Two professors at the University's Hershey Medical Center hope to revolutionize how the medical profession currently diagnoses disease.
Dr. Arthur Krieg, professor of pathology, and Dr. Michael Bongiovanni, assistant professor of pathology, have developed a computer model which can help physicians evaluate diagnostic tests. Bongiovanni said this model is potentially more cost-effective than current methods of diagnosis and Kreig said it can also be more accurate.
The model, developed along with Dr. Robert Beck of the Dartmouth University School of Medicine, is based on the understanding that there are more than just two possible results of a medical diagnostic test: namely, that the disease is present or it is absent.
Many of the current systems of evaluation in use, according to an article by Kreig, Beck and Bongiovanni in the Dec. 9, 1988 issue of the Journal of the American Medical Association, imply only these two possible states.
The new model, with sufficient input data such as environmental risk factors and clinical state (condition of the patient), can also imply several variable results such as disease type, stage of growth, or degree of seriousness of the illness.
"Let's say the physician is testing for AIDS in the inner city where many of the patients are drug users," Krieg said. "If the patient has a weakly positive test result for antibodies, the patient has a greater chance for actually having the AIDS virus (than) a patient who lives in a rural conservative community."
The system allows the physician to combine environmental factors such as this with the test results to arrive at a better diagnosis, Krieg said.
Bongiovanni said sufficient data must be fed into the system before it can be implemented. Studies are performed on a number of healthy as well as sick people to come up with visual spreadsheets which can function as probability curves. From these "dot plots" physicians can visually estimate how likely it is that the disease is actually present in the subject, Bongiovanni said.
The dot plot system will also be more cost-effective, Bongiovanni said, because one graph can easily be tailored to suit an individual's needs. Whereas current methods often require separate and costly diagnostic tests for each patient and each suspected disease, the new model, with sufficient data, can be altered to suit various diseases, patients, clinical states and environments, he added.
"Medical costs have been rising rapidly," Krieg said. "As we search for ways to reduce the cost of health care, we must evaluate which services are most effective."
Bongiovanni said, "I'm not saying (the current method) is ineffective, but it could be much better."
For now, the dot plot system is used more on an individual basis than routine use at the medical center, Bongiovanni said.
"For instance, a urologist might come to us to see if the system is suitable for a patient," he added.
In the long run, however, both hope to see the computer model make an impact on everyday methods. But, Bongiovanni admitted, more technology and research must be implemented before it can advance.
Krieg said, "Ideally, I'd like to see ongoing technology assessment for all diagnostic procedures, but for now, this simple program allows physicians and clinical scientists to evaluate commonly used diagnostic tests on their own desk top computers."



