Problem-oriented analysis of biological process data is hampered by the inability of existing tools to help an investigator identify a subset of data most relevant to solving a problem. Such a tool should 1) permit the entry of selection criteria directly by an investigator, 2) implement an algorithm for identifying the part of a data set that satisfies the specified criteria, and 3) support visualization of the retrieved data in a way that facilitates decision making speed and accuracy. We have implemented these criteria as a symbolic rule-based expert system called Problem-Oriented Temporal Analysis (PROTEMPA). It is a general-purpose framework for identifying subsets of ordered data (temporally or spatially) based on complex selection criteria. We are currently investigating its application to clinical laboratory, microarray, and protein sequence datasets.

Key personnel: Andrew Post, Jim Harrison, Valerie Monaco, Vanathi Gopalakrishnan (CBMI)

• PatientPatterns
• Temporal abstraction presentation (Feb 2004, 2.1MB Powerpoint file)
• Adaptive Clinical Laboratory Display based on PROTEMPA (proposal, 2.7MB PDF file)