Background: Artificial Intelligence (AI) is seen as a potential solution to alleviate workforce demands arising from growing use of magnetic resonance imaging (MRI) in prostate cancer (PCa) screening.
Objective: We aimed to synthesize the evidence on use of AI in prostate MRI readings in asymptomatic men in PCa screening settings.
Methods: We conducted a rapid scoping review following PRISMA-ScR guidelines and Cochrane rapid review methods performing the systematic search of major databases supplemented by grey literature search with no restrictions in study design and time-duration. We considered various aspects of utilization of AI in MRI interpretations and biopsy indications in the screening setting. Anticipating limited evidence on AI implementation in screening settings, we extended the review from ‘what is known’ to discussion on ‘key considerations for expected expansion’. 
Results: We identified 284 records with 47 studies assessed for eligibility. Although several studies have evaluated the performance of deep learning (DL) algorithms for prostate MRI interpretation and decision-making in clinical settings with proven effectiveness, only two studies (Thimansson et al. [2024] and Winkel et al. [2020] ) met the inclusion criteria, having evaluated DL algorithms for prostate MRI interpretation in screening settings. Both evaluated commercially available ProstateAI software tool to interpret prostate MRI. Agreement between deep learning-based algorithm of AI and expert radiologist ranged from poor to moderate (kappa 0.17–0.42). In both studies, AI demonstrated high tendency of over-detection and low specificity, leading to discordance with expert radiologists. The authors highlighted that reduced specificity in the screening context risks unnecessary biopsies.
Conclusions and practice implications: Both studies were prone to sampling error because of the limited sample size, which further limits the generalizability of the results. The AI tool (ProstateAI) used by both studies was trained using 2,170 bpMRI prostate examinations with more PI-RADS≥3 lesions than lesion-free cases. This might have led to over-detection with reduced specificity in both studies. As most AI algorithms have been trained and tested on diagnostic datasets or retrospective cohorts, deploying AI to improve quality of care in a screening setting is a challenge to prove. Current evidence on use of AI in prostate MRI interpretation in screening appears limited. Despite promising potential, DL tools need to be robust, largely trained using large-scale diverse MRI images collected in screening population to enhance reproducibility and accuracy. In addition, ethical, legal, and social implications are central to AI adoption. One key consideration is informed consent and autonomy of patients on the use of AI in care processes or on how AI influences clinical decisions. The ongoing multicentre screening trials (PRAISE-U and EU-CanScreen) on prostate cancer are best positioned to provide evidence on performance of AI in MRI in risk-stratified screening setting and associated outcomes. 

Flow diagram of studies identification and selection process for PRISMA-ScR reporting system