Optimizing CAPD Patient Monitoring Through Automated Vs Rule-Based Artificial Intelligence: A Systematic Comparative Review

Continuous Ambulatory Peritoneal Dialysis (CAPD) is a flexible renal replacement therapy that is widely used in developing and middle-income countries. Despite being beneficial, CAPD remains vulnerable to complications, such as peritonitis and fluid overload. In this systematic review, two prevailing artificial intelligence (AI) paradigms-rule-based systems and automatic machine learning approaches- were compared to enhance CAPD monitoring and decision-making. Literature published between January 1, 2020, to May 20, 2025, was assessed for clinical effectiveness, patient adherence, operational efficiency, cost, and usability. Automated AI systems for dialysate image classification have also been examined. Our findings suggest that automated AI systems provide greater precision and earlier detection, whereas rule-based models offer practical advantages in a low-resource structured environment such as Indonesia's healthcare system. These findings validate the value of integrating both paradigms, and propose a hybrid integration model to achieve the highest clinical accuracy, cost-effectiveness, and accessibility for CAPD monitoring. A total of 156 articles were identified, including 42 from PubMed, 37 from Scopus, 58 from Google Scholar, and 19 from IEE Xplore. Following screening and eligibility assessment, 24 studies were included for full synthesis. Of these, 12 investigated automated AI systems including machine learning based dialysate image classification and predictive modeling while 3 evaluated rule-based systems using predefined clinical logic. Overall 14 studies were identified as eligible studies that assessed the implementation of AI systems for the monitoring and management of CAPD. The proposed hybrid implementation model combines the strengths of both paradigms, tailored to national clinical guidelines and insurance schemes.