A Systematic Review on Biomarker Identification for Cancer Diagnosis and Prognosis in Multi-omics: From Computational Needs to Machine Learning and Deep Learning

Biomarkers, also known as biological markers, are substances like transcripts, deoxyribonucleic acid (DNA), genes, proteins, and metabolites that indicate whether a biological activity is normal or abnormal. Markers play an essential role in diagnosing and prognosis of diseases like cancer, diabetes, and Alzheimer's. In past years, in healthcare, an enormous amount of omics data, including genomics, proteomics, transcriptomic, metabolomics, and interatomic data, is becoming available, which helps researchers to find markers or signatures needed for disease diagnosis and prognosis and to provide the best potential course of therapy. Furthermore, integrative omics, often known as multi-omics data, are also proliferating in biomarker analysis. Therefore, various computational methods in healthcare engineering, including machine learning (ML) and deep learning (DL), have emerged to identify the markers from the complex multi-omics data. This study examines the current state of the art and computational methods, including feature selection strategies, ML and DL approaches, and accessible tools to uncover markers in single and multi-omics data. The underlying challenges, recurring problems, limitations of computational techniques, and future approaches in biomarker research have been discussed.