How will IBM Watson affect the Future of Healthcare? Will it replace physician functions or just be another tool for them?
IBM made healthcare news when it directed its Watson supercomputer and artificial intelligence (AI) research to target a high-profile target: CANCER. But it seems that many in the medical industry lack the imagination needed to envision the potential that Watson offers. That’s why I’m writing today.
I was disappointed by a Business Insider article, Doctors say IBM Watson is nowhere close to being the revolution in cancer treatment it was pitched to them as. It appeared overly critical and caused me to respond this way:
“Garbage in, garbage out” is a term expressing the notion that incorrect or poor quality input will produce faulty output, especially in the literal world of computer systems. It also applies to AI systems, where the value depends on what was learned and how well they’ve been taught.
The first part of this article about a STAT investigation seems defensive, like physicians threatened by the idea of machines challenging their expertise wrote it. Don’t they realize that computers are tools meant to help them, and that they have a responsibility to feed the right stuff in? Toward the end of the article, it finally became clear that Watson actually is indeed providing the kind of help it was designed for.
Obviously, some people had unrealistic expectations of what Watson could do and how quickly it could do it. That may be because of overly zealous reporters trying to sell articles. It could have come from some IBM reps trying to make a sale, or just impress, and improperly positioning Watson’s capabilities, or neglecting to emphasize customer responsibilities. But it’s also possible that people just wanted to believe that AI would progress faster than it has. I don’t see that as justification for such a negative article.
Toward the end of the article, the authors admitted that, “It took Facebook and Amazon more than 13 years to grow [to] $20 billion.” That’s about the size of IBM’s Watson Health business unit now, after just a few years. So, to them I say, have more patience and give Watson time.
As a retired IBM technologist (retired in 1999, way before Watson) and futurist, I’ve collected quotes and predictions by people who should have been in a place to accurately predict the future but turned out to be very wrong. Here’s an article with my Top 10 Favorite False Predictions.
Watson‘s Machine Learning versus Human Learning
According to Peter Rudin in his Singularity 2030 article, “Learning is the act of acquiring new or reinforcing existing knowledge, behaviors, skills or values. Humans have the ability to learn, however with the progress in artificial intelligence, machine learning has become a new resource that can augment or even replace human learning.”
He goes on to make an important point that learning does not happen all at once but builds upon and is shaped by previous knowledge. To that end, learning may be viewed as a process, rather than a collection of factual and procedural knowledge. He
To appreciate the magic of Watson and it’s potential, let’s step back and look at how a skilled oncologist acquires her proficiency. It starts as an infant as she learns to recognize shapes and patterns and then learns to speak, read and do things through play, coaching, and practice. It takes YEARS for her to graduate from high school and college and complete her residency. How many texts and medical journals does she read in the process? A few hundred? How many patients does she and treat? Several dozen? Her knowledge and experience accumulates and enables her to weigh options and develop useful insight. That’s what IBM is doing with Watson – developing systems with the ability to learn over time and offer actionable insight.
In 2011, Watson competed on Jeopardy! and won, defeating former champions Brad Rutter and Ken Jennings. In just 3 seconds, the time required to answer a question, Watson was able to parse and analyze the equivalent of over 200 million books on disk. For each clue, Watson displayed its three most probable “guesses” and consistently outperformed its human opponents.
Do the doctors critical of Watson fear for their jobs? Some do. Or do more doctors see Watson as a tool for improving their effectiveness? I hope so.
Since the now-famous Jeopardy! game, Watson has gained access to general information on the Internet. But still, almost all of its cancer treatment training has come from Memorial Sloan Kettering Cancer Center. That’s rather limiting.
But just imagine how Watson might evolve over the next 10-20 years, given the exponentially accelerating pace of Moore’s Law and tech innovation. Give Watson access to other cancer experts and patient medical records worldwide; and then imagine including their geographic location, genetic makeup and phenotypical behavior, which can also influence treatment effectiveness. Finally, imagine adding the benefit of being able to remotely monitor treatment progress in real-time – not just for cancer treatment but for other conditions too.
Today, Electronic Medical Records are still contained within individual hospital systems. They’re not widely shared, but if they were, a future version of Watson should be able to compare conditions, treatments and outcomes of hundreds of millions, or even billions of patients worldwide, in different cultures. Its accumulated knowledge and value could then easily surpass that of the best oncologists, and that surely has many of them worried. I’d be too.
You see, Watson is just a few years old now, and it’s getting smarter every day. There’s nothing to prevent this AI system from expanding beyond cancer treatment into other medical disciplines – nothing except a lack of imagination and will, or an industry that doesn’t want to change.