A Qualitative Expert System for Clinical Trial Assignment

May 1998

Proceedings of the Eleventh International Florida Artificial Intelligence Research Society Conference (FLAIRS 1998)

Volume

Proceedings of the Eleventh International Florida Artificial Intelligence Research Society Conference (FLAIRS 1998)

All Papers

Assigning patients into clinical trials is a knowledge and data intensive task. Determining the eligibility of a patient for admission into a clinical trial is based upon specific criteria. These criteria may be shared among several protocols or may be unique to one protocol. A major difficulty in assigning patients to clinical trial protocols is the absence of complete information regarding the patient. Much of the needed data can be time-consuming or expensive to obtain, or needed tests can cause pain or discomfort to the patient Another difficulty is that there are many open trials at an institution at any one time and it is very difficult to keep track of criteria for each trial. This paper investigates the use of a fuzzy expert, system joined with a depemiency analysis to handle uncertainty and sort needed data for several protocols in order of influence. The system’s output is an evaluation of the patient’s eligibility for one or more clinical trials. Preliminary tests show that the system presents important data as high priority data while finding an appropriate order to obtain in all needed data. We have implemented four breast cancer protocols and successfully tested 15 cases which were clinically eligible for one of the four protocols.