Report: Human Centered Systems in Health Care
Organizer & Chair: Simon Kasif, University of Illinois at Chicago

Panelists:	Joel Saltz, University of Maryland
	Barry Silverman, George Washington University
	Catherine Plaisant, University of Maryland

Medicine, more than any other single discipline stretches technology and imagination with respect to deployment of Human Centered Systems. Here are few reasons:

• Medicine is a knowledge rich field. Every procedure ranging from performing a surgery to examining X-Rays and conducting clinical studies relating genomic data to disease histories must be based on a deep arsennal of scientific and methodological knowledge.
• Medicine is a data rich field. The size of data depositories that include patient records, imaging information (e.g, MRI scans), laboratory tests, and anatomical information (Brain Maps), prohibit any feasible possibilities for browsing and analyzing this data by hand.
• Medicine is a technology-driven field. Many of the procedures are in the process of being automated. The automation trends to range from traditional problems such as scheduling procedures and staff, to more novel frameworks such as extracting information from medical reports and performing computer assisted surgery.
• Medicine is a people oriented field. This is perhaps the most important aspect of a proper deployment of Human Centered Systems in Medical practice, research and methodology.
• Medicine is a highly error-sensitive field. Any errors, especially in dissemination of knowledge, proper data sharing, clinical decision making, policies and procedures might be very costly.

The panel could not touch on all possible aspects of Human Centered Systems in Health Care and Medical Practice and Research. The panelists focused on the following topics:

• Human Computer Interfaces: the panelists discussed browsing medical records, and generally human centered approaches for accessing and retrieving structured medical data. Many basic problems require immediate attention: e.g, building dynamic query interfaces (that facilitate instant access to massive data). Other important developments include automated medical procedures such as robotic surgery, telesurgery, and human factors in computerized critical care. Human centered approaches play an important role here, carefully balancing automated procedures with giving the surgeon full control over the critical parts of the procedures.
• Virtual access to distributed image data and knowledge networks. The panelists discussed such problems as the virtual microscope that allows medical scientists to view (and share) image data across the internet. Related topics include computer assisted radiology.
• Medical Data Mining is another exciting direction in medical HCS. The wealth of unstructured and distributed data presents enormous opportunities for research related to information extraction from medical reports, and discovering association rules, complex multi-attribute correlations, and other difficult tasks involving discovering knowledge from data.
• Medical Decision Making: Continuing on the successful efforts to build effective automated clinical decision making tools (e.g, probabilistic networks) research must be done to move this technology closer to deployment and building new tools for integrating decision making algorithms with data mining technology. The basic problems here involve the interaction of knowledge and data and producing more human centered methodologies that increase the trust and confidence of medical staff in using intelligent systems for diagnosis.
• Network support for collaborative and shared HC environments.

The panelists included reports on several ground breaking systems. Prof. Simon Kasif discussed a system for automated discovery of coding regions in microbial DNA. The system is now in standard use by a central Genome Sequencing Center and has been deployed to annotate several new genomes. The system has been developed by Johns Hopkins University and TIGR (the institute for genomic research). A new prototype is being co-developed by the University of Illinois and TIGR.

Prof. Joel Saltz discussed several high performance systems built to support virtual access to medical images. A novel system has been built at the University of Maryland to support a virtual microscope at the Johns Hopkins Medical School.

Prof. Barry Silverman discussed a joint development with George Washington University and Harvard Medical School that supports a clinical database to be used in medical decision making.

Finally, Dr. Catherine Plaisant from the University of Maryland, discussed joint work with Prof. Ben Scheiderman to build effective human computer interfaces for browsing medical reports. These interfaces have been shown to improve the ability to focus instantly on the most important parts of the report and therefore improve the effectiveness of decision making.

These very successful (and in some cases already deployed) research efforts are just a small sample of the great potential of HCS in health care applications. The complex interaction of knowledge, data, technology, visualization, networking, cognitive skills and human capabilities cause health care and medical research to be a particularly challenging problem facing society. Therefore, this complex interaction creates uncountable possibilities for using HCS to enhance human intellect and capabilities in medical practice and research. The benefits to society and individual from deployment of HC systems in medicine (in terms of improving medical care, knowledge and cost) suggest the immediate need for increasing HCS research related to this fundamental application.

Simon Kasif
Department of Electrical Engineering and Computer Science
University of Illinois at Chicago
Chicago, IL 60607
312-355-0441
KASIF@EECS.UIC.EDU