Watson's next test: health care

IBM's Watson computer marched into pop-culture history last year by outscoring two former "Jeopardy!" champions on national television.

Now, the Minnesota-made computer system is moving into real-world business applications, and the state's health care industry is looking for ways to tap in.

Named after Thomas J. Watson Sr., the influential early leader of IBM, Watson is an artificial intelligence computer system whose brainpower comes from multiple computers that are engineered and manufactured at the IBM facility in Rochester.

With a souped-up processing system that can ingest reams of data, analyze it and spit out results in the blink of an eye, IBM officials believe that turning Watson's brainpower onto the medical system can reduce errors and improve the quality of care.

Watson is much more than a search engine with access to encyclopedic amounts information. It can understand the spoken word, and it can learn. With 16 terabytes of memory, twice the Library of Congress, Watson can store huge amounts of data ranging from patient health records to cutting-edge treatments. While a doctor may spend 10 hours a week reading the latest advances in medical journals; Watson can read 200 million pages of text in three seconds.

"This changes the world," said Dr. Gary Oftedahl, of the Institute for Clinical Systems Improvement in Bloomington. "Watson can shift the ability to access information in a way that's never been done before."

If a patient says, "I'm dizzy," Watson could assess whether that's a symptom worth exploring or ignoring, given the patient's medical history. It could compare symptoms with others in a vast database, draw in the latest scientific research from around the globe and weigh the value of a range of treatments that might include doing nothing, prescribing a certain medication or ordering more lab tests.

It took four years to teach Watson to play Jeopardy, said Kohn. When it comes to health care, "Right now, we're crawling."

A number of pilots are in the pipeline, however, and IBM has high hopes. Watson could generate $2.65 billion in revenue for IBM in 2015, or about 52 cents per share in earnings, according to analyst Ed Maguire of CLSA.

IBM began working with managed care company WellPoint last year on a program to automatically pre-authorize certain medical procedures. WellPoint also is working on a Watson project with Cedars-Sinai Medical Center in Los Angeles to build decision-making tools for oncologists treating breast, colon and lung cancer.

In March, Memorial Sloan-Kettering Cancer Center in New York became the first provider to buy into Watson's power to help its doctors diagnose and treat cancer, and Citibank became Watson's first financial services client.

While IBM hasn't announced Watson collaborations with any Minnesota health care organizations, it has had "conversations with most of the major players in town," said IBM's Barry Mason, vice president of global health care payers and senior location executive for the Twin Cities. Among those are the Mayo Clinic in Rochester and Minnetonka-based UnitedHealth Group, which operates the nation's largest insurance plan, and its health services division, Optum.

"We're very much focused on what we're working on now, but you've got to think bigger," said Susan Heichert, Allina Health's chief information officer. "Watson is taking it to the next level."

For all its promise of more accurate diagnosis and treatment, both fans and critics urge caution when bringing technology into the exam room.

Oftedahl, also an internist, said there are legitimate fears about patient privacy. Clinicians raise questions that Watson could be used against them in malpractice cases, or by insurance companies to direct or limit the kind of treatments that will be reimbursed. Health industry workers say human contact remains an essential part of medicine.

"Some of those are real, some are based on fear of the unknown and fear and uncertainty for the future," Oftedahl said. "It's hard to ignore the opportunities, yet we need to be cognizant of unintended consequences. There are always unintended consequences."

Watson's data mining potential holds the most promise, particularly as federal reform laws force hospital systems to be more accountable for keeping patients healthy, Heichert said. Under the law, hospitals will be penalized when patients bounce in and out of the hospitals, which is prompting some providers and insurers to set aside some of their adversarial differences and share data as never before.