A Comprehensive Review Of Methodologies In Clinical Data-Driven Decision Support Systems For Healthcare

Abstract

Objectives:
This review aims to analyze recent advancements in clinical data-driven decision support systems (CDSS) in healthcare on 15 years period (2010-2024), focusing on the evolution from rule-based to machine learning and AI-driven models in broad sense. The goal is to examine key methodologies, trends, domain-specific applications, and the challenges faced in deploying these systems in real-world clinical environments.

Methods:
A systematic review of IEEE Xplore, Web of Science, PubMed and Scopus was conducted covering publications from 2010 to 2024. Inclusion criteria targeted articles discussing data-driven CDSS methodologies, including supervised learning, deep learning, hybrid systems, and explainable AI (XAI). A total of 954 papers were initially identified, 213 met eligibility criteria, and 50 were selected for detailed analysis.

Results:
We selected 50 papers, 34 of which describe approaches in the data-driven AI area, 11 present purely classical rule-based CDSS, and 5 adopt hybrid approaches relying on both rule-based and data-driven AI.

Conclusions:

Recent studies in clinical data-driven Decision Support Systems (CDSSs) show a shift toward data-driven AI methods. These can work alone in purely data-driven systems or be combined with classical rule-based CDSS in hybrid systems. Hybrid approaches integrate machine learning with domain knowledge, creating a synergy that improves reliability and effectiveness.This combination also helps address key challenges in healthcare, especially transparency, interpretability, and explainability, which are now central to both AI and medical informatics research.