Computers can be such know-it-alls

One of the reasons I enjoy my Shabbat table is the eclectic group of guests that regularly join us for the meals. As we all tremendously enjoy my brilliant wife’s amazing cooking, the conversation moves from topic to topic. Some topics are very serious, and some are purely for the humor. But by the end of the meal, usually a couple of hours after we started, I often find that I have learned something new. Everyone should strive to learn something new every day. It gives a taste to life which is very special and enjoyable.

One of the critical things I have learned from these Shabbat discussions is that it is tremendously easy to make statements, even grand ones, without having to support these statements with references or data. It might sound funny to speak about research level support for a random statement amongst a group of friends. But when someone says that “the economy is doing better” or “the economy is doing worse”, it definitely would add quality of the conversation to know where these conclusions come from.

Obviously, everyone has the right to their opinion and can interpret the news and other data in any way they wish. I can definitely say, anecdotally, that the range of interpretations of “facts” can be extremely broad. And if you happen to have someone within your group that is expert in a topic, you often quickly find yourself being challenged with hard data that is hard to ignore. In practice, there are not a lot of people who are that expert in that many fields, so that the Shabbat table turns into an academic conference every week. It is much more common that everyone voices just an opinion [without numbered references] and by the time dessert comes, the world is a better place.

The environment in a physician’s office should be highly professional and yes, academic. As much as possible, a physician should always be able to support a claim or opinion or suggestive treatment, on the existing medical literature. This is what is called evidence-based medicine, where the physician uses accepted research conclusions in order to guide the care of his or her patients.

The patients themselves should always be comfortable and asking why the physician has chosen a particular path. a classic example is when a young patient presents with an ear infection. One would think that such a “simple” and common medical condition would have been studied so extensively that we really did know the best treatment. But as it turns out, the treatment of ear infections has changed dramatically over the last decade based on the continuing flow of new research.

I will state now that there are many physicians who, unfortunately, treat middle ear infections [otitis media] in the same way that they were taught, 10 or 40 years ago. Often, for these physicians, the presence of a fever and the most minimal redness over the eardrum is enough to write out a prescription for Augmentin [because it strong enough to be sure to kill everything]. Even 10 years ago, this might have been more universally accepted practice. But based on a great deal of more recent research, this approach is not only exaggerated but potentially harmful.

The standard today is to only treat very specific presentations where the findings of fever and middle ear signs are dramatic. In most cases, especially in a fully vaccinated child, pain control  with ibuprofen and time will do all the necessary work. In most cases, the child can be spared unnecessary antibiotics which is far better for the child and far better for the community. Knowing this does not demonstrate brilliance. All it demonstrates is a regular tendency to stay up-to-date with the medical literature. And if a doctor does not do this, it literally puts into question many if not most of his or her practices.

The thinking/cognitive computers, like IBM Watson, have one critical advantage when ever they express their “opinion”. Following Watson’s interpretation of the medical literature in regards to a question it is answering, there is a list of references. These references are scored in terms of their contribution to the final opinion presented by Watson. This list of references and scores offers the human physician the opportunity to review the evidence and to agree with or challenge the computer’s conclusion.

This use of Watson is nearly the utopic vision for how computers can contribute to our well-being. The human is in charge and makes all the final decisions. But all of the information necessary to responsibly make those decisions is presented cleanly and neatly and statistically to the decision-maker. This is the classic example of advanced computer systems assisting physicians to provide better care.

Even in a scenario where non-physician humans directly interact with the computers, it will still be the final decision of the human patient as to which path to take towards wellness. And of course, the patient could present Watson’s references and conclusions to a physician in order to get a further opinion as to how to proceed.

The following would be a fascinating scenario, and raises critical questions about the provision of medical care in the future. Let’s assume that Watson presents data from the medical literature that has an extremely high score in terms of medical value. In other words, Watson states that the clearly best treatment for the condition of the patient is to do “X”. And Watson adds that the likelihood that “X” is the most appropriate treatment is well over 99%.

There are very few decisions that we make in life where we know beforehand that there is a 99% chance that we are right. But what happens if the patient, even after consulting a physician, decides to take a different path and be treated  with a far less likely to be successful treatment. Obviously, the patient has a right to choose what’s best for him or her. But what if the insurance company says that it will not pay for the treatment, since the patient has ignored what is clearly best [and perhaps even cheaper]. When the evidence is so glaring as to the superiority of one treatment over another, can it be said that the patient is irresponsible to choose anything else? If so, this is effectively enforcing the “opinion” of the computer system.

What of the physicians? What if a physician says that he or she refuses to treat a patient who will not follow highly recommended therapy, based on the medical literature and presented by Watson? Should a physician be forced to provide care which is not ideal? Can a physician argue that the patient is clearly not able to make appropriate decisions, if he or she ignores glaringly ideal therapy?

The answers to these questions will be the foundation of medical and health policy in the coming future. There are no simple answers to these questions. And it could very well be that the answers will ultimately be based on simple financial considerations. While patients would be free to privately seek any treatment they wish, no insurance would support alternative therapies that do not have the evidence to back them up.

What this all comes down to is the following question: in the world with sufficient evidence to make the best decision, does one still have the right to choose differently? And if that person’s right to choose is maintained, is it still society’s responsibility to care for that person?

I wish I knew the best evidence supported answers.

You may enjoy reading the following (or it may scare you): The value and safety of listening to medical talk shows

Thanks for listening

My website is at http://mtc.expert