Philadelphia Reflections

The Wedding of Computers and Medicine: First Annual Fuller G. Sherman Lecture George Ross. Fisher, M.D.

The Wedding of Computers and Medicine:

First Annual Fuller G. Sherman Lecture

George Ross. Fisher, M.D.

October 1, 1987

Fuller G. Sherman, M.D. was born August 15, 1894, graduated with academic distinction from Jefferson Medical College in the class of 19** with his second doctorate degree, was certified by the American Board of Internal Medicine, practiced for many years in Woodbury New Jersey, and retired from practice in 19** to live in his native state of Maine.

The competitive strengths of Dr. Sherman’s character have actually been easier to see during the so-far thirty years of his retirement from medicine. Past the age of 90, he attended Bowdoin College, taking courses in Shakespeare and geology, plays par golf, holds a Masters’ certificate in tournament bridge, is a distinguished cabinet maker, and does creditable work in oil painting. Two notable achievements were once, on the day after his graduation, to have flattened an Associate Dean of this Medical School with a single punch; and secondly to have consistently outperformed the Dow Jones Industrial Average on the New York Stock Exchange. Because of the latter, of course, he was able to endow the lectureship we inaugurate today. In both of these adventures, he illustrated the truism that in life, everything is a matter of timing.

He has been my teacher, employer, referring physician, and friend. He is now my patient, allowing me to judge he has as good a chance as any of us to live another thirteen years. If he does, he will have the almost unheard-of opportunity to observe the practice of medicine in three different centuries. There can be little doubt he will study the next century harder than any of us, as his patronage of computer science demonstrates today.

My subject has three parts: yesterday, today, and tomorrow. The exhilarating nature of the computer world lies in only a little yesterday, a little of today, and a great deal of tomorrow. For our purposes, yesterday began about thirty-five years ago when the Chairman of the International Business Machines Corporation, Thomas Watson, made the decision to gamble the whole future of his successful typewriter and tabular company on mass-producing computers. There were then only a few dozen of those machines in existence, mostly owned by the military. They cost millions and were expensive to operate. Typically, a bushel of worn-out vacuum tubes were replaced every day. You could walk around inside them without stooping over. By 1960, IBM was selling a thousand of these machines a month to large corporations for about $ 4 million apiece. Technology in 1960 had greatly reduced the maintenance cost, but the University of Pennsylvania still had to rent them for $300 an hour at the academic rate. Machines of equal power can today be purchased for a thousand dollars, and are the size of typewriters. I own five of them; there are about twelve million others in existence, up from nine million last year. One surgeon recently told me he bought one of the best, about a year ago, but had not yet had time to take it out of the box. The cost of these miraculous machines was thus trivialized in a single generation, and each year the Sunday supplements have promised us that within two years, five at most, such things as a medical diagnosis would be relegated to computers. It never happened, of course, because science fiction writers had not heard Dr. Sherman’s professor Thomas McCrea (Dr. Maddrey’s predecessor by seven) repeatedly intone that “most diagnoses are missed because the doctor didn’t look not because he didn’t know”. The problem of diagnosis today, as then, is one of information gathering, not information manipulation.

A generalization can be offered. If you hear a prediction about computers, be fairly certain it will never happen, unless it already exists. So many brilliant minds are at work, with financial rewards providing unlimited resources, that the immediately achievable is achieved immediately. Mr. William Gates, a self-made billionaire at the age of 31, illustrates how among people who are successful in this field, there is no motivation for idle chatter.

The amazing drop in the cost of computers has made it possible to have a personal computer that is those dedicated to use by a single person. Personal, or stand-alone computers, can now do almost everything a large main-frame computer can do except cope with multitudes of users. But, having only a single master, they cater solely to his needs and undergo an unexpected transformation into tools not appropriate to big shared machines, becoming extensions of one user’s brain. Word-processors and spreadsheets transform the way we think and work; such generalized mind-expenders prove to be more powerful than programs which merely calculate acid-based balance or remind us of potential adverse drug interactions.

Word processing is a utility as revolutionary as Guttenberg’s invention of movable type; it can be expected to raise the standards of thought just as much as the standard of typing. A program costing less than $200 permits preliminary display on the cathode ray tube, where prose can be corrected and modified repeatedly before it is printed on paper. The machine will find all spelling errors and most grammatical errors, permit any character, paragraph or page to be replaced, repositioned or erased. It will index the material, automatically insert hyphens place footnotes and references in place, and allow unlimited experimentation with different margins, page size or paragraphing. When finally printed on paper, the right margin can be automatically justified, and the words become unified into important than these aesthetic advantages, word processing permits the author to revise repeatedly until what he writes finally says what he means.

A second innovative creation on a personal computer is a hybrid of two steps, the generalized data management system, and the spreadsheet. Many small stored globlets of information are aggregated on request, meaningful. If the unit of data is a single patient, with blue eyes can then be effortlessly linked with any other glob of information, such as antibodies to retroviruses. The spreadsheet concept then organizes such data cells into rows and columns which can be fed into formulas which operate serially on every row in a column, generating a new column of derivatives. The user need not, in fact commonly does not know statistical theory, but for example, can process anyway to command regression analysis on eye color and AIDS or any more plausible hypothesis in clinical research. These programs will then transform selected numbers into colored graphs on request (slide). The ability to use statistical tools without understanding them will, of course, create abuses of this system, which in the case of regression analysis would be to overemphasize the validity of 95% confidence limits. Ultimately, the value of the computer product will depend on the brainpower of the individual user. When convincingly packaged data can be processed in massive amounts by chimpanzees at the keyboard of $1000 machines, it is a little daunting to await the misinformation which will be generated by the 5% error content of mountains of data. By the rules of regression analysis, one conclusion in twenty will be reached by the operation of chance alone. Since editors are intrigued by papers which reach unexpected conclusions, thoroughly documented spreadsheets research which later turns into smoke will someday be their proper torment. The exciting future of computers is thus not to replace doctors in some profession-threatening way, but rather to extend the capabilities of their minds in powerful ways which continue to reflect the personality of the user. As has been said of corporations, the dedicated personal computer projects the lengthened shadows of the man.

Early steps in that direction would please Adam Smith, producing exalted results from trivial and mercenary motivations. Whether you like it or not, and whether cost-effective or not, the of medical practice with insurance and reimbursement is making it essential for every practicing physician to employ a computer; those who avoid it have done so out of fear of the disruptions, not because they deny the value of the office machine. Once the machine is installed, word processing is seen as a free bonus, and the financial affairs of the practice become raw materials for database program and spreadsheets. In this way, the practicing physician acquires a mind-expander when he merely sought to reduce his clerical expenses. He also, by way, merges into the mainstream of business computing, and like everyone else, will find that the new IBM model 50 has become the modern standard, just as the IBM Selectric typewriter became the business standard thirty years earlier.

To understand why this is so, notice that IBM continuously spends $5-6 billion annually on research but withholds most new products from the market. Then, about every seven years, a bundled package of innovations is released as a “new generation” which then makes existing machines obsolete and dominates the field for the next seven years. Throughout the seven-year gestational period, other companies also bring forward innovations but must recover their costs by releasing them immediately. IBM watches market reactions, preparing to submerge would-be pioneers in a tidal wave of releases. In effect, other companies test the market for IBM to exploit. Almost every component of the 1987 new generation is new and very little of it is unique to IBM (slide). However, an irresistible market standard is created when many innovations are released at once by the largest volume producer. The new 1987 machines seem mainly designed to permit for the first time several personal computer users to share one octopus machine, a mildly useful thing for the doctor and his secretaries. The main importance of using 32-bit technology lies in the fact that the doctor’s personal computer uses the same system and thus can talk the same language as the mainframes in hospitals, laboratories, insurance companies, and the Internal Revenue Service. The profession must not let itself get lost in the chatter of intercomputer communications; while he must adapt to the equipment that is available, the physician mostly needs many different mind-expanding applications on a single machine. The creation of a market standard will almost surely prove to be a dominant force even for physicians. The IBM model 50 will be what to buy until 1993, as will IBM stock, but physicians need to establish their own culture within the commercially available environment.

1993 will not, of course, be a far enough horizon for Dr. Sherman’s third century, so we might look across the valley over two intervening mountains. Fourteen years from now, the new 2001 models will also be composed of standardized refinements of whatever exciting advances may accumulate in the meantime. By that time, computers should be able to accept voice dictation since they can already understand a spoken vocabulary of about 1000 words. Computer scanners can now read pages of typing with 90% accuracy, so we can except much less typing for accession of much larger volumes of day-to-day information. I understand x-ray films require a resolution of 2000 by 1600 pixels; since advanced computer screens already achieve 1600 by 1200 pixels, the silver-coated films we use now should likely disappear.

In the shorter term, it would be a fair prediction that the exciting programs of the next seven years coming up will exploit the telecommunication power of 32-bit processing and the vast storage capacity of CD ROM’s. Large Multimillion dollar mainframe computers operate in units of 32, but personal computers now mostly use units of 16. Declining prices of transistor chips make 32-bit technology affordable for personal computers, so we can expect to see the doctor’s computer talking on equal terms with the big computer in the hospital, the drug store, and the Library of Medicine. It is exciting to predict role reversal, with once imperious mainframe owners outmaneuvered by users of agile PCs. Organizational whales like hospitals, department stores, banks, and the Internal Revenue Service have had their day of forcing everyone else to conform to their convenience; the little piranha fish will grow sharp teeth.

Since the profession of Medicine is after all in the knowledge business, it is breathtaking to contemplate the migration of medical journals and libraries from paper to electronic medium and the dissemination of libraries to the doctor’s consultation room. Compact disc technology is already the cheapest form of information storage, whose inevitable price decline has scarcely begun (slide). Grolier’s encyclopedia is now available on a plastic disc you could put in your shirt pocket; the entire encyclopedia only takes up 20% of the disc. This unerasable form of storage is sometimes called WORM (write once, read many). Inexpensive scanners can convert pages of print to a computer file in 20 seconds; rather accurate programs can convert foreign language to pidgin English. Between the two processes, whose only present limitation is price, whole libraries will surely soon be swallowed up on plastic disc, and medical journals may appear in that form as soon as someone figures out how to incorporate drug advertising. However, don’t run out and buy a CD-ROM just yet; there are twenty different types and they are incompatible with each other. The exasperating power of IBM is well illustrated by the fact that this titanic information storage revolution will not take a step forward until someone like IBM is able to impose a standard which will disciple the present Tower of Babel.

In closing, the point must be made that the main hindrances to adoption of computers by the medical profession will not be technical, they will be sociological. “Not in my back yard” is the Spirit with which most new things are greeted, even in the learned professions. A plain fact of human behavior is that how you stand is determined by where you sit. For years, banks have transferred patient payments to physician bank accounts without the creation of a single piece of paper. But electronic funds transfer has made very limited progress in 20 years, primarily because the payers and their banks do not wish to surrender the interest float which develops during the delay of transfer. Blue Shield of Pennsylvania has over a million floating dollars earn interest at all times while the obsolete paper check depositing process limps on. The videotape machines which are everywhere provide a warning example of how technical potential is easily frustrated. Instead of ten thousand college professors giving mediocre lectures on Hamlet, it is clear that some professor at Oxford could give the very best lecture on videotape which all students everywhere could watch at home without even paying tuition. Since it hasn’t happened, and it won’t happen, perhaps the point I am striving to make becomes clear.

Try to image the resistance which pharmacists would create to electronic drug ordering; indeed, the nursing profession is very resistant to physician orders which come in any way except handwriting on the floor chart. While I cannot identify the economic incentive which explains the delay, Jefferson Hospital has just installed a system in which laboratory results are instantaneously transmitted to the clinical floors. However, a similar system was installed at the old Philadelphia General Hospital in 1965. These and many other examples of apparently irrational delays most likely have their explanations in the motivation of people rather than the limitations of machines. Therefore, in predicting a revolution in medical information handling we must not, for example, underestimate the capability of printers and typesetters at the New England Journal of Medicine to hold up electronic publishing, or the librarians of the world to resist the destruction of their careers by plastic disks. IBM is in the business of setting bridal supplies and has repeatedly proved to be a shrewd judge of bride psychology. They obviously believe 1993 will be soon enough for the real wedding of medicine and computers; maybe it can wait for the year 2000. Meanwhile, it will not matter much that the bride’s father can easily afford the wedding, or the groom is anxious to perform. Medicine, the bride-to-be, hasn’t yet said “yes”.

Originally published: Tuesday, June 13, 2017; most-recently modified: Monday, June 03, 2019