Perspectives in Biology and Medicine

ATTEMPTS TO USE COMPUTERS AS DIAGNOSTIC AIDS IN MEDICAL DECISION MAKING: A THIRTY-YEAR EXPERIENCE RALPH L. ENGLE, JR.* Introduction In the 1950s, when electronic computers were first coming into use, there was already great interest in having the computer perform functions that had previously been performed only by the human brain. People were impressed by the memory, speed, and logical manipulations of the computer. Some felt that in decision making computers would rival the human brain. Some were even concerned that the computer of the future would be making all of the important decisions in certain fields. Still others felt that computers would never be able to match the decision-making capabilities of the brain. Yet all agreed that computers would be useful in various ways to decision making. We still do not have final answers to this basic question of the computer's ability to make decisions, and most experts have not changed their minds about it. Computers have found an important role as aids to decision making in the field of radiology, for example, where many modern imaging techniques depend entirely on their capabilities. Also, computers as analytic tools have been used to aid in decision making in many fields of medicine, as in the interpretation of electrocardiograms. Even less direct Over the years a number of people not mentioned in the text have contributed in important ways to the project. Drs. Leo Leveridge and George Ubogy participated as members of the group during 1967-1970. Programmers included Doris Cavalieri and Margaret Hurley. Gary Novick, Deborah Lasher, and Paul Aisen were summer student employees of IBM and participated actively in developing and testing computer programs. Consultants to the project during the early stages included Dr. Saul Amarel of RCA and Rutgers, Dr. Ivan Sublette of RCA, Dr. Gerald Goertzel of IBM, and Dr. Aubey Rotenberg , director of the Biomathematics Computer Center of Cornell University Medical College. *Departments of Public Health and Medicine, Cornell University Medical College. Address: 1 Country Club Lane, Pelham Manor, New York 10803.© 1992 by The University of Chicago. AU rights reserved. 003 1-5982/92/3502-0767$0 1 .00 Perspectives in Biology and Medicine, 35, 2 ¦ Winter 1992 207 aids are those designed to find relevant bibliographic sources. However, these applications, important as they are, fall short of the capabilities of physicians confronted with diagnostic decisions. Over the past 30 years our group has tried to develop computer programs to aid physicians as they make decisions about diagnosis in the medical subspecialty of hematology. Some members of the group conceived that the programs would eventually have all of the capabilities of a trained physician-hematologist. Others hoped that the programs would be close enough to a physician's thought process to be useful as a benchmark against which the physician-diagnostician could compare his or her own thought processes. All agreed that it would not be difficult to simply list certain pieces of relevant information. The experiments were designed from the start to mimic a physician's diagnostic thought processes in order to test the most difficult hypothesis. All of the programs made use of one or another decision-making algorithm. Documentation of the history of this project is justified by the light it sheds on the difficult process of instructing a computer to function as the human brain functions in decision making. Although, over the years, there have been frequent press announcements of medical diagnosis by computer, there are still no generally accepted, successful systems in use except in an experimental way. Early History ofProject and Work of Others In 1955, Dr. Martin Lipkin invited me to join a group of computer specialists and physicians meeting at the Rockefeller Institute to discuss and develop ways in which electronic computers could be used to help physicians. This group was inspired by Dr. Vladimir K. Zworykin, inventor of the iconoscope that became the basis for the television camera. Zworykin had recently retired from RCA and had established the Medical Electronics Center at the Rockefeller Institute for Medical Research, now the Rockefeller University. Although Zworykin had broad interests, the group decided to focus on computer aids to medical decision making , particularly diagnosis, since this...