The greatest problem of health care in United States – the world leader in health inequality – isn’t actually about the quality of care. The greatest problem we have is access to care. According to the CDC, nearly 20% of adults in the United States have no regular source of healthcare. One of the places this is most stark is in lifespan – where the wealthiest Americans have benefits from steady gains, about five years of additional longevity from 2000-2014 – versus the poorest, for whom, during the same period, life expectancy hasn’t changed at all.
There are many factors that contribute to this growing divide in mortality, socioeconomic and medical – but one of the biggest is simply not enough physicians in the right places. The best doctors and providers are drawn to similar circumstances: top hospitals, with the top tier of colleagues, in the most desirable places to live, with patients that can pay for services. But soaring prices of medical care is also a critical part of the growing problem. The costs of treating chronic conditions like diabetes continues to grow with the aging population. The rising cost of doctors (already high, and rising at 7-10% per year), pharmaceuticals, and expensive medical technology lies squarely on what’s referred to as Eroom’s law, the evil twin of Moore’s law, where the cost of healthcare exponentially increases over time.
This leaves us with more needs, but fewer, more expensive providers. The pressing question today is: Can new technologies slow or even reverse the exponentially rising costs to help truly democratize healthcare? The wealthiest patients today benefit not only from being able to afford the top medical services – but also maybe even to fly somewhere to get the opinion of more than one of the top doctors in the world. Imagine we could all do this – if to diagnose any condition, every patient called in, say, a conference call of the top 50 specialists in their field, who all drew upon their unique experiences and knowledge to confer and reach a consensus on an accurate diagnoses and treatment for that patient – who they had been following for years and years. Pretty good medical care, right? And probably the current best possible treatment for that patient. Unfortunately there’s no way in which this kind of approach or scenario is cost feasible – or scalable.
Magnifying and speeding up the human skill of evidence gathering and analysis is exactly what Artificial Intelligence and Machine Learning do best. They can bring 50 experts to bear for a single patient – by codifying the knowledge, taxonomy, and understanding of those experts. Machine learning is built on what the best doctors have learned, and now know: whether a suspicious looking mole is malignant or benign, whether an irregular heartbeat might be atrial fibrillation. Machine learning would be nothing without this essential human input; the technology trains on and scales the knowledge of the best doctors. And modern AI has the remarkable ability to keep learning, continuing to identify new features in the data which will give the most accurate diagnoses. This data is drawn not from a handful patients seen in an exam room but from thousands and thousands of examples – more than most specialists will ever see in a lifetime.
AI’s broadest and most important application may be its amplification of our own collective crowd wisdom.
Now imagine that your doctor had the ability to follow your individual history over time, thinking not just about the heart flutter that brought you in, or the suspicious mole, but knowing your entire history with perfect memory and recall. This is what’s called longitudinal data: understanding what your health has been like over time, and what’s anomalous for you versus what’s anomalous for the broader population. Like the best doctors, AI can constantly be retrained with new data sets to improve its accuracy, just the way you learn something new from each patient, each case. But the unique ability of AI to apply time-series methods to understand a patient’s deviation from baseline on a granular level may allow us to achieve a statistical understanding of causality for the first time – figuring out exactly what elements of your particular lifestyle and/or treatment have led to your current state. In other words, while a good doctor might guess that a man might have prostate cancer because his PSA levels have risen above a “normal” threshold, a great doctor might suspect prostate cancer not because his PSA levels were high compared to the population, but high compared to his own baseline. In fact, this is precisely how doctors discovered Ben Stiller’s cancer so early. AI understands how you have changed over time more than any human could – and this, it turns out, is much more predictive.
AI’s broadest and most important application may be its amplification of our own collective crowd wisdom. When you look at it that way, it begins to seem absurd that we rely on the opinion of any single doctor (or two, or three!), looking at data from only one person, drawn from only one moment of time. No matter how superb that doctor might be, individuals can, and inevitably do, make mistakes. But the wisdom of a crowd of doctors – hundreds, thousands of them – and the data of thousands and thousands of patients, with more coming every day – is very strong. The opinion of two doctors will never match terabytes of data. This is how human learning is scaled, in just the same way that the internet enabled the spread of knowledge to go faster than reading printed books. Imagine if doctors could telepathically teach each other their new findings. For modern AI methods, this is exactly what is happening.
Perhaps the most important way AI’s capabilities are super human may be the fact that AI can be replicated. Trivially. And at low cost. AI approaches are already often driven with relatively modest computational requirements, sometimes with a single GPU or a few CPUs. Thanks to Moore’s law’s continued push in this space, the cost of compute resources will soon be essentially free. So those 50 person conference calls for a single patient, tracking the patient’s health over a lifetime, are now no longer looking impossible. They are beginning to look cheap, and easy. And with the potential to reach corners of the world rife with doctor shortages, from near and far – places like prisons, or rural areas in the US, or developing countries – not with just a doctor, but the very best doctor humanly possible.
But democratization of healthcare will not happen on its own. The standard of care would need to change to incorporate this new technology. Using AI should be seen as amplifying and scaling the best human skills – and as such has a natural place in virtually all areas of care, including prevention, diagnosis, and treatment, from sending patients to the doctor at very early (previously undetectable) stages of disease to improving both outcomes and decreasing costs.
Scaling the doctor won’t replace doctors. It will magnify them, extend their reach, making it possible to recreate the advice of 10,000 doctors quickly and easily at lower costs – and bringing the best medical care to any corner of our country or the world.
Scaling the doctor won’t replace doctors. It will magnify them, extend their reach, making it possible to recreate the advice of 10,000 doctors quickly and easily at lower costs – and bringing the best medical care to any corner of our country or the world. It might even reinvent what we think of as patient-doctor interactions altogether. In the not so far off future, you might wake up, take a look in the mirror, use the toilet, and brush your teeth… where the mirror is AI enabled to look for dermatology, ophthalmology, and muscular-skelatory issues; the toilet will run a urinanalysis on analytes in your urine; and the toothbrush will gather DNA from saliva – with clinicians getting updates as needed to give you the very best care. Making your own bathroom the doctor’s office for a mini physical every single morning would give a longitudinal analysis from months to years to decades of information about you and your deviation from your personal baseline. Imagine the benefits of the very best doctors assessing each of us daily, no matter how remote, how rural the area, around the globe, every day, for our entire lives. This has the potential to give each of us the very best standard of care derived from not just your own, but billions of people’s longitudinal data sets.
When it comes to AI and healthcare, it’s actually the status quo we should be afraid of. Without these new technological tools, inequality will certainly continue getting worse. With AI, we have the potential to give everyone the best doctor, the best tests, the best analysis, anywhere in the world and at low cost – the potential to truly democratize healthcare.
This op-ed originally appeared in Forbes.