A Review of Artificial Intelligence Techniques for Biomarker Discovery

Abstract

Biomarkers play a crucial role in medical research, helping guide diagnosis, prognosis, and treatment decisions. Traditionally, discovering these biomarkers has been a labor-intensive process, relying on experiments and statistical analysis that often capture only a small part of biological complexity. The introduction of Artificial Intelligence has brought a significant shift in this field. Machine learning approaches first allowed researchers to uncover patterns in genomic and clinical data that were previously difficult to recognize, and later, deep learning expanded these possibilities to include multi-omics data and medical imaging. Recently, explainable AI techniques have addressed the challenge of trust and interpretability, enabling clinicians to understand and validate AI-derived biomarkers. This review presents the journey of biomarker discovery with AI, highlighting key developments, current trends, and challenges, while discussing the potential for AI to enhance the clinical impact of biomarker research.