Artificial Intelligence in Clinical Decision-Making: A Scoping Review of Rule-Based Systems and Their Applications in Medicine.

Artificial intelligence (AI) has become increasingly integrated into clinical workflows, with rule-based clinical decision support systems (CDSS) emerging as one of its most mature and widely adopted applications. These systems rely on rule engines, that is, software components that apply predefined conditional logic (if/then rules) to patient data, to deliver alerts, diagnostic suggestions, or treatment recommendations. By embedding expert knowledge into structured rule sets and utilizing inference engines to process them, rule-based CDSS provides transparent, interpretable, and adaptable decision support. Although their use has expanded significantly over the past decade, evolving from simple decision aids to advanced tools incorporating AI and real-time analytics, a comprehensive synthesis of their applications, effectiveness, and technological evolution remains lacking. This scoping review aims to examine the current landscape of rule engine implementations in medicine, focusing on their clinical functionalities, evaluated outcomes, technological characteristics, and geographic adoption patterns across different medical domains. Following established scoping review methodology, we conducted a systematic search of PubMed and Scopus databases (2007-2023). Of 437 initially identified records, 28 studies met our inclusion criteria after rigorous screening. Data were extracted on study characteristics, clinical applications, rule engine technologies, and implementation outcomes, with particular attention to temporal trends and geographic distribution. The analysis revealed several key findings. The United States accounted for 46.42% of studies, demonstrating significant geographic concentration. Technologically, implementations evolved from early SQL-based systems to contemporary approaches integrating machine learning and natural language processing. Clinically, the rule engine showed particular effectiveness in chronic disease management (approximately 30% focused on diabetes care) and demonstrated measurable improvements, such as 30% reductions in adverse drug events. However, challenges persisted in system interoperability and clinician adoption across multiple studies. Our analysis of 28 studies demonstrates that rule engines have demonstrated substantial potential to enhance clinical decision-making and healthcare efficiency, though their adoption remains uneven geographically and is technically constrained in some settings. Based on our findings, we recommend: (1) developing standardized implementation frameworks to address interoperability challenges, (2) expanding research and deployment in underrepresented regions, and (3) investing in hybrid systems that combine rule-based logic with machine learning capabilities. These insights provide valuable guidance for healthcare organizations seeking to implement or optimize rule engine technologies in clinical practice.