Towards the identification of tissue-based proxy biomarkers.

Area of Science:

• Biomedical data mining
• Computational pathology
• Personalized medicine

Background:

• Accurate patient stratification is crucial for personalized medicine.
• Enhanced biomarkers are needed for precise patient identification.
• Current data exploitation methods have limitations.

Purpose of the Study:

• To develop a novel computational framework for patient stratification.
• To create a tissue-based biomarker using multimodal data.
• To identify a high-risk patient population for colorectal cancer.

Main Methods:

• Integration of gene expression data and histopathology images.
• Development of a computational framework for biomarker discovery.
• Validation in a colorectal cancer patient cohort.

Main Results:

• A novel tissue-based biomarker was successfully constructed.
• The biomarker effectively identified a high-risk patient subgroup.
• The biomarker served as a proxy for a prognostic gene expression signature.

Conclusions:

• The developed framework enables multimodal biomedical data mining.
• The tissue-based biomarker can be readily implemented in routine pathology.
• This approach advances biomarker discovery and clinical practice for colorectal cancer.