Center for Advanced Medical Informatics at Stanford (CAMIS) (1990-1998)

[Taken from a 1994 brochure describing the CAMIS resource]

Principal Investigator:	Edward Shortliffe (1992-1997)
Resource Director:	Thomas Rindfleisch (1992-1996)

What was CAMIS?

The Center for Advanced Medical Informatics at Stanford (CAMIS) was established by the Dean of the Medical School in 1990 to create links between academic, research, and administrative computing activities in the School of Medicine, and to serve as a source of advice to the Dean’s office on informatics issues. The Center was funded in 1992 by the National Library of Medicine (NLM) to create a shared computing resource to support research activities in biomedical informatics. This resource was also named CAMIS and supported an active core research program, the computing infrastructure and day-to-day computing needs of a large number of researchers, each of whom has external support from the National Institutes of Health (NIH) or other agencies. Though the CAMIS resource was Stanford-based, its mission had a national character, which included support of national electronic mail distribution lists, extensive on-line information resources, an annual CAMIS symposium, two academic training programs, a visiting scholar program and a national advisory council.

For CAMIS, Medical Informatics is the rapidly developing scientific field that deals with biomedical information, data, and knowledge — their storage, retrieval, and optimal use for problem-solving and decision-making. Medical Informatics touches on all basic and applied fields in biomedical science and is closely tied to modern information technologies, notably in the areas of computing and communication. The emergence of medical informatics as a new discipline was due in large part to advances in computing and communications technology, to an increasing awareness that the knowledge base of medicine is essentially unmanageable by traditional paper-based methods, and to a growing conviction that the process of knowledge retrieval and expert decision making is as important to modern biomedicine as is the fact base on which clinical decisions or research plans are made.

Background and History

CAMIS emerged from a long and rich history of biomedical and artificial intelligence work at Stanford University. Beginning in the mid-1960s, with the ACME Computer resource, researchers including Joshua Lederberg, Edward Feigenbaum, and Gio Wiederhold helped define the use of large time-shared computing resources for biomedical research. Shortly thereafter, a confluence of computer science, department of chemistry, and medical school interests resulted in a shared facility for the national networked community of collaborators who applied the techniques of artificial intelligence (AI ) to biomedicine. This facility was called SUMEX-AIM and it supported some of the seminal work in biomedical informatics, including DENDRAL, SECS (University of California at Santa Cruz), MYCIN, EMYCIN, INTERNIST/QMR (University of Pittsburgh), PUFF/VM, ONCOCIN, PROTEAN, PROTÉGÉ, and T-HELPER.

By the late 1980’s, researchers had learned to adapt AI, along with many other methodologies, to the task of promoting more effective biomedical computing and to assisting biomedical scientists in their basic science, clinical research, patient care, and education. In short, while the AI component of the work was important, it was only one facet of work that became more broadly focused on the entire field of medical informatics.

CAMIS brings together many of the same investigators who previously worked under the SUMEX-AIM resource, but it adds others and builds on symbiotic ties with other computing and communications activities in Stanford Medical School as well as the national bioinformatics research community. CAMIS retained its strong links with researchers in the Stanford Computer Science Department, especially in the Heuristic Programming Project under Professors Feigenbaum and Fikes.

CAMIS Resource Goals and Activities

The CAMIS resource, under Professor Edward H. Shortliffe as principal investigator and Mr. Thomas C. Rindfleisch as resource director, is guided by a number of long-term goals:

• to promote scientific communication, collaboration, and sharing within a distributed research environment;
• to provide a superb setting for exploring biomedical informatics topics;
• to pursue a strong basic research effort in biomedical informatics and to develop improved tools for building more complex and effective applications systems;
• to help define new ways to disseminate computing and information-management technologies into real-world settings;
• to create a state-of-the-art computing and communication environment for our research work.

CAMIS consisted of three main parts – the core research community, the collaborative research community, and the computing and communications infrastructure. Within this structure, CAMIS undertook work in four principal areas: core research and development; collaborative research; infrastructure development and service; and training, education, and dissemination. These are described in more detail below.

Core Research and Development

Work in this area was intended to provide the long-term technological base for collaborative research in the CAMIS resource and infrastructure services. Initially it support core research projects in several areas:

• Clinical Trials Workstation (CTW) Project — generalize and abstract system-design concepts and implementation solutions from our ONCOCIN and T-HELPER projects so that they can be applied to facilitate the more rapid development of integrated workstations in any domain of clinical-trials management [Ted Shortliffe, Mark Musen, Larry Fagan, Robert Carlson (Oncology), Angel Puerta, Samson Tu, John Gennari, Josef Schreiner, and Elizabeth Miller];

• Understanding the Code for Biological Molecules — develop and generalize tools for discovering and understanding the structural codes for biological macromolecules based on belief-network methodologies that efficiently capture three-dimensional structure relationships [Douglas Brutlag (Biochemistry), Russ Altman, and Ross Shachter (Engineering-Economic Systems)];

• Advanced Computing Systems and Environments — build upon the extensive distributed-computing environment already in place and improve, extend, and adapt new computing tools as necessary to support the CAMIS community. Research areas include gesture-based portable computing, information resource navigation and retrieval tools, email management tools, high-speed networks, large-volume file storage, and new workstation technologies [Thomas Rindfleisch, Larry Fagan, Richard Acuff, Torsten Heycke, Christopher Lane, Michael Macgirvin, William Yeager, and Nicholas Veizades].

Collaborative Research

A community of 15 collaborative research projects was supported by the CAMIS resource, representing work in areas such as AIDS protocol management, methods for uncertain reasoning, bioinformatics, health care outcomes and economics, basic artificial intelligence research, human-computer interfaces, multimedia authoring and delivery tools for medical education programs, and distributed library resources. Additional collaborations were encouraged through improved mechanisms for inter- and intra-group communications.

Infrastructure Development and Service

CAMIS provided effective and widely accessible distributed communication and computing facilities, consulting services, and systems management support to the research community. Specific activities included:

• Integrating and operating advanced network systems for the CAMIS community, including high-speed or wireless networks as they become available. (CAMIS network services are coordinated with the Medical School Information Systems Group (ISG) which is funded by the Dean’s office and administers general network services (MedNet) in the Medical Center.)
• Providing commonly needed, general purpose host and distributed computing services for the CAMIS community as determined by requirements derived from close contact with core and collaborative research groups.
• Assisting CAMIS research groups in exploiting computing and communication services through consulting on system purchases, software systems, integration issues, troubleshooting, and application development.
• Providing documentation and training for CAMIS user groups to increase awareness and proficiency in the use of information management, system development, and computing tools.
• Serving as the liaison between the CAMIS resource and other information systems groups, including the ISG, elements of the Library and Information Resources office, the Lane Library system, Computer Science and School of Engineering facilities, and other local and Internet research groups contributing expertise and technology beneficial to CAMIS goals.

Training, Education, and Dissemination

CAMIS supported two formal graduate degree programs (in medical information sciences and in health-services research), and sought improved ways to disseminate research results through publications, on-line information resources, video presentations, visiting scholar programs, and an annual CAMIS symposium.

Medical Information Sciences (MIS) program

The Medical Information Sciences (MIS) program began in October 1982, as a special program offering MS and PhD degrees to individuals with a career commitment to applying computer science and related disciplines to the field of medicine. The MIS training program was directed by Dr. Shortliffe, was co-directed by Dr. Fagan, and was overseen by a group of university faculty from several schools in the university. We graduated a total of 33 students as of 1992, 14 with PhD degrees and 19 with MS degrees. Of these graduates, 12 are now in academic faculty positions, 3 are in private practice, 1 has a hospital position, 5 are finishing their clinical training, and 12 are working in industry. The MIS program is supported by a training grant from the NLM.

Health Services Research (HSR) program

The post-doctoral fellowship program in Health Services Research (HSR) was also based in the Division of General Internal Medicine’s Section on Medical Informatics. All trainees in this program were physicians, mostly general internists, who came to Stanford to obtain specialized training in preparation for faculty positions in medical schools. Fellows generally obtained an MS degree in Health Services Research (offered through our Department of Health Research and Policy), although any MS program in the university (including the MIS degree) was a potential course of study for these trainees. Students stayed at Stanford for three years, including an intensive third research year after their formal coursework is complete. The program was supported by the Agency for Health Care Policy and Research.

On-Line Information Resources

CAMIS offered a set of on-line information resources over the Internet, through anonymous file transfer (FTP) and the then new gopher protocol. Contents included information about our local computing resources and technologies, indexed abstracts for our technical report series, the Macintosh Info-Mac software distribution database, archives of the AI-in-Medicine discussion group, information about research funding opportunities, and pointers to many other relevant information repositories on the Internet. The on-line resource with the broadest following and longest history of operation was the Info-Mac service. Info-Mac was started in 1984 as a means of sharing software for the Apple Macintosh computer among academic developers.

Video Tapes and Films

Various CAMIS groups prepared video tapes that provided an overview of the research and methodologies underlying their work and that demonstrated the capabilities of particular systems. Tapes included descriptions of the T-HELPER project, the Speech Understanding project, and the SUMMIT project. These were available to visitors and other groups upon request.

Visiting Scholars Program

We were able to accept scientific visitors for periods of six months to a year to work with us, to learn the techniques used by CAMIS investigators, and to collaborate with us on specific projects. Our ability to accommodate visitors was limited primarily by support resources — space, computing, and personnel. Each academic year we offered two positions to visitors: one for a senior investigator, generally one who was known to us professionally, and the other for a more junior person who formally applied for a one-year visitor’s position.

Annual CAMIS Symposium

Under the CAMIS resource, we initiated an annual CAMIS Symposium series, beginning in the spring of 1993. The goal of this symposium was to provide a formal focus for resource scientists to present their work, for students to develop presentations of their research, for resource members to inform those outside of CAMIS at Stanford and at other institutions about what we were doing, and for selecting a few themes for special courses or workshops in conjunction with the symposium.

Electronic Mail Hub

CAMIS offered Internet electronic mail support for appropriate members of the national medical informatics community. While we could not support mail accounts for the entire community, we did provide a reliable routing system for community members and we supported distribution lists relevant to CAMIS users and alumni, special interest groups for particular scholarly topics, and communication among members of informatics professional societies.

Advisory Council

The CAMIS research resource was managed by the principal investigator and resource director on a day-to-day basis. An Advisory Council has been appointed by the PI to review the status of the resource and to provide external advice on matters such as future directions for technology integration or development, new research opportunities, and the role of the resource in the national community to help guide the administration of CAMIS. The Advisory Council met at least once per year (in conjunction with the CAMIS symposium). Its membership included:

Joshua Lederberg, Ph.D. Chair University Professor The Rockefeller University	Bruce G. Buchanan, Ph.D. Vice-chair Professor Department of Computer Science University of Pittsburgh
Milton Corn, M.D. Ex-officio member Director of Extramural Programs National Library of Medicine	John Seely Brown, Ph.D. (1993-1994) Vice President Xerox Palo Alto Research Center
Paul Clayton, Ph.D. (1993-1995) Professor and Director Center for Medical Informatics Columbia Presbyterian Medical Center	Sherrilynne S. Fuller, Ph.D. (1993-1996) Director, Health Sciences Library and Information Center University of Washington
G. Anthony Gorry, Ph.D. (1993-1996) Vice President Research and Information Technology Rice University	Fred Richards, Ph.D. (1993-1995) Professor Molecular Biophysics and Biochemistry Yale University
Peter Szolovits, Ph.D. (1993-1994) Associate Professor Laboratory for Computer Science and Engineering Massachusetts Institute of Technology

Research Environment

Funding

The National Library of Medicine funded the CAMIS resource (under grant LM05305), including the core research projects, and the infrastructure which supported the computing needs of our collaborative research community. Each of the collaborative projects had external support from NIH or other agencies through grants and contracts.

Computing Resources

Under the aegis of the Symbolic Systems Resources Group, CAMIS developed and operated its own computing resources tailored to the needs of its individual research projects. Computing resources consisted of a networked mixture of personal workstations and shared central server computers, reflecting the evolving hardware technology available for biomedical informatics research. Central hosts were 2 Sun 4/490 servers and a Sun SPARC IPC network information server (this machine also maintained access for those familiar with the old SUMEX-AIM name). These ran the UNIX operating system and formed the core of the CAMIS resource. Research program development and routine computing functions, such as electronic mail reading and composition, text processing, and information querying, were provided by distributed user workstations. Workstations included about 100 Macintosh IIs (including IIci’s and an FX), 15 NeXT machines, 15 Sun SPARCstations, a DECstation 5000, and 3 Hewlett-Packard 720s. Additionally, several portable Macintosh PowerBooks were available as were an NCI portable pen-based computer and IBM PC clone (INTEL 80486) development system for mobile computing.

CAMIS facilities were integrated with other biomedical and computer science resources at Stanford through the campus Ethernet network (SUNet) and to external Internet resources through the Bay Area Regional Research Network (BARRNet) link to the NSFNET.

CAMIS Collaborative Research Project Summary

The following are short resumes of each of the collaborative research projects that were part of the CAMIS resource.

• Therapy-Helper Project (T-HELPER)

  The T-HELPER project is developing a graphical medical record and decision-support system for health-care workers to use in caring for AIDS patients to encourage enrollment in and compliance with protocol-based therapy. An evaluation of the system in two local AIDS clinics will assess independently the effect of the electronic record and the decision-support components on the rate of patient enrollment in protocol-based care, completeness of data collection, protocol compliance, and physician satisfaction. [Ted Shortliffe, Mark Musen, Bob Carlson (Oncology), Larry Fagan, Tom Rindfleisch, Terry Blaschke (Clinical Pharmacology), Samson Tu, Josef Schreiner, Christopher Lane, Tse Lai (Biostatistics), Elizabeth Miller, Carol Kemper (Santa Clara Valley Medical Center), Jeanette Sison (SCVMC), Stanley Deresinski (SCVMC)]

• PROTÉGÉ Project

  The PROTÉGÉ project is building tools to help in the construction of clinically useful decision-support systems. Research includes implementing a meta-level knowledge-acquisition tool that generates task-specific tools so that physicians can enter knowledge of clinical trials; exploring new problem-solving methods; evaluating PROTÉGÉ-II for entry of AIDS clinical trials protocols; and developing methods for designing new clinical protocols. [Mark Musen, Angel Puerta, Samson Tu, John Gennari]

• Reasoning under Uncertainty in Medical Decision-Support Systems

  This project is exploring the use of probability theory as a representation of uncertainty in medical diagnostic systems. Key challenges include developing tractable methods for knowledge acquisition and probabilistic inference, developing a generalizable multi-attribute utility model, and methods for providing good explanations of probabilistic inferences. The current approach is based on Bayesian belief networks. [Ted Shortliffe, Larry Fagan, Max Henrion, Blackford Middleton (General Internal Medicine), Ross Shachter (Engineering-Economic Systems)]

• VentPlan Project: Dynamic Selection of Models

  The VentPlan project is designing a data-interpretation and therapy-planning system for the intensive care unit (ICU). Basic research issues in temporal reasoning include real-time assimilation of incoming data from multiple instruments, representation of time-oriented intervals, and description of ongoing physiological processes. The system will determine the meaning of the measurements with reference to the particular clinical context and the patient’s individual response patterns, and suggest settings for the mechanical ventilator that supports the patient’s respiratory function following surgery. [Ted Shortliffe, Larry Fagan, Adam Seiver (Clinical Consultant)]

• Guardian Project

  Guardian is an intelligent-agent architecture designed to operate in intensive-care environments. The project addresses problems in real-time intelligent monitoring and control, including: continuous acquisition and interpretation of patient data; construction of short- and long-term therapy plans and monitoring the execution of such plans; diagnosis and alarm setting for unexpected disorders and complications; demonstrating agility in reaction to problems requiring immediate attention; explanation of physiologic and pathophysiologic phenomena; and closed-loop control of therapeutic devices on demand. [Barbara Hayes-Roth, Serdar Uckun, Lee Brownston, Adam Seiver (Clinical Collaborator), Juliana Barr (Research Fellow in Anesthesia, Thomas Feeley (Clinical Collaborator)]

• Spoken Input to Medical Systems

  This project explores the use of commercially available speech-recognition equipment as an interface to medical information and decision-support systems. We are using both isolated word-recognition systems and continuous-speech systems to understand when speech input is a preferred modality, and what are the relative benefits compared to other input devices, such as graphical interfaces and pen-based systems. We are experimenting with spoken interfaces for the ONCOCIN therapy-planning system and the QMR diagnostic program. [Ted Shortliffe, Larry Fagan, Christopher Lane, Smadar Shiffman]

• COPE-CAPIS Project

  This project free-text data and advanced modeling methods to assess quality improvement in medicine. Goals include developing a robust natural-language (NL) processing system to capture free-text information in medical records, and applying this system to monitoring and alerting systems; assessing the appropriateness of care using admission and discharge summaries and existing practice guidelines; and applying severity-of-illness models, based on a variety of data sources (potentially including free-text data) consistent with statistically-based continuous quality improvement paradigms advocated by Deming and Juran. [Leslie Lenert, (Medicine)]

• Bioinformatics Group

  The goal of this group is to understand the flow of information from genes to molecular structures, to biochemical function, and, finally, to biological behavior. This collaborative research project is strongly biological in character (rather than emphasizing medical informatics research per se), and draws heavily upon the core medical informatics research project, “Understanding the Structural Code of Biological Molecules”. Specific research areas include: knowledge representation and reasoning to simulate DNA metabolism; pattern recognition and sequence classification in biological-sequence databases; and use of parallel computer architectures to implement biological search. [Douglas Brutlag (Biochemistry), Russ Altman, Ross Shachter (Engineering Economic Systems), Dalit Naor (Postdoctoral Fellow in Biochemistry)]

• Helix Project

  The Helix project is interested in the analysis of macromolecular structure — especially in the analysis and characterization of the levels of structural resolution and uncertainty in these structures. We have developed a probabilistic algorithm for structure determination (from NMR data in the case of proteins, or from predicted base-pairing schemes for nucleic acids) which simultaneously determines the uncertainty in a structure and the structure itself. We are applying this algorithm to important structural problems in molecular biology, and are investigating the possibilities of parallel, distributed and networked implementations. We are also analyzing known structures to characterize the ways in which uncertainty in atomic positions affects understanding of structure/function. [Russ Altman, Oleg Jardetzky (Stanford Magnetic Resonance Laboratory), Doug Brutlag (Biochemistry), Michael Levitt (Cell Biology). Iain Johnstone (Health Research and Policy; Medicine), Ross Shachter (Engineering Economic Systems)]

• Informatics applications in gene mapping and sequencing

  The goals of this project are to apply statistical and computational methods to the development of new technologies for DNA sequencing and new methods for gene mapping, in collaboration with genome researchers at Stanford. [Michael Walker, David Botstein (Genetics), Ron Davis (Biochemistry), Pat Brown (Howard Hughes Medical Institute)]

• How Things Work: Knowledge-Based Modeling of Physical Devices

  The How Things Work (HTW) project seeks to develop representation techniques for encoding knowledge about engineered devices in a form that enables the knowledge to be used in multiple systems for multiple reasoning tasks. It also seeks to develop reasoning methods that enable the encoded knowledge to be effectively applied to the performance of the core engineering task of simulating and analyzing device behavior. [Edward Feigenbaum (Computer Science), Richard Fikes (Computer Science), Robert Engelmore (Computer Science), Thomas Gruber (Computer Science), Yumi Iwasaki (Computer Science), James Rice (Computer Science)]

• Penguin Project

  The Penguin project seeks to demonstrate a workstation for researchers dealing with biomedical experiments and information. Initial work focuses on making information for multiple types of experiment designers sharable by providing an object-oriented presentation of information stored in sharable relational databases, suitable for use in expert systems that aid experimenters. Future work will synthesize the use of database and artificial-intelligence techniques into expert database systems (EDSs) resulting in a set of mediators —software modules that exploit encoded knowledge about some sets or subsets of data to create information. [Gio Wiederhold (Medicine and Computer Science), Arthur Keller (Computer Science), Larry Fagan, Lee Herzenberg (Genetics), Kincho Law (Civil Engineering), Toshiyuki Matsushima (Genetics), Ernest Wood (Center for Integrated Systems)]

• Federated Data Base Project

  Federated databases are autonomous, cooperating, information resources that cannot be directly integrated, both because of inherent differences in their deep models and ontologies, and because of the need to keep maintenance local and distinct. Our research focuses on dealing with the problems of semantic mismatch among distributed knowledge and databases, and the management of the large collections of knowledge needed to abstract and merge information from federated resources. [Gio Wiederhold (Computer Science), Arthur Keller (Computer Science), Peter Rathmann (Computer Science), Yuri Breitbart (University of Kentucky)]

• Health Economics and Health-Policy Research

  The health economics and health policy research project aims to assess the cost-effectiveness of health interventions and factors associated with the utilization of nursing home care. [Alan Garber (Medicine), Thomas MaCurdy (Hoover Institution), Victor Fuchs (Economics; Health Research and Policy), Richard Olshen (Health Research and Policy)]

• HIV Technology Assessment Project

  The HIV Technology Assessment Group is a collaborative project to develop an approach to comprehensive technology assessment beginning with the evaluation of the performance of diagnostic tests and extending through the production of customized screening and/or practice guidelines based on formal, normative decision models. The initial application domain is diagnostic and screening recommendations for HIV infection. Research issues include the development of new approaches for technology assessment and methods for producing screening and practice guidelines. [Douglas Owens (Medicine), Alan Garber (Medicine), Margaret Brandeau (Medicine), Robert Nease (Dartmouth Medical School), J. Sanford Schwartz (University of Pennsylvania), Bruce Kinosian (University of Pennsylvania), Mark Holodniy (Infectious Disease Consultant), Lincoln Moses (Statistical Consultant)]

• Lane Library – Networked Services

  Lane Library is the single, centrally supported library and information service for Stanford University Medical Center. The aim of Lane’s Networked Services is to bring electronic library-based information services to its user constituency when and where these services are needed. In the context of the rapid proliferation of network communications and computer workstation tools, this is a natural extension of traditional library services. [Peter Stangl (Director, Lane Library), Dick Miller (Head of Technical Services, Lane Library), Randy Woelfel (Lane Library)]

• Stanford University Medical Media and Information Technologies (SUMMIT) Project

  The Stanford University Medical Media and Information Technologies (SUMMIT) project is a research and development project that is exploring new methods of exchanging, organizing and presenting information, particularly in the areas of medical research and education. Our interests encompass all aspects of medicine, including interactive educational media, computer-driven surgical aids, and research into computer-assisted patient care. Specific project goals are to create technology to help faculty and researchers explore the interactive presentation of information, and to help the medical community explore electronic information exchange. [Parvati Dev (Office of Medical Students and Graduate Affairs), Ramon Felciano (Office of Medical Students and Graduate Affairs), Phil Constantinou (Office of Medical Students and Graduate Affairs)]