By Alan L Wagner MD, FACS, FICS
Somewhere around the time that Neanderthals and Homo Sapiens began to mingle and create communities, the first efforts of healthcare began to form. All the interactions were based on first hand observation. The “cures” were mystical, or if on occasion successful, based on little more than random chance. As rudimentary communal living developed, oral traditions and symbology helped pass on valuable survival and life-saving information. As recordation improved, society advanced, and it became possible for a body of knowledge to develop, move from one generation to the next, and help members of those groups in need. Great value was placed on the heads of individuals that could remember and recite ancient lessons. Seeing seemingly unrelated individual events as fitting into patterns, recognizing them together as similar to past experience, and predicting their combined outcome, raised that person to the highly-praised status of Oracle.
The scientific method and ancient Egyptian and Ayurvedic medicine cultures developed the same way. Very exact and clearly defined physician observations, along with experience, were required to make the decision of a diagnosis and apply one of the treatments from the historic wisdom tradition. They, too, were Oracles.
To transmit this body of knowledge, much like religions, great care was taken to protect and make holy this information. Only a select chosen few from each generation were ”worthy” of such significant societal responsibility.
Only recently has this begun to change in Medicine, and the speed of the change is about to accelerate. The challenge we all face is that the norms that have protected our community from suboptimal quality of care are being challenged and rewritten by the concomitant evolution in our society’s mores (a rise in pluralism) fueled by jaw-dropping advances in ever more readily-available technology. Most remarkable of these technologies is prediction and its rapidly decreasing cost.
You may be more familiar with the overarching concept of artificial intelligence, AI, and heard of ”deep learning”, and IBM’s Watson. Artificial intelligence is all based on prediction. The better you can predict something, the better you can forecast and more accurately decide an appropriate course to take given the information at hand. A growing pre-existing knowledge base, real-time evaluation and subsequent prediction can take the place of the fabled and highly valuable Oracle.
Before you throw yourself into a new career with this news, remember that it was up to the ancient leaders to determine what to do with the Oracle’s information. The Kings, Queens, and generals were highly valued for their judgement to make good decisions, and to evaluate whether or not, and how, the Oracle’s predictions were to be employed for the public good.
As physicians, and more recently non-physicians, we have had the combined role of Oracle and leader, determining the best course to take for our patients and communities.
But what if the role of Oracle changes? What if our frequently performed prediction actions are replaced by ”prediction machines” that are inexpensive and complete these tasks as well as, if not better, than we have performed as a group of humans traditionally?
I believe better prediction, AI, will open wondrous opportunities for improved healthcare delivery and access, reduce suffering, and eventually lead to lower costs to society. Better information upfront leads to better decisions. For example, as the data becomes more robust and the neural network deep learning matures, the frequency of biopsies should decrease as surety in the predictions increase. We will quickly abandon the multitude of unnecessary follow-up visits, and bring others in much sooner. Similarly, the lowering cost of AI allows us to understand the cost of our errors, and see more clearly where near and long-term improvements can be made. Application of AI strengthens the role of the physicians and team leaders as judgement becomes more valuable. AI will not be the answer for the rare conditions in the near term. A seasoned professional’s judgement will remain central for the foreseeable future.
AI combined with ever improving data gathering will bring about fundamental changes in the location and process of healthcare delivery. Who says you need to go to a doctor’s office, or hospital, to find out how your heart is doing? For less than $100 you can do that right now with your cell phone and an app. Your smart watch can do about the same thing, if not better. Need an accurate screen for colon cancer depending upon your family history? Head to the endoscopy suite? Instead, find out through the mail by sending a test sample from home. Blood pressure testing used to be the province of hospitals, clinics, and doctors’ offices–now you can take it anywhere, anytime. A cellphone image of a questionable bump on your skin can be interpreted within the day. Radiological and imaging studies are being read at all times of the day and night all over the world, regardless of where the studies were gained. In a few places, the initial report is being generated by an automated system and checked by expert humans. Automated autonomous readings are now available for fundus images for diabetic and hypertensive patients, relaying risks of heart disease and blindness.
Considering these examples, did we howl when the Coulter Counter appeared and made hand counting the peripheral blood smear all but vanish? The automation of sample evaluation improved the quality of the service delivered, and opened the door to a multitude of healthcare advancements. Buried in the mists of time resides the resistance that some pathologists had to allow non-physicians to peer through a microscope and report on a blood sample.
Given what is coming our way, we must consider the true meaning of tasks. A task is a collection of decisions. Our day-to-day tasks in medicine are a series of decisions requiring interpretation and valuation of multiple predictions. As prediction generation becomes much less expensive, and can be applied more broadly, more opportunities for decision making will present themselves, and make those making the decisions even more valuable. We will be able to create new questions to answer more complex problems. ith better prediction and automation, the tasks we have historically performed are newly made simple and will be relegated to others. Our workflow as doctors and care providers will have to change to respond to, and lead, this earthquake of an evolutionary shift in our relationship to a patient’s individual data and the world’s rapidly gathering intellectual resources. Our responsibility is to lead, not stifle, innovation.
The buggy whip manufacturers protested the advent of the automobile. Wisconsin banned the sale of margarine. Marie Curie’s “invention” was deemed dangerous by many. Major medical advances have always been met with protectionism. There are corners of our profession that resist online care delivery services, provider consultation, purchase of medications, refractions/eyeglasses/contact lenses, etc. or automation of any of our present tasks.
Imagine the excitement we would feel if it were 1816 and Rene´ Laennec had just given us the first stethoscope ever made to help our patients. With AI increasingly available to allow us to shape the future the way we believe is best for our patients, we should have that same feeling of wonder right now. To make the burnout causing constraints of the mundane and repetitive disappear, I urge you to take these new prediction tools robustly into your hands. Imagine, create, and bravely build a new paradigm that adds value to our noble profession!
Alan L. Wagner, MD, FACS founded the Wagner Macula & Retina Center in 1987. A Board certified ophthalmologist specializing in vitreoretinal surgery, Dr. Wagner received his medical degree from Vanderbilt University School of Medicine. He completed his residency in Ophthalmology at EVMS, and furthered his training as the Dyson Fellow in vitreoretinal disease and surgery at Weill Cornell University Medical Center. wagnerretina.com