Foundations of Data-Driven Healthcare Decision Support Prior to Clinical Artificial Intelligence

Abstract: Data-driven decision support enables the consolidation, organization, and analysis of amassed health data for more effective decision-making. It complements AI solutions by directly addressing core user needs prior to the availability of clinical AI. Decision-theoretic models provide a normative framework for qualitative and quantitative models. Methods for evidence synthesis underpin clinical decisions supported by data without requiring AI. Technologies that enable these capabilities comprise database systems that collect and store data, tools that facilitate querying and exploration, statistical approaches that extract patterns and structure, and predictive models that summarize risk. The ability to perform medical decision-making has always been constrained by the complexity and uncertainty of the clinical environment. Data-driven decision support refers to a set of processes, methods, and supporting technologies that consolidate, organize, and analyze the accumulated clinical data to make decision-making easier and more reliable. Data- driven decision support does not rely on Artificial Intelligence (AI) methods, although it can help pave the way for them. Recent advancements in Natural Language Processing and Convolutional Neural Networks have made AI tools seem right around the corner, but it is important to remember that these technological trends might not deliver AI solutions for all clinical tasks any time soon. Data-driven decision support can provide answers when AI methods are either not available or cannot be trusted, such as in the early phase of development for a given medical task.

Keywords: Data-Driven; Clinical Decision Support; Decision Theory; Evidence Synthesis; Statistical Methods; Health Informatics; Clinical Decision Support Systems (CDSS); Evidence-Based Medicine; Health Informatics; Medical Data Analytics; Rule-Based Expert Systems; Electronic Health Records (EHR); Statistical Decision-Making; Knowledge- Based Systems; Clinical Guidelines and Protocols; Data Quality and Standardization.