By doing this, the hospital will be extending its newest and most striking trend, which is to foster innovation, later to be picked up by other, nonac-ademic institutions. The absurd end-point of such a trend would be for the hospital to direct personally the diagnosis and therapy of a patient who never enters the hospital. Absurd as it may be, it is already happening in the case of many patients treated at Logan Airport. It will happen more often, in other ways, in the future.

Of the almost limitless spectrum of potential technological advance, we can concentrate here on two areas of imminent advance, television and computers. One ought to say that they have been imminent for a long time; a decade ago one heard that computers were about to revolutionize medicine, and one still hears it today. It obviously hasn't happened yet. Indeed, neither television nor the computer has made much difference yet to routine hospital functioning. Television is employed on occasion for student teaching; it is used in a small way for dispatching blood samples and other items; it has some application in X-ray technology, in terms of image-intensification systems. Computers remain primarily the plaything of researchers. At the MGH there is now a computer program to help in running the clinical chemistry lab, and a computer to help in billing and patient record-keeping, but the computer and television as direct aids in patient care have not made their appearance.

In contrast, the Tele-Diagnosis system at Logan Airport uses computers and TV in direct confrontation with the patient. The system is expensive and in some ways primitive. Also, its present thrust is diagnostic; therapy, the steps following diagnosis, will still be directly carried out by a doctor, nurse, or the patient himself. There are no machines to do this, unless one stretches the definition to include renal-dialysis machines, exercise machines, and the like.

In general, diagnostic automation appears much closer than therapeutic automation-and is much more readily acceptable to physicians. Consider, then, diagnostic automation first.

The first and most striking feature of the Logan system is that diagnosis can occur at a distance. The doctor's stethoscope is three miles long. But, oddly, that diagnosis at a distance is very old and has some humorous elements. Beginning around

a.d. 900, for example, the practice of uroscopy, or "water casting," came into vogue. It was felt that the amount of information obtainable from inspection of urine was unlimited. The urine of a sick man was often sent many miles to be examined by a prominent physician.

David Riesman cites a typical medieval interpretation of urine:

The urine is pale pink, thick above, thin below, becoming gray or dark toward the surface. The grayness and obscurity is caused by overheating of the material. The symptoms are these: pain in the head, especially in the temples, sourness of the breath, pains in the back from bile descending to the loins and kidneys, with paroxysms every day or every second day, usually coming on after dinner time.

In medieval literature there are many discussions of the hazards to the physician of uroscopy; even in those days, diagnosis at a distance had its risks. The Spanish physician Arnold of Villanova, who lived in the thirteenth century, wrote:

With regard to urines, we must consider the precautions to protect ourselves against people who wish to deceive us. The very first shall consist in finding out whether the urine be of man or of another animal or another fluid.

The second precaution is with regard to the individual who brings the urine. You must look at him sharply and keep your eyes straight on him or on his face; and if he wishes to deceive you he will start laughing, or the color of his face will change, and then you must curse him forever and in all eternity.

The third precaution is also with regard to the individual who brings the urine, whether man or woman, for you must see whether he or she is pale, and after you have ascertained that this is the individual's urine, say to him: "Verily, this urine resembles you," and talk about the pallor, because immediately you will hear all about his illness…

The fourth precaution is with regard to sex. An old woman wants to have your opinion. You inquire whose urine it is, and the old woman will say to you: "Don't you know it?" Then look at her in a certain way from the corner of your eye, and ask: "What relation is it of yours?" And if she is not too crooked, she will say that the patient is a male or female relation, or something from which you can distinguish the sex… Or ask what the patient used to do when he was in good health, and from the patient's doing you can recognize or deduce the sex…

The list continues through nineteen precautions, all designed to enable the physician to pry information from the person bringing the urine, and to prevent deception. Arnold was not above a little deception himself, however:

You may not find anything about the case. Then say that he has an obstruction of the liver, and particularly use the word, obstruction, because they do not understand what it means, and it helps greatly that a term is not understood by the people.

The modern counterpart of this medieval guessing game over urine is the telephone conversation between physician and patient. For years after the telephone became common, physicians resisted making telephone diagnoses, and they still frown on them. But every practicing doctor now spends a substantial part of his day talking to patients on the phone, and he is resigned to making a large number of decisions, some of them uneasily, by

phone.

Closed-circuit television, while far from the ideal of a personal examination, is vastly superior to the telephone alone, and in many cases it is surprisingly adequate. This does not mean that future patients will all be seen by closed-circuit television, with neither doctor nor patient leaving home. What it does mean is that television will probably work in certain very special applications. One of these is the Logan application-providing a doctor to a clinic during low-use periods. Another obvious use would be specialist consultations. A hospital or clinic that needs a neurologist only a few times a year cannot afford to staff one. Nor could it find one, even if it could afford it. Television is perfectly suited to such consultation.

At the same time, a system such as that at Logan makes possible a routine physical examination, but goes no further-and there are suggestions that technology will ultimately change the very nature of physical examination. Here the historical trend is clear.

Consider the innovations in physical diagnosis. In the nineteenth century, there were three of great importance-the stethoscope, the blood-pressure cuff, and the thermometer. Each of these is really nothing more than a precise way to determine what can be inaccurately determined by other means. Thus the thermometer is superior to the hand on the forehead; the stethoscope superior to the ear against the chest [For the purposes of this argument, I will ignore the fact that the stethoscope really initiated auscultation as a useful examination procedure. In truth, ears were not pressed against the chest with much regularity before Laennec invented the stethoscope and described auscultation.]; and the blood-pressure cuff superior to a finger compressing the artery to test its pressure.

Now, the first two advances of the twentieth century were quite different: the X ray and electrocardiogram provided new information not obtainable by physical contact. No amount of squeezing and touching the patient will tell you anything directly about the electrical currents in his heart. You may deduce this information from other findings, but you cannot extract it directly. Similarly, X rays represent a new kind of vision, providing a new kind of information.

At the present time a variety of examination procedures are being tested. These include ther-mography, ultraviolet light, ultrasonic sound, as well as mapping electrical currents in the skin. Except for thermography, these all represent "new" sensory information for the doctor.

Thus the initial trend was to measure the patient more exactly, and later, to measure the patient in new ways. The first approach has been to find new sorts of measurements and new sensory information. But a second approach, now in its infancy, concerns translation of old information into new forms. The computer will be helpful here in a number of ways, in producing what is called "derivative information."

In a simple way, this is already being done. The human computer [Defmed as the only computer that can be produced by unskilled labor] and the electrocardiogram are a clear example. The electrocardiogram measures electrical currents within the heart muscle-the current that makes it contract and beat. Often, when a physician looks at an electrocardiogram, he wants specific electrical information. He wants to know about rate and rhythm, about conduction of impulses, and so on. At other times, he wants nonelectrical information. He may want to know how thick a part of the heart wall is, for instance. In this case, he derives the information from the electrical information.

But there are more complex forms of derived in-

formation. A physician examining a patient with heart disease may be interested in knowing the cardiac output-exactly how much blood the heart is pumping per minute. This is the product of heart rate (easily determined) and volume of blood ejected per beat (very difficult to determine). Because cardiac output is so hard to assess, it is not much used in diagnosis and therapy. However, by measuring heart rate and the shape of the arterial pulse (both easily done) a computer can calculate cardiac output and can perform these calculations continuously over a period of days, if necessary. If a physician needs to know cardiac output, he can have this information. He can have it for as long as the patient is connected to the computer.

Does the physician really need cardiac output? At the moment, he can't be sure. For centuries he's had to content himself with other information. There is reason to believe, however, that cardiac output will be useful in a variety of ways, as will other derived information.

An interesting technological application concerns the reverse of the coin: determining which information the physician already has but does not need. This is not to say that the information is inaccurate, but only that it does not have diagnostic significance and is therefore not worth obtaining. At present, the physician naturally tries to avoid gathering useless information, but in certain circumstances he cannot perform as well as a computer. Multiple discriminant analysis is a case in point. As one observer notes, "There is a limita-