Project Description

Clinical treatment protocols are represented in an intention-based temporal representation language to overcome the drawbacks of vague or ill-structured problem definitions (e.g., missing functional dependencies). These representations are used to guide the patient, to provide necessary explanations, and to observe and critique whether the patient obeys the instructions of the health-care provider.

Patient Advocate's Functionality

The improvement of technical equipment facilitates more frequent and accurate monitoring of patient's health condition. There are several kinds of data about the patient's health condition available from different clinical analyses and devices. All this information is time-stamped and time-varying as a result of changing therapeutic actions and reactions of the human body. The health care provider has to analyze and interpret as much data as possible to derive diagnoses and to recommend appropriate therapeutic actions. In particular the patients have to comply with health care providers' instructions and advice, to exercise health-related common sense, and to monitor themselves for condition-specific danger signs over a period of time. Medical consulting time is expensive. Some patients cannot afford the expense or the time for adequate and regular medical consulting meetings.

Therefore, the central aim of the Patient Advocate project is to assist in clinical practice by helping patients to get a closer insight into their health conditions. Task-adequate methods give a global, comprehensive picture of all information available (e.g., observed parameters available at home, physicians'/nurses' instructions and advice) including explanations about the patients' needs, preferences, and experiences as well as the degree of severity of a situation. Such a comprehensive picture can be achieved by context-sensitive interweaving of different knowledge-based approaches to classify input data and by adequate visualization techniques. Currently, our attention is directed at this interweaving process, even though it is a very complex and partially domain-specific task.

The monitoring and consultation tasks consist of data selection, data validation, data abstraction, data visualization, recommendations, and explanations of data and vision.

Hundreds of medical information resources around the world are available on the World Wide Web (WWW), including information from federal agencies (e.g., NIH, FDA), clinical guidelines and protocols, literature and library services, medical encyclopedias, continuing medical education resources and, of course, many other non-medical services. Exploring the WWW is a time-consuming and unreliable task because there are too much data available. The variety of candidate sites and paths results in failing to retrieve necessary and accurate information in a reasonable time.

The Patient Advocate will facilitate a context-sensitive access to these various resources, providing the patient with additional explanation or teaching utilities on request. The guiding principle is that only those sites and resources are activated which are meaningful for the current patient's health condition. These sites and resources are annotated with an importance ranking to simplify the search process.

The patient often requires, or benefits from, interaction with different health care professionals, other patients, or support groups. Therefore, the patient is usually confronted with several coordination issues. She/He has to schedule a new appointment with the health care provider, remember scheduled or unscheduled therapy, find other patients with similar clinical conditions or support groups, and more. The Patient Advocate will provide the patients with necessary tools to assist these coordination tasks, such as an appointment scheduler, email connection to the medical staff or other patients, or access to supporting newsgroups.

In the current phase of the Patient Advocate project we are working on a prototype to implement an intelligent assistant performing the monitoring and consulting of the patient's health condition. Figure 1 shows a mock-up of a prototype which monitors and consults women with gestational diabetes mellitus.

Project Members

• Computer Scientists:
• Barbara Hayes-Roth
• Silvia Miksch
• Kenneth Cheng

• Medical Collaborators:

California diabetes & pregnancy program "Sweet Success",
Stanford University Medical Center
• Anne Regenstein, MD, Medical Director
• Gretchen D. Flanagan, RD, Clinical Nutritionist, Program Coordinator
• Lylia Needham, RN, MPA, Nurse Specialist

Research Funding

Publications

• S. Miksch, K. Cheng, & B. Hayes-Roth. The Patient Advocate: A Cooperative Agent to Support Patient-Centered Needs and Demands. Knowledge Systems Laboratory, KSL-96-14 , March 1996 & in 1996 AMIA Annual Fall Symposium, October 26-30, 1996, Washington, DC.
• Y. Shahar, S. Miksch, & P. Johnson. An Intention-Based Language for Representing Clinical Guidelines. Knowledge Systems Laboratory, KSL-96-15 , March 1996 & in 1996 AMIA Annual Fall Symposium, October 26-30, 1996, Washington, DC.
• S. Miksch, K. Cheng, & B. Hayes-Roth. An Intelligent Assistant for Patient Health Care. Knowledge Systems Laboratory, KSL-96-14 , July 1996.
• Y. Shahar, S. Miksch, & P. Johnson. A Task-Specific Ontology for Design and Execution of Time-Oriented Skeletal Plans. SMI-96-0608, July 1996 & in Tenth Knowledge Acquisition for Knowledge-Based Systems Workshop (KAW-96), Banff, Canada, November 9-14, 1996.