Guest blog by Rafael J Grossmann, MD, FACS (Twitter @ZGJR)
Healthcare is in trouble—and unless Artificial Intelligence (AI) and Deep Learning systems come to the rescue, the prospects of any reasonable fix, anytime soon, are very unlikely.
What facts support the affirmation of severe problems in the state of medical care systems today?
Well, I would list at least these three obvious ones:
- Cost. The approximate USA healthcare related expenses in the years 2000-2009 were $2.3 trillion, or the equivalent of 18% of the GDP. This is absurd, gross, shameful and, most importantly, unsustainable, even in the short term.
- Deficit of medical providers. According to WHO data, there is an approximate global shortage of 4.3 million. In the USA, the projected deficit for 2025 is greater than 120,000. Figures from the World Economic Forum (WEF) show, as an example, that in Nigeria, for the year 2030, the shortage will be in the magnitude of x12, and that it would take $51 billion and 300 years to correct!
- Medical errors. Not just mistakes, but mishaps that cause morbidity (more disease) or mortality (death) on those patients. The latest data by the Leapfrog group pointed out that there are about 440,000 deaths/ per year, caused by preventable medical errors in the US alone (Journal of Patient Safety: September 2013 - Volume 9 - Issue 3 - p 122–128
Some people doubt that number. I'm neutral about it. ("OK Glass: Disrupt Healthcare")
http://youtu.be/DVzkw7y4_u4). But let's say that’s not the number, but a half...or a third...or a quarter...or a tenth of 440,000. Forty four thousand (or even many less than that) is still an unacceptable figure, especially when the etiology is preventable.
Even One Death, regardless of Cause and if Preventable, is ONE TOO MANY, when AI-Deep Learning Can Help Make Us Better Physicians
In the last few years, and in fact, in the last several weeks, developments in AI and machine Deep Learning have enabled computers to start performing activities usually adjudicated to humans (recognize images, read, talk, learn from mistakes, etc.). The development is exponential, which means that MORE is happening FASTER!
Computers are now starting to becoming capable of performing about 80 percent of activities that humans now do and earn "a living" by doing. I believe that the medical field is no exception to this change.
I refer you to a fantastic talk by Jeremy Howard at TEDxBrussels 2014: "The wonderful and terrifying implications of computers that can learn.”
The binomium machine (computer with Deep Learning algorithms) plus the Human (us), interacting, will augment and improve the way that we “live” in a way not dreamed of before. I strongly believe this. And I think we will be much better physicians and improve on every variable and outcome because of it. Our future is making smart use of these technological resources.
I’m not talking about Vinod Khosla’s assertion from 2013, where he states that 80 percent of physicians will be substituted by technology in the near future (watch him at Stanford’s MedicineX 2013), but I do think that the many functions we have, along with many “traditional” steps that we take in caring for our patients, will be radically changed and improved…very soon.
He said, “Eighty percent of what doctors do, tech can do at a fraction of the cost—especially your rural doctor in India,” he continued. Still, he granted, tech won’t be able to replace the top doctors at research hospitals anytime soon. That is very important. We are not talking replacement of doctors, but certainly replacement of “old” and inefficient, expensive ways in which many providers deliver healthcare.
As a full-time surgeon, working in “the trenches,” and dealing with complex regulatory and administrative requirements, that sometimes even make our clinical work appear simple, I can be a witness to the fact that we would greatly benefit from some AI in our daily decision-making.
In regard to our medical functions, AI-Deep Learning systems could (will) free our memories and intellects to allow us to concentrate in the “art of medicine” instead, favoring the so needed human-side approach, making the experience much better for the suffering patient.
It is almost a paradox that this technology—these AI-Deep Learning systems—in the end, could indeed free us to be more human, better healers.
Follow my blog at Rgrosssz.com