Biochemical recurrence (BCR) following radical prostatectomy (RP) serves as a critical indicator of distant metastases and increased mortality risk in prostate cancer (PCa) patients. This systematic review aims to assess the effectiveness of artificial intelligence (AI) models in predicting BCR after RP.

A thorough literature search was performed across databases including Medline, Embase, Web of Science and IEEE Xplore, in line with PRISMA guidelines. Studies were included if they utilised AI to predict BCR in patients after RP. Studies were excluded if patients received radiotherapy or salvage RP.

Out of 9,764 articles screened, 24 studies met the inclusion criteria, which involving a total of 27,216 patients, of which 7,267 experienced BCR. AI models incorporating radiological parameters achieved higher predictive accuracy (median AUROC of 0.90) compared to models based only on pathological features (median AUROC of 0.74) or combined clinicopathological factors (median AUROC of 0.81). Evaluation using the Prediction Model Risk of Bias Assessment Tool (PROBAST) revealed an unclear risk of bias in three studies due to unclear inclusion criteria and significant exclusion of patients due to missing follow-up data.

AI shows promise in predicting BCR post-RP, particularly when radiological data was used in its development. These AI algorithms often outperform traditional methods of BCR prediction. However, significant variability in AI performance and study methodologies highlights the need for larger, more standardised prospective studies with external validation prior to clinical application.