Biomedical Informatics projects

Biomedical Informatics projects

Temporal pattern analysis in clinical data
Clinical decision-making using a data-driven display
Identification of clinical concepts in free text
An XML syntax for clinical laboratory procedure manuals

 
Temporal pattern analysis in clinical data
protempa_flow.pngPersonnel: James Harrison, Andrew Post

Development of a software framework allowing detection, aggregation and reporting of arbitrary patterns of change in time course data. An implementation in Java called PROTEMPA has been developed in collaboration with Andrew Post. PROTEMPA identifies time intervals over which patterns hold in raw data using software “detectors” specific for those patterns, and aggregates these intervals into higher order patterns using rules that define the allowable temporal relationships between intervals. PROTEMPA can be used for retrospective or prospective analysis.

Clinical decision-making using a data-driven display
Personnel: James Harrison (PI), Andrew Post

Support: National Library of Medicine, 9/30/2003 – 9/29/2006

Develop and test an adaptive user interface that presents patient data to clinicians in a graphical display and organizes the display based on patterns of change (detected by PROTEMPA) contained within the data. A preliminary study of the use of the interface by medical residents carrying out patient monitoring tasks is being evaluated.

Identification of clinical concepts in free text
Personnel: James Harrison

Support: NCI Shared Pathology Informatics Network, 8/13/2001-07/31/2006

A collaboration among pathologists at the University of Pittsburgh, Harvard and UCLA to create a corpus of annotated pathology reports with textual references to organs, procedures and key diagnostic concepts identified. Will form the basis for testing auto-coding systems.

An XML syntax for clinical laboratory procedure manuals
Personnel: James Harrison

Creation of a new XML vocabulary that can express the range of types of procedures typically found in a clinical laboratory, and evaluation of the syntax for ease of use and document complexity as compared with the generic structural markup provided by the HL7 Clinical Document Architecture “narrative block.”