FACTS (Find Appropriate Clinical Trials) for You: A Computer-Based Decision Support System for Breast Cancer Patients

摘要

We have built a prototype of a web-based system that matches patients to NCI-listed clinical trial eligibility criteria. The system consists of a protocol encoder, an expression evaluator and inference engine that can handle uncertainty (via Bayesian belief networks), and user interfaces customized for the patient and the health care provider. We use an object-oriented model and standard nomenclature (UMLS) in this system. We use measures of value of information in the dynamic creation of data entry forms. We have completed a formative evaluation of the system, by comparing its ranked list of trials to those of two oncologists for the same retrospective cases of stage IV breast cancer collected from the Brigham and Women's Hospital. We have also elicited the oncologists' impressions of the quality of recommendations given by the system. Although the number of cases and protocols was limited in this experiment, the results suggest that systems such as this one can select appropriate trials for a given patient, even when presented with incomplete or uncertain information.