Prostate cancer (Pca) is one of the most common cancers among men worldwide. The current screening methods lack effectiveness such as prostate-specific antigen (PSA) and Magnetic resonance imaging (MRI), and some others come with pain such as biopsy. Understanding the genomic behavior of the disease may play a key part in designing more effective, accurate, and less invasive diagnosis measures. Pca has many clinical features to describe the spread and the aggressiveness of the tumor including Gleason score, TNM staging system, and the location of the tumor in the prostate gland which is known as laterality.
Machine learning models were recently utilized to predict the outcomes of Pca, and to find potential biomarkers for the clinical features of the disease. In this study, we review recent machine learning methods for finding biomarkers for Pca clinical features including Pca progression, Gleason score, and laterality. The supervised models were built on gene expressions and next-generation sequencing data to find genes or genes transcripts that are associated with these clinical features. The results show high performance in the three models with an accuracy of more than 90%. The three models reported many biomarkers genes and genes transcripts including but not restricted to CARNA22, DOCK9, FLVCR2, IK2F3, USP13, PTGFR, and CLASP1 genes for Pca progression.  UBE2V2, GPR137, and EPB41L1 for different Gleason scores. And FBXO21, RTN1, NDUFA5, ALG5, Z99129, SNAI2, MRI1, HLA-DMB, SRSF6, and EIF4G2 for laterality prediction.
Keywords: Machine learning, prostate cancer diagnosis, next-generation sequencing, gene expression.

Machine learning, prostate cancer diagnosis, next-generation sequencing, gene expression.