THERAPY-PLANNING STRATEGIES FOR CONSULTATION BY COMPUTER

点击翻译按钮获取中文摘要

Therapy planning represents an important area for the development of computer-based clinical consultation systems. Even when a difficult diagnosis has been made, the optimum treatment may be far from clear. This is particularly true when adjustments to therapy are required after observing a patient's response to initial interventions. We propose basic research into the representation and use of therapy-planning strategies of the type routinely employed by expert clinicians as they monitor and adjust the treatment for patients with serious disease entities. In addition, we propose to implement and test these new techniques in the setting of an existing consultation system known as ONCOCIN. The ONCOCIN program is an oncology protocol management system that encodes the knowledge of highly structured cancer chemotherapy protocols and gives advice on treatment. Although ONCOCIN is currently in routine experimental use by oncologists in the Stanford oncology clinic, the program is limited in its ability to manage patients who have unexpected responses to therapy. A generalized approach to therapy planning that can be applied to ONCOCIN will broaden the system's scope and increase its clinical utility. In summary, this proposal combines basic research in the representation of clinical knowledge with a useful application. The specific aims of the research are: (1) to conduct basic research into possible representations of the therapy planning process, particularly how plans are modified in response to unusual conditions prior to or during treatment cycles; (2) to develop a computer program to represent the therapy planning processes of expert clinicians; and (3) to interface the planning program with ONCOCIN.