LabScanner is a simple data mining program that reads text (ascii) files exported from clinical laboratory information systems or other clinical databases. It parses these files, using data specifications and rules entered by the user, to find and display longitudinal patterns in clinical test results (for example, three of five sequential test results for a particular patient above a certain value). It can display results graphically for inspection on-screen, export graphs for use in other programs, print reports containing graphs of data, and calculate incidence and persistence statistics for data patterns of interest in a group of patients. A description of a precursor to LabScanner can be found in J.H. Harrison and P. Rainey, Identification of patients for pharmacologic review by computer analysis of clinical laboratory drug concentration data, Am. J. Clin. Pathol., 103:710, 1995.Although LabScanner can be used to detect "critical values" and for delta-checking, it's real purpose is to identify more subtle patterns of change, such as a set of sequential results above or below a certain value, increasing or decreasing trends, excess variability in test results or excess frequency of testing. LabScanner was originally developed to identify "problem" results patterns in therapeutic drug monitoring, but these principles can be applied to other settings, such as tumor marker trend analysis, platelet survival and utilization, etc. Rules for identifying data patterns of interest are entered and modified by the user and thus can be specific for a particular application. LabScanner can be useful for identifying cases for clinical follow-up or training, in quality assurance studies, and in support of clinical pathology consultation. For example, over the past 5 years LabScanner's precursor has formed the basis for a clinical pharmacology rotation in the 4th year of medical school, identifying cases for student follow-up and evaluation.
LabScanner is intended primarily to aid clinical pathologists and increase their value and effectiveness in clinical laboratories by allowing their expertise to be focused on the critical review of selected cases. It was developed to demonstrate new avenues for the application of clinical expertise by pathologists for the improvement of patient care and clinical laboratory operations. Unless such links between clinical expertise and laboratory/patient management are made clear, the role of pathologists in clinical laboratories is likely to diminish--to the detriment of both clinical laboratories and the field of pathology. The effective use of computer-based tools for data analysis and case follow-up is one way to combat that trend.
The program is presently "beta"-level software, which means that it may contain bugs and it is intended to evolve to support both changes in current features and new features. Pathologists and laboratory managers who work with me to investigate LabScanner will define its ultimate feature set and user interface, and improve its reliability and accuracy. Because this software is currently developmental, no warranty can be made regarding accuracy and reliability. Though I test the software to the best of my ability, I am developing LabScanner personally on my own equipment and I cannot test on even a fraction of the variety of system configurations available. Though unlikely, it is possible that a bug in LabScanner might damage disk files on some computers, or that LabScanner might produce inaccurate results. To avoid problems, good laboratory practices similar to any developmental test should be followed:
LabScanner is copyrighted but is provided free of charge to the Pathology community. Its use is unrestricted except that it should not be distributed. New users should obtain the software from http://thelab.upmc.edu/jharrison/labscanner/ or an approved mirror site so that they can obtain the most recent version of the program and the documentation.
- Do not run development releases of LabScanner on systems used for critical production work or containing critical data files.
- Make sure to back up files from machines prior to LabScanner testing.
- Test LabScanner against the supplied data files initially to confirm correct operation.
- Create test data files from your own system with known contents to evaluate LabScanner's output and to test the performance of any new rules you create in LabScanner.
James H. Harrison, Jr., M.D., Ph.D.
Director, Division of Clinical Informatics
Department of Public Health Sciences
University of Virginia
West Complex 3181
Charlottesville, VA 22908-0717
434-924-8054
| Last modified: 3/9/01 | LabScanner Home | LabScanner Operation |