JMIR Publications

ABSTRACT

Background:

Machine learning (ML) algorithms have been drawing attention at the joining of pathology and radiology in prostate cancer (PCa). However, due to their inner complexity and variability, there is an ongoing need to analyse their performance.

Objective:

This study assesses the source of heterogeneity and the performance of ML applied to radiomic, genomic and clinical biomarkers for the diagnosis of PCa. One research focus of this study has been on clearly identifying problems and issues related to the implementation of ML in clinical studies.

Methods:

Following the PRISMA protocol, 816 titles were identified from the PubMed, Scopus, and OvidSP databases. Studies using ML algorithms to detect PCa and providing performance measures were included in our analysis. The quality of the eligible studies was assessed using QUADAS-2 tool. The hierarchical multivariate model was applied to the pooled data in a meta-analysis. To investigate the heterogeneity among studies, I2 statistics were performed along with visual evaluation of coupled forest plots. Due to the internal heterogeneity among ML algorithms, a subgroup analysis was carried out to investigate the diagnostic capability of artificial intelligence systems in the clinical practise.

Results:

37 studies were included in the final analysis, of which 29 entered the meta-analysis pooling. The analysis of traditional ML to detect PCa reveals the limited usage of the methods and the lack of standards that refrain the implementation of ML in clinical applications.

Conclusions:

The performance of ML for diagnosis of PCa was considered satisfactory for several studies investigating the multiparametric magnetic resonance imaging (mpMRI) and urine biomarkers; however, given the limitations indicated in our study, further studies are warranted to extend the potential use of ML to clinical settings. Recommendations on the use of ML techniques were also provided to help researchers in designing robust studies to facilitate evidence generation from the use of radiomic and genomic biomarkers.