A Framework to Design Successful Clinical Decision Support Systems

Abstract

This paper presents seven principles for successful modeling of the clinical process, forming a framework for clinical decision support systems design. Modeling methods should incorporate data interactions during clinical decisions and should mimic the cognitive skills of clinicians. Predictive models need to be interactive, regenerating predictions in response to new clinical information, or clinician feedback. Any decision support method needs to consider trends of physiological measurements. Temporal trends can be stronger predictors of health outcomes, than cross sectional values. Finally, clinical decision support methods should be outcomes based, in an effort to avoid a 'historical decision' bias. These principles, can contribute to optimized modeling methodologies in healthcare settings, improving the response of health systems to decision making challenges.