Over the last few blogs of our series, “How will Artificial Intelligence (AI) Impact Healthcare”, we have been discussing the known vulnerabilities and risks of Artificial Intelligence (AI). This week, we will go over a few suggestions on how to mitigate these risks, yet the value of AI remains overwhelming and compelling. In Part 7, we wrote about the unknown origins of AI and we showed a timeline going back to the 1950s when the elements that are now AI were being discussed.
Some have gone back and documented that AI is really part of a technological evolution that is now on its sixth wave, making AI the natural seventh wave. The first was the development of the Mainframe Computer, mostly developed by IBM and referred to as “IBM and the Seven Dwarfs” with Burroughs, UNIVAC, NCR, Control Data, Honeywell, and General Electric all going back to the same 1950s. After the Mainframe, we entered what we would call a Timeshare of those Mainframe or Multiple Virtual Storage in the 1970s, that allowed companies to share, or couple, the computing resources.
This was followed by what I remember well as the dawning of the Personal Computer Age, with new predominant players including Texas Instrument, Tandy, Commodore, and Sinclair, until IBM (reluctantly) entered this market in the 1980s. We can talk or write a lot about how IBM heavily missed the following trends. Perhaps a lack of imagination or reluctance to dilute sales in their mainframe business; however, this demonstrates that to be entrepreneurial, one must understand the new technologies and be quick to pivot when needed.
The fourth trend was the actual Internet, which reached common consciousness of the Internet in the 1990s. The fifth wave was the Mobile Internet that we all have lived though in the early 2000s. Lastly was the sixth wave, and we have all experienced Cloud Computing in the early 2010s.
Following in that line of thinking, AI is the natural and connected next step, and is clearly carrying forward the DNA of its predecessors back to the 1950s.
From our view, healthcare has not been transformed by these trends to the same degree as biotech. In fact, one may argue that if it were not for the wave developing as it has, biotech may not even exist – at least not as it is today. We have seen amazing technological advancements applied to healthcare from the computer chip that enables everything from ultrasound machines, MRI, CT, PET scanners, as well as cellphones. Some can say that even “Dr. Google” has advanced the state of healthcare; however, the reality is that the chip has enabled three (3) technologies that have had a much larger impact on healthcare – the phone, the pager, and the fax machine.
AI has an opportunity, and we predict it will have the most impactful influence on healthcare than those other six waves combined.
First, AI has a scalar effect on all the other technologies. Just as each one mentioned above, multiplied the reach of those behind so AI will continue to impact how drugs and drug combinations are developed. AI will continue to help doctors in diagnostics, and we think most importantly in predicting health and wellness for patients.
Healthcare is the largest industry in the U.S. by GDP and by number of employees, yet it lags behind many industries in the use of technology per-employee. Healthcare has amassed a staggering amount of data, but relatively little intelligence, partly because so much of it is unstructured. The average patient in the U.S. generates roughly 80 megabytes of data every year; in part, this is due to the massive adoption of Electronic Health Records (EHR), and that data has been growing at a rate of over 30% per year. In 2019, it was estimated that in the entire world, we were using 2,314 exabytes (one exabyte is equal to 1 million gigabytes) of storage to just to hold medical information.
The problem we have created in the EHR industry is that we have continued to create wider and deeper ‘data lakes’; however, few are connected to one another and there is no ‘policing’ or vetting of the information. Information is literally just dumped in the ‘lake’, some deeply unstructured, unvetted, and much of it de-identified. In the U.S., we have the wishful thinking of creating “interoperability” but that has not worked. What incentive do the tech providers have in bringing it to fruition after making tons of money seeking that impossible dream in our view? What incentive is there from elsewhere to complete the ‘dream’ when that huge amount of data has not impacted quality of life, cost of care, or time to diagnose?
Some are proposing to use AI to help doctors with diagnosis. My experience in 95% or more cases, doctors don’t need help in diagnostics. The clear value in AI is for example, assisting with imaging – though that will save lives by nearly flawless early diagnostics, that is not where the impact can be mostly felt.
We think the big opportunity is in using AI to see into all those data lakes, restructure the data so that future health can be better predicted, and suggest changes that a consumer can make to live a better, healthier, and longer life.
On the other side, once properly vetted, structured, and de-identified, that data can be used by universities, biotech developers, pharma, government, and even employers to not only confirm what we know about life, drugs, and cost, but also detect more important factors of what we don’t know. We will be able to see longitudinal patient and consumer data in a way that is not possible in 95% of healthcare today – even though there are many different providers touching the same patient. That huge healthcare industry we mentioned earlier becomes progressively less efficient as the system becomes more complex.
The takeaway we see is the looming transformation in healthcare, from institutional control of isolated healthcare data, to one where the consumer controls and directs their own integrated, “Universal”, Personal Wellness electronic Record®.
More to come later.
-Noel J. Guillama, Chairman