HCLSIG/Project Ideas/ClinicalDecisionSupport

Clinical Decision Support Systems

Proposal

Few commercial healthcare information systems provide a full complement of proactive decision support. The richness of applying a broad range of medical knowledge to influence physicians' orders through the responsible, appropriate use of (both discrete and probabilistic) knowledge representation is still an unachieved goal.

[[

  Only when both patient data and clinical knowledge reside in the system in machine-understandable format can the system provide additional support to the clinician making decisions.  For example, encoded medical knowledge about the meaning and significance of changing laboratory-test results would allow a system to provide alerts, an active function, in addition to the passive data retrieval function.

  Similarly, if the system could match the patient context with relevant clinical guidelines, it could present ordering options consistent with the appropriate guidelines, it could present ordering options consistent with the appropriate guidelines.  The clinician is responsible for the definitive decision, but the system can actively provide options and explanations that improve the clinician's efficiency and compliance with accepted guidelines of practice.

]]

• The Computer-based Patient Record: An Essential Technology for Health Care - Institute of Medicine, 1997

Clinical decision support requirements were the main catalyst for an earlier burst of research (MYCIN for example) around the intersection of Artificial Intelligence and medical record systems. The goal at the time was to build medical (rule-based) expert systems. The idea of using rule-based systems that have facilities for explaining inferences as a framework for supporting clinicians at the point of care is still quite relevant. The advent of Semantic Web technologies introduce a framework for cross-linked hypermedia artifacts that can facilitate the process of cross referencing relevant guidelines.

This task group would learn from the lessons of the previous (related) task group: Adaptive Clinical Pathways (ProtocolsHCLS/ACPPTaskForce), earlier medical expert system endeavors such as MYCIN, and current capabilities of Semantic Web inference engines. The goal would be to identify the core requirements (with a few supporting use cases) that are necessary for a successful decision support component to a medical record system that leverages semantic web technologies.