IBM, Nuance to Tune Watson Supercomputer for Use in Health Care

IBM will incorporate Nuance CLU speech-recognition applications into the Watson supercomputer to provide information that assists doctors as they make diagnoses.

IBM will continue its longtime collaboration with speech-recognition software developer Nuance Communications to bring the analytics capabilities of supercomputer Watson into the health care field. Under a research agreement announced Feb. 17, Nuance will feed its CLU (Clinical Language Understanding) applications into IBM's Watson hardware.

Nuance makes the Dragon speech-recognition software.

Meanwhile, IBM will incorporate its own Deep Question Answering (QA), Natural Language Processing and Machine Learning capabilities into the supercomputer.

Combining the CLU language capabilities of Nuance in a supercomputer such as Watson could lead to the next generation of EHRs (electronic health records) and decision-support applications, according to Dr. Eliot Siegel, director of the Maryland Imaging Research Technologies Laboratory (MIRTL) at the UMD School of Medicine. "We believe that this has the potential to usher in a new era of computer-assisted personalized medicine into health care to improve diagnostic accuracy, efficiency and patient safety," Siegel said in a statement.

A commercial product will be available in 18 to 24 months, IBM and Nuance report.

Columbia University Medical Center and UMD (University of Maryland) School of Medicine will contribute medical expertise that will enable Watson to work effectively in health care.

Watson, named after IBM founder Thomas J. Watson, is coming off a $1 million win on "Jeopardy." Big Blue has donated the prize to charity.

The supercomputer will be used to help doctors make diagnoses and analyze a vast amount of health care resources, including EHRs and medical journals, in ways that doctors and nurses may not be able to.

"What it can do much faster than a person is collect that information, analyze it and use it as an additional resource this huge array of health care literature or most-recent journals and provide feedback on that information to the physician," Dr. Marty Cohn, associate director for IBM Healthcare Analytics, told eWEEK.

"Just as Watson collects information and understands the questions on 'Jeopardy'-the subtlety of the puns-it looks at the language, understands what it really means and can bring information from the vast array of health care literature that is relevant to the physician's and patient's joint effort to come up with a proper diagnosis," Cohn said.

"I've been in health care in 45 years, and this is one of the most exciting things I've come across, with the greatest potential," he added.

With Watson's ability to understand natural language and respond in a humanlike manner, it will be able to understand patients' verbal descriptions of their symptoms, such as chest pains or dropping blood pressure, as well as collect medical data from EHRs, physician notes and family history to help doctors make recommendations on a patients' condition, Cohn explained.

Watson will also prove helpful in spotting potential drug interactions and highlighting missing test results, according to IBM. In addition, the supercomputer can guard against the bias of a particular doctor's past experiences.

Information overload from all the resources available contributes to 15 percent of inaccurate diagnoses, according to Harvard Business Review.

For Watson, however, the more data it's fed, the smarter it will get, Janet Dillione, Nuance's executive vice president and general manager of health care, told eWEEK. And the doctor or patient won't be aware of the short time Watson takes to use the Nuance software to come up with an answer. "The answer will be pushed to them," Dillione said.

Cohn noted that a timely answer is important to get physicians to adopt the supercomputing help. "If it takes them more time, they're not going to use it," he said.

Watson won't replace doctors, just provide additional relevant information for them to make diagnoses in a timely manner, Cohn stressed. In fact, Watson may not be used in emergency rooms, a setting where time is life or death, he added.

"Watson has the potential to help doctors reduce the time needed to evaluate and determine the correct diagnosis for a patient," Dr. Herbert Chase, professor of Clinical Medicine at Columbia University College of Physicians and Surgeons, said in a statement. Watson could help doctors personalize treatment according to a patient's needs, Chase added.

The supercomputer's hardware encompasses 10 racks of IBM Power 750 servers (a total of about 90 individual server units) with 2880 Power7 cores, which can run at 80 teraflops. The supercomputer also has 500G bps of on-chip bandwidth, a 10GB Ethernet network, 15 terabytes of memory and 20TBs of clustered disk storage.

In October, IBM and Nuance announced a collaboration to incorporate structured data into EHRs.