| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 377869 | Artificial Intelligence in Medicine | 2009 | 17 Pages |
SummaryBackgroundHypertension is one of the most prevalent chronic conditions and is directly correlated to deadly risks; yet, despite the availability of effective treatment, there is still clear room for improving patient outcomes. Use of relational databases is widespread for storing patient data, but formulating queries to identify patients whose clinical management can be improved is challenging due to the temporal nature of chronic illness and the mismatch in levels of abstraction between key management concepts and coded clinical data.ObjectiveThe objective of this work is to develop a sharable and extensible analysis tool that can be used to identify hypertensive patients who satisfy any of a set of evidence-based criteria for quality improvement potential.MethodsWe developed an ontology driven framework to enhance and facilitate some important temporal querying requirements in general practice medicine, focusing on prescribing for hypertension. The Web Ontology Language has been used to develop the ontology and the specific queries have been written in Semantic Query-enhanced Web Rule Language. We have used production electronic medical record (EMR) data from a General Medical Practice in New Zealand to populate the ontology.ResultsA unified patient management ontology consisting of a disease management ontology, a patient data ontology, and a plan violation taxonomy has been created and populated with EMR data. We have queried this ontology to determine patient cohorts that satisfy any of eight quality audit criteria, thereby identifying patients whose clinical management can be improved. A prescription timeline visualisation tool has also been developed to aid a clinician in understanding a patient's antihypertensive prescribing patterns, as well as visually validating the query results.ConclusionsThe presented framework shows potential to provide answers to clinically relevant queries with complex temporal relationships. The framework can be used to successfully identify hypertensive patients who need to be followed-up/recalled.
