IBM, Nuance Envision Watson Helping Doctors

IBM and speech recognition products vendor Nuance have announced a five-year research agreement to explore, develop and commercialize IBM's Watson computing system's advanced analytics capabilities for use in healthcare. Yes, this is the same Watson technology that's good at playing Jeopardy.

The two companies plan to combine the Watson technology with Nuance's speech recognition and clinical language understanding capabilities to create a "physician assistant" that will help guide doctors to more accurate diagnoses and better treatment decisions by providing more data sources. The Watson technology includes deep question answering (QA), natural language processing and machine learning capabilities.

The combined technology would be used in mining data from healthcare providers' existing electronic systems, including electronic health records, lab and medical imaging systems, as well as dictated reports. Data from outside sources, such as evidence-based research literature and best practices could also be factored in.

"The data could be from inside and outside the firewall," said Peter Durlach, a senior VP in Nuance's healthcare business unit. "Watson allows us to mine a large swath of information."

Nuance's speech recognition products are used by clinicians to turn spoken dictation, such as medical reports, into text-based unstructured and structured data. Watson would then mine that data.

Key information such as a list of a patient's allergies can be extracted, creating structured data that clinicians need to take into account about a patient, Durlach said.

In addition to data that's created within the physician's exam room, medical research is creating incredible volumes of new information each year that can take more than a decade to get put into practice by clinicians.

"The amount of data in the clinical arena is growing explosively, and for the human brain, it's hard to keep up," said Dr. Martin Kohn, IBM associate director of healthcare analytics research, and a physician who practiced emergency medicine for 30 years. "Giving physicians access to that information during the process of making diagnoses is powerful," he said.

Watson has the ability to understand the subtleties of language, and "use that and other resources to seek out the best answers" said Kohn.

The tools could be used by clinicians to make a real-time, personalized treatment decision about a specific patient, as well as by hospital administrators, medical directors and researchers looking to establish"the right treatment protocol for hundreds of patients, he said.

The use of Watson in healthcare will likely entail doctors providing the system with information about a patient, such as symptoms. Watson might mine and analyze other sources of data, including the patient's own records. It then might ask the clinician for additional information about the patient, before providing a list of best answers.

Watson's analysis could also take into account data about outcomes of other patients diagnosed with a particular disorder and how well they've responded to various treatments.

The answers the system provides could help clinicians validate their own hypotheses and provide possibilities they hadn't considered.

Nuance and IBM are also working with Columbia University Medical Center and the University Of Maryland School of Medicine. They will help identify critical issues in the practice of medicine where combined Watson-Nuance technology could help clinicians make better healthcare diagnosis and treatment decisions.

Nuance expects to bring the first Watson products to market within 24 months.

Need help sorting one electronic health record vendor's pitch from the next? Get the new issue of InformationWeek Healthcare. Download it now.