical physics often focus on imaging, but are increasingly applied to biological data in radiation biology.
• Radiogenomics integrates genomic data with medical imaging to understand treatment response.

Purpose of the Study:

• To review radiogenomics modeling frameworks and efforts toward genomically guided radiotherapy.
• To discuss the development of precision biomarkers in medical oncology.
• To explore clinical assays for normal tissue or tumor radiosensitivity.

Main Methods:

• Review of existing literature on radiogenomics and machine learning in radiation biology.
• Discussion of modeling frameworks for radiosensitivity.
• Analysis of the evolution of ML for predictive radiogenomics models.

Main Results:

• Machine learning offers a framework for analyzing complex biological data in radiation biology.
• Efforts are underway to develop precision biomarkers and clinical assays for radiosensitivity.
• ML is evolving to create predictive models for radiogenomics.

Conclusions:

• Radiogenomics, powered by machine learning, holds significant potential for advancing genomically guided radiotherapy.
• Predictive models can improve our understanding of tumor and normal tissue radiosensitivity.
• Further integration of ML with biological data is crucial for precision medicine in radiation oncology.