What is cognitive computing, and what does it look like in healthcare?

A person can walk into a bank pretty much anywhere in the world and use his or her ATM card to draw out cash, thanks largely to an infrastructure built by IBM in the 1970s. Now, imagine the Watson cognitive computer being just as ubiquitous.

That's IBM's vision for its former Jeopardy! champion, and it's quickly becoming reality. Already, there's cognitive cooking for your kitchen and cognitive coaching for your daily run. What's Watson's vision for medicine?

“There's probably a thousand different use cases, and that's what makes my job hard because there's no single thing we're trying to do,” acknowledged Bill Evans, chief marketing officer for Watson Health.

Evans played host to the MM&M editorial team during its visit to the Watson Experience Center in New York in February. Nearing the one-year anniversary of the launch of the Watson Health business unit, we came to its Astor Place offices seeking to learn more about the potential of the Watson supercomputer in healthcare.

We'd been struck not only by the unit's $4 billion in acquisitions in less than 12 months, but also by the number of biopharma, medtech, and pharmacy giants Watson's been partnering with—Apple, EHR provider Epic, Johnson & Johnson, Medtronic, Teva, Novo Nordisk, and CVS, to name a few.

Some of the use cases Evans alluded to can be seen in these partnerships. With J&J, it's leveraging cognitive insights to develop apps for patients preparing for and recovering from knee surgery, and with Medtronic, it's live-streaming data coming from insulin pumps to help diabetes patients predict a life-threatening event.

This month Watson announced another pharma collaboration: with Pfizer to create a remote-monitoring system to support patients with Parkinson's disease. “If there's a pharmaceutical company we're not talking to at this point, I'd be surprised,” said Evans.

I asked him why IBM thinks it can succeed in harnessing big data for things like the Triple Aim where other data and analytics providers have not.

“We're not a Johnny-Come-Lately company,” he replied. “This is a company that's a hundred and some years old; it's 350,000 people worldwide. It's got a legacy of being invaluable to the businesses that it serves.”

More from Watson Week: Challenges exist, even as excitement grows for big data market

But IBM has been a B-to-B healthcare company all these years. For instance, its life sciences division wrote the first pediatric medical record in the late 1960's and has worked with health systems for decades.

With the launch of Watson Health, the company is trying to be more front-and-center in the eyes of the consumer, and to use cognitive insights to improve lives and reduce waste.

That's a tall order, but Watson is no longer a new technology. “While we have to be diligent and focused on our future, we've got a head start on everybody else and our breadth of capability in data that right now is unmatched,” said Evans.

To test this theory, and to learn more about IBM's big bet on healthcare, starting today we've embarked on a weeklong series of articles. Let's start by answering some basic questions.


“A cognitive system understands, reasons, and learns,” explained Evans (pictured). “So you teach it much like you would teach a child. You don't program it.”

Indeed, Watson has ingested all 23 million documents in the Medline repository. As part of medical residencies at famous teaching hospitals, it ingested tens of millions of journal articles and clinical texts.

As it chews through all that data, its handlers start asking it questions. The computer makes connections and correlations of the data, ranking them by confidence score. The cognitive system constantly adapts its reasoning and evidence chain based on new inputs.

More from Watson Week: How Watson for Oncology is advancing cancer care

“As new publications come in, it can re-correlate all the data to have different confidence and different inputs based on the information it receives,” added Evans.

In this way, it's constantly updating itself — what humans call learning. “It's constantly getting smarter and smarter as you use it,” Evans said. “The more data we can feed it, the smarter it gets.”

Watson was introduced in 2011 on the TV show Jeopardy!, where it smoked two human contestants. Since then it's slimmed down physically. Its massive servers no longer take up a whole room; Watson now runs on a single server in a cloud computing environment.

And Watson execs are no longer intent on winning quiz shows but helping customers in a range of industries succeed in their jobs. Its first use case in healthcare was with the HMO Wellpoint on claims adjudication.

Several years ago, cancer hospitals Memorial Sloan Kettering Cancer Center and the University of Texas MD Anderson Cancer Center began training it for use in advising oncologists on treatments. IBM is now focused on embedding access to Watson in a host of products and services.

See also: Oncologists hope to bring the computing power of IBM's Watson to bear on the complexity of treating cancer

Is there no limit to the technologies to which its cognitive-computing engine can be applied?

“I hate to sound flippant,” said Evans. “Anywhere there's data and a problem, Watson can be of service.”

His statement is borne out by what IBM refers to as its ecosystem partnerships — businesses built on the back of APIs. By plugging their APIs into Watson's cloud-based system, third-party companies can create cognitive apps that bring Watson directly to their users.

Welltok, a population health management company, is one example, and another is Pathway Genomics, which debuted a genomic wellness app powered by Watson this year.

The network of coders not only accelerates the adoption of cognitive technologies, but stands as a key proof point for the fact that “this is not science fiction,” said Evans.

These complement its big beachhead partnerships with the likes of Pfizer and J&J. The overarching goal is to use cognitive insights to help people relate to their own health, thereby improving lives, reducing costs, improving access to care, or creating value economically.

More from Watson Week: Second-generation IBM star pushes for partnerships

“We know at IBM it's going to take a collaboration with other players in the industry to really tackle this problem,” Evans said. “We're not going to be able to do it on our own, so when we look across the health landscape we need partners alongside of us.”

Watson Health has also plunked down about $4 billion on acquisitions, including Phytel, Explorys, Merge Healthcare, Cúram Software, and Truven Health Analytics, to build what it claims is the largest healthcare data trove in the world.

On the acquisition of Truven, which amasses claims data, Evans said, “If you think about how the industry is moving toward an outcomes-based model, you need real-world evidence attached to the payment value, attached to the treatment paradigm. Being able to model all of that data gives us very good insights into what's working and what's not.”

The aforementioned use cases are just the tip of the iceberg in terms of what's possible. To borrow Evans's slightly over-hyped description: Watson's cognitive computing is basically limited only by the problem you want to give it and the data that it has access to.

Frost & Sullivan predicts that by 2017, 10% of all computers will learn like Watson does.

Will the “Powered by cognitive computing” sign become as ubiquitous as the ATM? Maybe, but cognitive capabilities are certainly on their way to becoming a bigger part of everyday healthcare.

Check back throughout this week to read MM&M's Watson Week series, where we will examine Watson's work in oncology, provide a behind-the-scenes look at the company's partnerships with pharmaceutical companies, and discuss the challenges for technology giants like IBM and Apple as they enter the U.S. healthcare market.