Methods Inf Med 2002; 41(02): 89-97
DOI: 10.1055/s-0038-1634291
Original Article
Schattauer GmbH

Clinical Data Retrieval: 25 Years of Temporal Query Management at the University of Vienna Medical School

W. Dorda
1   Section of Medical Information and Retrieval Systems, Department of Medical Computer Sciences, University of Vienna Medical School, Austria
,
W. Gall
1   Section of Medical Information and Retrieval Systems, Department of Medical Computer Sciences, University of Vienna Medical School, Austria
,
G. Duftschmid
1   Section of Medical Information and Retrieval Systems, Department of Medical Computer Sciences, University of Vienna Medical School, Austria
› Author Affiliations
Further Information

Publication History

Received: 27 June 2000

Accepted: 14 September 2001

Publication Date:
07 February 2018 (online)

Zoom Image

Summary

Objectives: Today, many clinical information systems include analysis components which allow clinicians to apply a selection of predefined statistical functions that satisfy typical cases. They are mostly to inflexible to handle complex, non-standard problems, however. The focus of this paper, therefore, is to present an approach that enables clinicians to autonomously create ad hoc queries including temporal relations in an interactive environment.

Methods: We developed the query language AMAS, which was specifically customized for users from the medical domain to flexibly retrieve and interpret temporal, clinical data. AMAS provides for a significant temporal expressiveness in data retrieval using timestamped clinical databases and relies on an operator-operand concept for the specification of a query.

Results: Within the last 25 years, four different clinical retrieval systems have been implemented at the Department of Medical Computer Sciences, based on the AMAS query language. Currently, these systems allow access to the medical records of more than 2 million patients. Physicians of 46 different departments at the University of Vienna and Graz Medical Schools have made extensive use of these systems in the course of clinical research and patient care, executing more than 10.000 queries per year.

Conclusions: We discuss a list of 20 issues that represent the most essential lessons we have learned in the development of the four systems mentioned above. Amongst others, our experiences indicate that the operator-operand concept allows an intuitive specification of complex, temporal queries. Further, customization to different user classes, based on their statistical background, is essential.