Fifty Years in Medical Informatics
Total Page:16
File Type:pdf, Size:1020Kb
174 © 2006 IMIA and Schattauer GmbH Fifty Years in Medical Informatics Morris Frank Collen Director Emeritus, Division of Research, Kaiser Permanente, Oakland, CA, USA Summary Life has brought me many nice sur- At the end of the war, I stayed with Objectives: An overview of personal experiences in medical prises. One was being invited by my Kaiser Permanente. I enjoyed being in informatics based on Dr. Morris Collen’s 50 years of research in the long-time friends, Reinhold Haux and an organization that allowed me to field. Casimir Kulikowski, to write this article practice medicine, establish a residency Methods: A personal reminiscence and historical overview, focusing for the 2006 IMIA Yearbook and share training program, and do research, just on the first two decades of medical informatics, when Dr. Collen my personal reminiscences. As I reflect as I had hoped to do in academia. began working with Dr. Sidney Garfield, the founder of Kaiser on my 50 years of involvement in medi- Garfield knew that I had training in Permanente, leading to his involvement in computer-based medical cal informatics, a myriad of faces e- electrical engineering, so in 1961 he care, through the development of the pioneering Automated merge, and I recall many people that I asked me to go to a conference on Multiphasic Health Testing (AMHT) system, which they introduced admire and respect for their significant biomedical electronics in New York into Kaiser clinics in Oakland and San Francisco. contributions to the field. My own ca- City. He thought the time had come for Results: Statistical models for medical decision-making based on consultations with Jerzy Neyman and George Dantzig were reer owes much to these worthy people. doctors to use computers in patient incorporated into the AMHT, and tested on a large database of I feel that I was born with an interest care. Of course, he was right; and I cases. Meetings with other pioneers in medical informatics at the in data, beginning with my birth date, found this conference to be very excit- Karolinska Institute led to the formation of the early society Salutas a series of three consecutive, two-digit ing. On my return, Garfield established Unitas, and the many national and international collaborations numbers: (mo-dd-yr) 11-12-13. As a the Department of Medical Methods which followed during the first two decades helped coalesce the field teen-ager, I enjoyed playing with Research (MMR) in Oakland, Cali- as clinicians and researchers investigated problems of medical data, electric gadgets, so I decided to study fornia, with the primary objective of decision support, and laboratory, hospital, and library information electrical engineering and then go on developing computer-based applica- systems. to graduate school to get a doctorate. tions for patient care. Conclusion: Dr. Collen’s research and his many medical informatics However, my life changed when I met On considering where to start, I decided activities significantly contributed to the growth of the field. The U.S. contributions are covered extensively in his book, A History of and married a nurse, named Bobbie, that it would be easiest and safest to Medical Informatics in the United States, 1950-1990. Washington, who advised me to go to medical school. first use computers in the process of DC: Am Med Informatics Association 1995. Although I planned to end up in aca- examining persons who come in for a Haux R, Kulikowski C, editors. IMIA Yearbook of Medical Informatics demia, my life again changed dramat- health checkup. The traditional health 2006. Methods Inf Med 2006; 45 Suppl 1: S174-9. ically while I was completing my medi- evaluation is a fairly routine and repe- cal residency at the Los Angeles County titive process; and for about 80% of Keywords General Hospital, and Pearl Harbor was Kaiser Permanente health-checkup History of medical informatics, Kaiser Permanente Automated bombed on December 7, 1941. Because examinees we usually did not find any Multiphasic Health Testing System, pioneering research in medical of my bronchial asthma, I did not join significant abnormalities. Furthermore, decision-making, professional organizations and meetings in the group of County Hospital physicians Garfield was committed to preventive medical informatics, evolution of research and leaders in medical sent to India. Instead, I worked for medicine and he advocated periodic Sidney Garfield in the Kaiser shipyards health checkups that could categorize in Richmond, California. Second only examinees as “well”, “worried-well”, to my wife, Garfield had the greatest “asymptomatic sick”, or “sick”[1]. So influence on my career. Trained as a I decided to develop a method for auto- surgeon, he was the visionary who, with mating as much as possible the multiple the support of industrialist Henry J. components of an automated multi- Kaiser, created Kaiser Permanente and phasic health testing (AMHT) system. instilled it with the principles of pre- To develop such a system I needed to paid, group practice, medical care. study some automated clinical labora- IMIA Yearbook of Medical Informatics 2006 175 Fifty Years in Medical Informatics tory systems. I found to be very help- cal for us to have eight of these ma- at Beth Israel Hospital in Boston, where ful the writings of George Williams and chines running side-by-side to do the Howard Bleich developed his famous his associates at the NIH Clinical Cen- eight chemistry tests I wanted, so I asked acid-base algorithms. Charles Safran ter, who developed one of the earliest Whitehead if he could make me an and Robert Greenes joined them to de- comprehensive laboratory systems. eight-channel, automated chemical ana- velop their impressive Beth Israel Hos- I soon visited Octo Barnett, who had lyzer. After a few moments of thought, pital information system. developed the laboratory system at the he asked me how much was budgeted We initiated the first automated multi- Massachusetts General Hospital for this. I replied only $25,000; to my phasic health testing (AMHT) system (MGH). I heard that the administrator surprise, he agreed to make it. The first in our Oakland medical center in 1962 of MGH found the laboratory system Technicon multichannel AutoAnalyzer [2]. My first encounter with a computer to be so efficient he would never give was delivered several weeks later and was with an IBM 1440 installed in an it up, no matter what happened to their became fully operational in our Oak- air-conditioned basement room with a other computer systems. A brilliant land AMHT system. Soon he delivered myriad of cables on the floor. We gen- informatician, Barnett worked with his a second machine in our San Francisco erated a deck of punched cards for each associates at MGH to develop MUMPS, AMHT system. We used these machines examinee; and as they went from one which became one of the most com- for many years, until Garfield pur- testing station to another, the test re- monly used programming languages; chased an AutoChemist, developed by sults were entered onto the cards. On and COSTAR, a computer-based ambu- Gunnar Jungner in Sweden, that used completing the health checkup, an latory record system for the Harvard piston pumps to keep the serum speci- online printout of a summary report Community Health Plan. Barnett also mens separate; and could do up to 40 listed any abnormal tests found and trained many leaders in medical infor- different chemical analyses. advised secondary follow-up tests to be matics, including Robert Greenes and We tested several modes of conducting performed by using simple decision Jerome Grossman. Octo and I became self-administered patient question- rules: “If this abnormality is present, good friends, who could disagree on naires, and settled on using a sort-box then do this test.” Suggested diagnoses many items, such as the best approach of punched cards with a question on for the physicians to consider were also to developing a medical information sy- each card. In the 1960s, when I was printed on the summary report. Pro- stem. He took a subsystem “modular” visiting my daughter, Roberta, then a fessors George Dantzig (called the “fa- approach, whereas I advocated a “total medical student at the University of ther of linear programming”) and Jerzy systems” approach. History has proven Wisconsin, I met Warner Slack, who Neyman met regularly with us at the his approach was better. I learned from was developing a computer-based, self- University of California at Berkeley him that nurses are the best informa- administered, patient questionnaire us- during the 1960s and helped to develop tion systems analysts, and have ben- ing a LINC computer. The LINC was our approach to providing computer- efited over the years from conversations developed at the Lincoln Laboratory at aided diagnoses [3]. Although most oth- with Rita Zielstorff, Virginia Saba, Judy the Massachusetts Institute of Tech- ers used Bayes’ conditional probability Ozbolt, Patricia Brennan, Kathleen nology by Wesley Clark and Charles theorem, we used a likelihood ratio McCormick, and Susan Grobe. Molner. The LINC computer became approach because we did not always I needed an automated chemical ana- the basis of the PDP minicomputer se- know the prior probability (prevalence) lyzer for the AMHT system being devel- ries that helped to advance medical of a diagnosis Bayes’ theorem required. oped in our Oakland medical center, so informatics. Slack described watching After the completion of the multiphasic I visited Edwin Whitehead, the presi- his secretary fill out an early version health checkup, all of the patient test dent of Technicon Corporation in of the questionnaire. When she came data were stored in the multiphasic da- Tarrytown, New York. Technicon had to questions related to sexual activities, tabase.