(Projektbeschreibung nur in Englisch)
K. Boegl1,
K.-P. Adlassnig1, G. Kolarz2,
A. Hatvan3
| 1 | Department of Medical Computer Sciences,
Section on Medical Expert and Knowledge-Based Systems, University of
Vienna Medical School, Spitalgasse 23, A-1090 Vienna, Austria e-mail: karl.boegl@univie.ac.at |
| 2 | Clinic for Rheumatic Diseases of the Social
Insurance Company for Trade and Industry, Adolfine-Melcher-Gasse 1, A-2500 Baden, Austria |
| 3 | Software Unlimited, Billrothstraße 31/18, A-1190 Vienna,
Austria e-mail: hatvan@swun.com |
Background
The Austrian Society for Rheumatology (Univ.-Doz. Dr. F. Singer and Univ.-Prof. Dr. G.
Kolarz) developed a consensus paper for a systematized documentation of signs and
findings in rheumatic diseases that offers a basic guideline for general practicioners [1].
Objective
The goal of this project was to adopt the representations of the consensus paper to an
easy-to-handle computer program for an electronic documentation of medical patient
data. An integration of the software in existing computer environments should be feasible
and a basic differential diagnosis support should be offered.
Material and Methods
The software consists of five data input screens (administrative patient data, history of
illness, clinical data, laboratory exams, and X-ray findings) and an output screen with
proposals of differential diagnoses. (Fig. 1) The differential diagnosis spectrum that is
covered by the underlying medical knowledge base comprises the following major groups
of rheumatologic diseases:
The knowledge base of RHEUMexpert comprises 350 findings, signs, and lab tests and 20 rheumatological diseases. The “raw” findings are converted into 71 higher-level medical concepts by 37 partly simple, partly complex rules (data-to-symbol conversion). The relationship between signs and diseases are specified by certainty factors, that either represent a positive or negative relationship. The inference algorithm is based on these certainty factors and a specifically designed gradient method. As a result, a number of diagnostic hypotheses (with a maximum of three diagnoses) are presented. In distinct cases, a further differential diagnosis support with more specific sub-diagnoses such as suspicion of rheumatoid arthritis and psoriatic arthropathy is provided. RHEUMexpert’s integrated database allows a persistent storage of patient data. Interfaces for data export and import support data transfer to and from other programs. In addition, the system contains an international classification scheme of rheumatological diseases.
Results
In an evaluation study 75 patients were tested and the overall accuracy of the top-level
diagnostic hypothesis generated by the system was 91% [2]. However, sensitivity and
specificity vary considerably among the various diagnostic groups. As an example, the
study showed that the sensitivity of well-defined disorders (e.g., rheumatoid arthritis)
reaches almost 100%, whereas it is as low as 50% in some other diseases (e.g., gout)
whose characteristic findings and symptoms are suppressed by treatment (drug medication)
in many cases.
Technical specification
RHEUMexpert was implemented in C++ and primarily designed for Windows95/NT. Currently a German version is available. In the course of a running master thesis project the
program is re-implemented using Java and the object-oriented medical expert system
framework MedFrame to allow a Web-based distribution and access of the system [3].
Conclusion
Our results showed that a computer-based documentation of rheumatic diseases facilitates
the systematized and standardized documentation of patient data. However, a few
modifications of the knowledge base as well as the knowledge representation formalisms
(i.e., to cope with the problems that are related with a concurrent therapy) are necessary.
These modifications will be implemented in the next release (RHEUMexpert-II).
References