Current role of artificial intelligence in the management of benign prostatic hyperplasia: a systematic review

INTRODUCTION: Artificial intelligence (AI) and its subsets, including machine learning (ML) and deep learning (DL), have revolutionized various fields, including urology. This review examines the application of AI in managing benign prostatic hyperplasia (BPH), focusing on its potential to enhance diagnosis, treatment, and patient education. EVIDENCE ACQUISITION: PubMed, Embase, Web of Science, IeeeXplore, and Cochrane databases were searched till December 2024. PRISMA guidelines were followed to identify eligible studies on artificial intelligence and benign prostatic hyperplasia. Data were divided into two broad categories: Large language models (LLMs) for patient education and clinical AI applications. Clinical AI applications were further divided in: biomarker identification, functional diagnostics, imaging, severity diagnosis, treatment prediction, and histological examination. Data extraction included study characteristics, data sources, model development and validation methods, handling of missing data, calibration, threshold selection, interpretability, and safeguards against data leakage. The risk of bias was assessed using the QUADAS-2, PROBAST, and ROBINS-I tools. EVIDENCE SYNTHESIS: Twenty-three studies were included: four evaluated large language models (LLMs) for patient education, and 19 investigated clinical AI applications, including biomarker identification, functional evaluation, imaging, severity diagnosis, and histological examination. The LLMs studies investigated AI chatbots like ChatGPT-3.5, ChatGPT-4, and New Bing Chat have shown high accuracy in answering BPH-related questions, aiding patient education. Biomarker Identification: Studies utilizing ML algorithms identified key diagnostic genes and biomarkers for BPH, improving diagnostic accuracy; Functional Evaluation: AI models accurately predict urinary flow patterns and diagnose bladder dysfunctions, reducing the need for invasive procedures; Imaging: AI improves the accuracy of prostate segmentation and volume measurement, aiding better diagnosis and treatment planning; Severity Diagnosis and Treatment: AI systems provide accurate severity assessments and predict treatment outcomes, supporting clinical decision-making; Histological Examination: AI-based models demonstrate high accuracy in diagnosing BPH and predicting tissue responses to treatments. CONCLUSIONS: AI has demonstrated significant promise in BPH management, offering enhanced diagnostic accuracy, personalized treatment strategies, and improved patient engagement. However, challenges remain, including data heterogeneity, small and institution-specific datasets, and a lack of prospective validation studies. Ethical concerns, regulatory oversight, and physician trust also need to be addressed before widespread clinical adoption. Future research should focus on refining AI models through multi-institutional collaborations, developing standardized datasets, and conducting real-world validation studies. With continued advancements, AI has the potential to revolutionize BPH management, ultimately improving patient outcomes and healthcare efficiency.