IBM researcher Martin Kohn talks about what goes into the training of super computer, Watson.
This content is brought to you by IBM. GOOD and IBM have teamed up to bring you the Figures of Progress series to explore the different ways that information has revolutionized our world. Click here to read more stories.
What type of tools could we give physicians so they could determine a diagnosis faster, and more accurately? The Watson research team at IBM is working to help bring their exclusive Watson technology to the exam room, enabling physicians and nurses to gain insights using the analytical power of the groundbreaking supercomputer.
Dr. Martin Kohn, chief medical scientist for IBM Research, is one of the lead team members working to bring the power of this supercomputer— known as Watson— to healthcare applications. One of IBM’s most unique and advanced machines, Watson is a cognitive system with natural language processing (the ability to understand and interpret human language) and evidence-based learning and hypothesis generation capabilities, Watson was introduced to most people through its 2011 appearance on the TV quiz show Jeopardy!, when it defeated the two reigning champions.
In his keynote speech at the CBMi 2012 Healthcare Informatics Symposium, Kohn details the place Watson could hold in healthcare. Noting that very often there is not a single answer or easy diagnosis in healthcare, Kohn said Watson could help fill the gap for doctors by presenting a list of several suggestions, alongside a score that indicates its confidence level in the solution, which the doctor could then use to make a final decision on treatment or diagnosis.
With one in five patients often receiving incorrect or incomplete diagnoses, Watson could become a welcome addition to hospitals and physicians offices. It’s estimated that 80 percent of patient data is unstructured data, which Kohn says is rich with information and potential insight if processed correctly. In healthcare unstructured data can be items like hand written notes and charts, or items not organized in a predefined manner that can’t easily be processed by traditional technologies.
With the ability to analyze data at incredible speed (Watson processed more than 200 million pages in three seconds on Jeopardy!), the technology can prioritize data in a patient’s chart, while also letting doctors know what important information might be missing to give a complete diagnosis. When a doctor is faced with making an immediate decision in the midst of hundreds of notes, charts and medical history pages about a patient, Watson’s ability to data crunch and sift through the medical terminology, could quite literally be a lifesaver.
Kohn notes, though, in an Atlantic interview, that Watson isn’t intended to replace doctors, but rather to serve as a treatment advisor.
So how will Watson put its abilities to use? For the past year Watson has been at medical centers, where doctors and nurses are training it for the benefit of all oncologists and the entire medical community. According to Kohn, both real-life and test medical cases have been submitted to Watson in order to give it further knowledge on conditions and treatment options. As of February 2013, Watson has taken in more than 600,000 pieces of medical evidence and two million pages of text from medical journals and clinical trials. And much like any medical resident, Watson is currently continuing to learn while on the job.
Read more from leaders like Kohn at Figures of Progress, including interviews with Matthew Stinchcomb , VP at Etsy, Jennifer Pahlka, founder of Code for America; Adam Brotman, chief digital officer of Starbucks; Rachel Sterne, CIO of the city of New York; and Oliver Hurst-Hiller, CTO of Donorschoose.org.