ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING IN PREDICTING PROGRESSION AND RECURRENCE OF PROSTATE AND BLADDER CANCER: CURRENT INSIGHTS AND FUTURE DIRECTIONS

Urologic cancers such as prostate and bladder malignancies are characterized by considerable heterogeneity in their biological behavior and clinical outcomes. Accurate prediction of disease progression and recurrence plays a critical role in improving treatment planning, optimizing follow-up strategies, and advancing personalized medicine. Traditional prognostic tools, which rely primarily on clinical and pathological features, often lack the precision required for individualized risk assessment. In recent years, Artificial Intelligence and Machine Learning have emerged as powerful tools in oncological research, offering approaches to prognostic modelling based on large-scale, high-dimensional data. This review synthesizes findings from thirty-nine recent studies that investigated the use of artificial intelligence in predicting progression, recurrence, survival, and treatment outcomes in prostate and bladder cancers. The included works applied diverse machine learning techniques to data types such as magnetic resonance imaging, whole-slide pathology images, gene expression profiles, and electronic health records. Many models demonstrated improved predictive performance over traditional methods, particularly when integrating multimodal datasets. Furthermore, external validation in multicenter cohorts was increasingly reported, supporting the generalizability of results. Despite promising advances, the widespread clinical application of artificial intelligence in urologic oncology remains limited. Challenges include variability in data sources, lack of standardization, limited interpretability of models, and ethical concerns related to data privacy and fairness. Nonetheless, the integration of artificial intelligence into clinical decision-making workflows holds substantial promises for enhancing prognostic accuracy and supporting personalized management strategies. Future efforts should focus on harmonized methodologies, prospective validation, and transparent reporting to fully realize the clinical potential of these technologies.