The use of patient-specific Decision Support Systems (DSS) may improve the quality and efficiency of health care, while reducing its costs at the same time. The adoption of such a system is largely compatible with the principles of " Evidence Based Medicine " and patient oriented care. PROMEDAS (PRObabilistic MEdical Diagnostic Advisory System) is a prototype DSS, based on a probabilistic model and advanced computational techniques. The system offers patient specific diagnostic advice. It presents a differential diagnosis and it supports the diagnostic process by indicating the most useful next step in the diagnostic process. The system is intended to support diagnosis making in the setting of the outpatient clinic and for educational purposes. Its target-users are general internists, super specialists (i.e. endocrinologists, rheumatologists), interns and residents, medical students and others working in the hospital environment. Currently, PROMEDAS is a stand alone application. In the future PROMEDAS may be integrated with a Hospital Information System and an Electronic Patient Record. This will facilitate its use in practice, and may augment its acceptance. PROMEDAS is based on medical expert knowledge, acquired from the literature by the medical specialists in our project team. The acquired knowledge is stored in a database, in such a way that extension and maintenance of the expert knowledge is facilitated. Currently, the database covers large parts of endocrinology and lymphoma diagnostics. In near future, parts of vasculair medicine will be entered as well. From (parts of) this database, Bayesian network and an interface for PROMEDAS is automatically compiled. The network is the underlying model of PROMEDAS. Bayesian inference is used to query the system. The PROMEDAS project is funded by STW (project nr: NNN5322). The goal of the project is to demonstrate that an accurate diagnostic DSS covering a large diagnostic repertoire in internal medicine is possible. The key technical innovation is the use of advanced approximate inference methods which allow Bayesian inference to be applied to large problem instances (patent submitted). 3 4 " Promedas " , a prototype DSS
[1]
S. W. Holman.
A challenge.
,
1955,
Medical technicians bulletin.
[2]
R J Lilford,et al.
Decision analysis and the implementation of research findings
,
1998,
BMJ.
[3]
Marc Berg,et al.
Rationalizing Medical Work: Decision-support Techniques and Medical Practices
,
2022
.
[4]
Hilbert J. Kappen,et al.
Novel iteration schemes for the Cluster Variation Method
,
2001,
NIPS.
[5]
T. Heskes,et al.
Expectation propagation for approximate inference in dynamic bayesian networks
,
2002,
UAI 2002.
[6]
A. L. Baker,et al.
Performance of four computer-based diagnostic systems.
,
1994,
The New England journal of medicine.
[7]
Bruce D'Ambrosio,et al.
Multiplicative Factorization of Noisy-Max
,
1999,
UAI.
[8]
P. Trott,et al.
International Classification of Diseases for Oncology
,
1977
.
[9]
Tom Heskes,et al.
Probability assessment with maximum entropy in Bayesian networks
,
2001
.
[10]
R Haux,et al.
An integrated approach for a knowledge-based clinical workstation: architecture and experience.
,
1998,
Methods of information in medicine.
[11]
Judea Pearl,et al.
Probabilistic reasoning in intelligent systems - networks of plausible inference
,
1991,
Morgan Kaufmann series in representation and reasoning.
[12]
E S Berner,et al.
Relationships among performance scores of four diagnostic decision support systems.
,
1996,
Journal of the American Medical Informatics Association : JAMIA.
[13]
D. Heckerman,et al.
Toward Normative Expert Systems: Part I The Pathfinder Project
,
1992,
Methods of Information in Medicine.
[14]
D E Heckerman,et al.
Toward Normative Expert Systems: Part II Probability-Based Representations for Efficient Knowledge Acquisition and Inference
,
1992,
Methods of Information in Medicine.
[15]
Tom Heskes,et al.
IPF for Discrete Chain Factor Graphs
,
2002,
UAI.