In previous posts I remarked that computers are taking over cognitive functions and how in many ways they have surpassed human abilities, especially with rapid calculations and data storage.  Computers have beaten human champions in such games as chess and Jeopardy. Given computers’ superior abilities in certain cognitive spheres,  it is fair to ask whether they should remain human servants, or if, under certain circumstances,  they should be allowed determine and control human behaviors. Almost everyone would counter that this is a question for science fiction in the distant future.  We needn’t concern ourselves with for a long time to come. I am here to tell you that day has already arrived.

In medicine we know how useful computers have become for data storage and especially in my area of neurology, computers have revolutionized brain imaging.  But in my hospital work, I am struck by the degree that computers dictate not only how to behave but how to think and even practice medicine. Much of medical practice is driven by protocols and it is very difficult for the clinician to depart from these in any meaningful way.  Computers are designed to guarantee this and allow for very little deviation.   Most of my day as a clinician is spent conforming to computer’s dictates, both in patent description (consultation) and in ordering therapies. Examples abound. As a specialist and clinical expert I try to hone in on a diagnostic problem. For example, I might find a genetic disease that runs through the mother’s family and describe affected family members that leads to that patient’s diagnosis.and talk about the patient’s sibs.  Everyone else who sees the patient will lazily report the mother and father had no relevant illness. They probably did not even bother to ask. In that case my extensive rendering of the mother’s family will be viewed as a deficiency for failing to mention the father. Or I might write a long paragraph describing  nystagmus and eye movements in detail in a patent with multiple sclerosis that makes the diagnosis, while colleagues will not notice anything at all and merely report all cranial nerves as being “normal.”  In that case my chart will be called into question for failing to mention the fifth cranial nerve since that does control eye movements.  Computers look for all categories of minutiae but generally do not attach values to a data stream and can’t evaluate relevance, only presence of certain data elements. For that you need an expert system, namely a trained human person. Frustratingly working with computers, most of us are asked to surrender finely honed judgment at the door, or worse, our cognitive skills are not valued at all. They ask us to be robots, fill in the form like we told you to. The people making the demands are often not doctors either, but managers.

A lot of people who read my patient notes come up to me to tell me how much enjoy reading them as they tend to be colorful and pithily describe my insights as to the patient’s disease state and condition. I hope they learn what is really going on in those notes as I think about them and take pride in what I do. I am not writing this at all  to brag about my notes, but only to make a critical point. As a seasoned clinician I have certain insights about disease states and their most efficient treatment. Why shouldn’t I? After all I’ve been practicing medicine for 35 years!! The computer knows nothing about a reasonable diagnostic formulation or a plan of action which derives from judgement and other cognitive functions, the very skills that have been honed through years of practice. It can only check for things like the presence of the father’s side of the family history. It can’t tell if a patient is thoroughly examined or if the doctor used any of her cognitive skills in formulating a diagnosis or plan. Computers can only determine that certain sections have been filled out. For example, I have done this a number of times, when I can’t answer a question, just filling in the blank with a random string, ldskfjslfjf, to go onto the next section and complete my work. It is OK by the computer. Yes, for the computer, nonsense = thought just as long as nonsense is placed in the correct slot.

The computer seeks to ensure that there is a certain basic standard product or widget, in this case a history and physical. Hospitals need to provide this product in order to get paid. Hospital systems are now the chief recipients of medical payments.  We may call the efficient production of standard output a factory model of medical care. The goal or values have to do with maximizing the output of standard widgets called patient encounters.

Computer programs are useful for inexperienced individuals in medicine at approximately the medical student or at max the resident level. These students have to learn what is basically necessary and what’s expected of them and are inexperienced in the areas they will hopefully one day be experts at, assessing patient’s real needs and responding to them. For beginners,  computers help by checking on little errors and deficiencies. Even at the learning stage, the students game the system. They quickly find that cognitive skills diagnosis careful listening are no longer necessary. There is no incentive  at all for constructing a narrative for the individual patient, or finding a telltale physical sign on exam. For example, many residents will find an approving checkmark for rhonchi or wheezes in examining a patient but then fail to note that a whole person gradually becomes short of breath in time to intervene, until the poor patient collapses and needs emergency treatment and intubation. It’s all because no provision is made for abstract reasoning and evaluating a whole unit, only for the presence or absence of details.  Good clinicians on the other hand, will be hampered by interminable data entry, and attention to irrelevant smothering time-consuming detail, diverting energy away from their primary mission which is to help the patient. Productivity may well be improved at the level of the medical student, who now may pay almost no attention to a true encounter or exam, only fulfilling minuscule criteria. But  experienced specialists see productivity decrease when they are made to handle such mundane tasks as click and drag, typing and fulfillment of irrelevant criteria that distract them from their real goal, constructing that patient narritive.   Even worse, diagnostic formulation, evaluative, judgement, diagnostic plans the true meaning of the patient encounter get derailed,  as in general, computers are not value driven as humans are.

Surprisingly, we didn’t expect this in 2014 but more by 2025 or 2030s, computers have become our masters, not only on automobile assembly lines, but as regards cognitive functions and expert systems and in ways no one predicted. Computers generally serve to standardize output and assure basic criteria for widgets. When you order a Chevrolet, you want to be sure, you will get what you ordered. Hospitals have converted to the factory floors of output units, called patient encounters. Computers are very helpful in assuring that there is some dumbed down standard product across all hospital systems.  On the other hand, computers will give no quarter to inventive or idiosyncratic thought processes, expert formulations, or human cognitive intervention of any sort, at least in this stage of computer software development. Should computers raise a bar for the standard of care, in other words bring up the stragglers, the effect will be salutary, even at sacrifice of the  reasoning of the seasoned expert, so the current reasoning by hospital managers goes.

In closing I have to say all this reminds me a bit of descriptions we had of Soviet trained physicians a few decades ago. In the Soviet Union medicine was highly standardized and everything treated according to protocol. After all that is how things were done in Communist countries. By reputation medical care was always very poor in the Soviet Union, and worse, subject to the dictates of the state. For example, electro-convulsive and even insulin shock therapies were used routinely on political dissidents. When some of these physicians left the then Soviet Union,  they compiled a dismal record. Most of them did not know how to think, only follow protocols which were no longer available to them.  In fields with large numbers of knowledge workers,  such as medicine, medicine is by far not the only such field of course, there have to be basic standards which are followed, But you, at least for the foreseeable future, before computers advance beyond human capacities, need some people who really know what they are doing, who have learned the whys and wherefores. That is the seasoned clinician. It’s not quite time to get rid of him or her.