This particular bit of news leaves your friend Doc Gumshoe shaking his head in puzzlement. To wit, in the US it is expected that there will be a substantial shortage of physicians in the upcoming years. The American Association of Medical Colleges predicts that this shortage may reach about 122,000 physicians by the year 2032 – about 56,000 primary care physicians and about 66,000 specialists.
How can this be? Physicians are well paid; their work is interesting and satisfying; they are among the most respected members of the community. On the other hand, the education and training of a physician is long and arduous, and many physicians are drained and frustrated by demands that have little relation to the healing arts, such as keeping electronic health records and navigating the complexities of the means employed by patients to cover their medical expenses, whether private or public.
A big part of it is demographics. Overall, the US population is growing at a moderate rate. The Census Bureau estimates that the population will have expanded by about 10% by the end of the next decade. However, the big change will be in the age distribution. The number of people over the age of 65 is expected to increase by about 48% during that same decade. And the aging population will affect the physician supply. About one-third of all active physicians will pass the 65 year marker by the end of the decade. Some of them may decide to continue practicing, but some will certainly hang up their stethoscopes.
Furthermore, large segments of the US populace are underserved when it comes to health care. Race, insurance coverage, and geographic location all have major effects on a person’s likelihood of receiving appropriate health care. If medical care were equalized across all population segments, we would need an additional 95,000 physicians immediately, and even more as time went on.
Can artificial intelligence (or what currently passes for AI) make up for the shortage of live human medical providers? Can we entrust our health to digital devices?
If the track record of the Apple Watch is an example, the answer appears to be a big “NO!”
The Apple Heart Study
This was, supposedly, a genuine clinical study, designed to produce information of value concerning the Apple Watch, which – again, supposedly – can detect episodes of atrial fibrillation. Atrial fibrillation, commonly referred to as atrial fib or afib, is a condition during which the contractions of the four chambers of the heart are irregular. The sequence of contractions in the heart, when it is functioning normally are:
- Blood from the venous system returning to the heart enters the right atrium, which contracts, pushing the blood into the right ventricle.
- The right ventricle contracts, sending blood into the lungs, where it will absorb oxygen.
- Newly oxygenated blood from the lungs enters the left atrium, which contracts, pushing the blood into the left ventricle.
- The left ventricle contracts, sending the oxygenated blood into the aorta, to be distributed throughout the body.
But in afib, the two upper chambers, the atria, contract rapidly and chaotically, and do not coordinate with the contractions of the lower chambers, the ventricles.
As a result, blood is not completely expelled from the upper chambers – the atria – and the lower chambers also do not expel as much blood as they are supposed to.
The greatest risk of atrial fib is that a certain amount of blood may remain in the upper chambers, where it pools and is more likely to form clots than blood that is circulating normally. As a result, the greatest risk from atrial fibrillation is that one or more of those blood clots will find its way into the brain and cause a stroke.
Clots may also circulate and wind up in other vital organs, leading to blocked blood flow and the potential for life-threatening harm.
The Apple Watch is supposed to detect cardiac arrhythmias and send information about the wearer’s heart rhythms to Apple for evaluation. Apple paid Stanford University more than $8 million to conduct the study, which was intended to determine whether the Apple Watch did indeed do what it was supposed to do. (This may sound like serious money, but it’s a small fraction of what a well-controlled clinical trial would cost, and also just a tiny nibble of Apple’s profits.)
Apple Watch owners were invited to participate in the study, and nearly 420,000 people did so. Of these, a majority – 219,179, or 52% — were under age 40. Of those, only 341, or 0.16%, were notified that they had a cardiac irregularity. And of those 341, only 9, or 0.004%, actually had atrial fibrillation.
Only 24,626 persons in the study were 65 years of age or older. Of these, 775, or 3.14%, were notified of an irregularity. But in that group of 775 who had been told that they were experiencing some kind of cardiac arrhythmia, only 63 people, or 0.26%, actually had atrial fib.
Comparing the percentages of people whose atrial fibrillation was actually detected by the Apple Watch with the prevalence of atrial fib in the American population tells us something about the Apple Watch that Apple probably does not want to acknowledge. Atrial fib is present in approximately 2% in people under the age of 65, and in 9% of people age 65 or more. But the Apple Watch study detected real atrial fibrillation in only about one quarter of one percent of the participants who were 65 or older. What happened to the other eight and three quarters percent of those 65-year-olds? Did the Apple Watch just fail to spot their atrial fib?
So, in sum, what did the Apple Watch do in this very large study? Mostly, it frightened the 1,116 people that supposedly had a “cardiac irregularity,” while only 72 of them had anything of medical consequence. And at the same time, the Apple Watch missed identifying thousands of people with real atrial fib.