Innovator Chat: How Watson Can Transform Healthcare

Dr. Martin Kohn, Chief Medical Scientist for IBM Research, tackles questions on how Watson is transforming the healthcare industry.
Dr. Martin Kohn, Chief Medical Scientist for IBM Research

According to one expert, only 20 percent of the knowledge physicians use to make diagnosis and treatment decisions today is evidence based. The result? One in five diagnoses are incorrect or incomplete and nearly 1.5 million medication errors are made in the U.S. every year. Given the growing complexity of medical decision making, how can health care providers address these problems?

Dr. Martin Kohn, Chief Medical Scientist for IBM Research, responds to our questions on how Watson is transforming the healthcare industry.

Topic #1: What is Watson?
Q: Watson has been described as the first of a new generation of computers. Can you talk about the shift to cognitive computing and what that evolution means for our relationship with computers?
A: We now are entering the Cognitive Systems Era, in which a new generation of computing systems is emerging with embedded data analytics, automated management and data-centric architectures in which the storage, memory, switching and processing are moving ever closer to the data. In today's big data era, computers essentially process a series of "if then, what" equations, which enables cognitive systems to learn, adapt, and ultimately hypothesize and suggest answers. Delivering these capabilities will require a fundamental shift in the way computing progress has been achieved for decades.

Q: Big data experts say that 80% of all data is unstructured. Including traditional datasets as well as tweets, blog posts, magazine articles, et al., how important is it to apply a broad definition of data to recognize Watson's potential?
A: It is estimated that 2.5 quintillion bytes of new data are created daily with an estimated 80% of this produced as "unstructured" data. Unlocking this data holds huge potential. Companies know this data is rich in information and potential insight, but are unable to handle the volume and pace with which it arrives. As a result, decision-making suffers. Using advances in natural language processing and analytics, the Watson technology can process information in a way that supports decision-making, representing a significant shift in the ability for organizations to quickly analyze, understand and respond to vast amounts of big data. The ability to use Watson to answer complex questions posed in natural language with speed, accuracy and confidence has enormous potential to improve decision making across a variety of industries, such as health care, retail, telecommunications and financial services.

Topic #2: Watson in Healthcare
Q: What is the opportunity for Watson in healthcare? Why not another industry?
A: The amount of medical information doubles every five years, while, in a survey, 81% of physicians say they can spare, on average, less than five hours a week keeping up. Combining its abilities to navigate the complexities of human language and to analyze massive amounts of data exceptionally quickly (more than 200 million pages in three seconds on Jeopardy!), Watson has the potential to take advantage of new research studies, published reports and articles, as well as patient outcomes and interactions, to help doctors make evidence-based decisions.

Q: We often hear about inefficiencies in medical treatment, from unnecessary procedures to misdiagnoses. What impact will Watson have for both doctors and patients in this regard?
A: Studies suggest that the complexities associated with healthcare have caused one in five healthcare patients to receive a wrong or incomplete diagnosis. These statistics, coupled with a data explosion of medical information that is doubling every five years, represents an unprecedented opportunity for the healthcare industry and next generation cognitive computing systems, to combine forces in new ways to improve how medicine is taught, practiced and paid for.

The information medical professionals need to support improved decision making is available. Medical journals publish new treatments and discoveries every day. Patient histories give clues. Vast amounts of electronic medical record data provide deep wells of knowledge. Some would argue that in this information is the insight needed to avoid every improper diagnosis or erroneous treatment. Watson uses natural language capabilities, hypothesis generation, and evidence-based learning to support medical professionals as they make decisions..

Q: It's clear that Watson isn't intended to replace doctors, but that it has the potential to dramatically transform patient care and outcomes. Could you describe how Watson will be used in the clinical setting?
A: Watson will serve as a treatment advisor that will help doctors and other medical professionals to more effectively personalize treatment. For example, a physician can use Watson to assist in treating patients. First the physician might pose a query to the system, describing symptoms and other related factors. Watson begins by parsing the input to identify the key pieces of information. The system has been taught to understand medical terminology, extending Watson's natural language processing capabilities.

Watson then mines the patient data to find relevant facts about family history, current medications and other existing conditions. It combines this information with current data such as test results, and then examines all available data sources to form hypotheses and test them. Watson can incorporate treatment guidelines, electronic medical record data, doctor's and nurse's notes, research, clinical studies, journal articles, and patient information into the data available for analysis.

Watson will then provide a list of potential treatment options along with a score that indicates the level of confidence for the relevance of each hypothesis. The ability to take context into account during the hypothesis generation and scoring phases of the processing pipeline allows Watson to address these complex problems, helping the doctor -- and patient -- make more informed and accurate decisions.

Q: These solutions have been years in the making and have just recently moved from pilot to production. How are they be using today?
A: Memorial Sloan-Kettering and Wellpoint continue to push test cases through Watson, a training process for Watson that will continue for the foreseeable future. In the current cycle there are a handful of oncologists that are working with Watson to help it understand lung cancer. As the first early adopters, WestMed and Maine Center for Cancer Medicine will be receiving access to the treatment advisor that they will run in parallel to their normal mode of practice. The purpose of this beta phase is to provide us feedback on the experience (such as what did they like? what was confusing? what observations did they have regarding Watson's recommendations?). The doctors will test using real patient historical data in the clinical/hospital setting, but not live with real patients at this time. Once the beta program is completed, we will work on full deployment in the hospital setting to be used with live patients.

Having been announced in February, the technology has not yet made its way into the clinical setting where it will be used with live cancer patients. This fine-tuning stage is a significant step in the implementation and, as as you can imagine, something of this magnitude doesn't happen overnight.

Q: On a more macro level, what impact could Watson have on soaring healthcare costs in the U.S.? Why?
A: Consider that some 75% of all healthcare dollars are spent on patients with one or more chronic conditions. Imagine applying Watson to harness the latest cancer or heart disease research to tackle disease challenges plaguing our citizens and alleviating the economic impact they have on our nation. Watson is part of the discussion about what's next in health care. The U.S. government, and the public and private sector together have made large investments in health care. This is about how do we truly start delivering value and improving how medicine is practiced as a result of this investment.

Additionally, according to the Institute of Medicine, 30 percent of the $2.3 trillion dollars spent on health care in the United States annually is wasted. While there are many factors contributing to this statistic, one step toward reducing waste is helping to personalize healthcare decisions so that diagnostic and therapeutic choices are more likely to be beneficial to the individual. We know that each patient is unique and not all patients respond in the same way to medications, for example. 

Healthcare payers also want to support personalized health care so that they pay for interventions that have value. That requires better use of evidence to improve the utilization management (UM) process, which governs the pre-approval of healthcare insurance coverage for many medical procedures. Improving response time, accuracy and consistency in the UM review process has been a goal of the entire industry. Attaining this goal is challenging, in part because of the volume of data that is analyzed in making UM decisions.

Q: Much has been made of Watson's ability to provide evidence behind any recommendation it suggests. What difference will this make for doctors, patients and insurance companies?
A: Some estimates suggest that as little as 20 percent of the clinical decisions are evidence-based. Watson is able to provide supporting evidence with every suggestion, both to provide transparency and to aid in the doctor's decision-making process. Watson will also point out areas in which more information is needed and will update its suggestions as new data is added. Ultimately, Watson is expected to facilitate access to the best of oncology's collective wisdom.

Q: How did the partnerships with WellPoint and Memorial Sloan-Kettering Cancer Center come about? What do each of them bring to the table for Watson's development?
A: To help improve the quality of care delivered, WellPoint teamed up with IBM on a new approach to UM: using the cognitive system IBM Watson to provide approval suggestions to nursing staff that is based on clinical and patient data. For more than a year during a pilot phase, WellPoint trained Watson with 18,000 historical cases. Now Watson uses hypothesis generation and evidence-based learning to generate confidence-scored recommendations that help nurses make decisions about utilization management.

And in March 2012, IBM launched a partnership with Memorial Sloan-Kettering Cancer Center and Wellpoint, where work is under way to teach Watson about oncology diagnosis and treatment options. Research centers such as MSKCC publish innovative findings in peer-reviewed journals, which are the most common medium doctors use to gather new medical information. Nevertheless, keeping up with the medical literature can take as many as 160 hours a week. It's not surprising that only about 20 percent of the criteria that clinicians use today is evidence-based.

MSKCC began looking for a way to expand the accessibility and usability of medical evidence to improve patient outcomes across the field of oncology. It wanted to find a technology solution that could provide personalized diagnosis and treatment suggestions for individual patients. Beginning with breast and lung cancers, the organizations are consolidating clinical expertise, molecular and genomic data, and a vast repository of cancer case histories into an evidence-based solution. The cancer center's world-renowned oncologists are training Watson to compare a patient's medical information against a vast array of treatment guidelines, published research and other insights to provide individualized, confidence-scored recommendations to physicians.

As a result of these partnerships, IBM, Memorial Sloan-Kettering and WellPoint have introduced the first commercially based products based on Watson. These innovations represent a breakthrough in how medical professionals can apply advances in analytics and natural language processing to "big data," combined with the clinical knowledge base, including genomic data, in order to create evidence based decision support systems.

These Watson-based systems are designed to assist doctors, researchers, medical centers, and insurance carriers, and ultimately enhance the quality and speed of care. The new products, announced in February 2013, include the Interactive Care Insights for Oncology, powered by Watson, in collaboration with IBM, Memorial Sloan-Kettering and WellPoint and The WellPoint Interactive Care Guide and Interactive Care Reviewer, powered by Watson, designed for utilization management in collaboration with WellPoint and IBM.

Q: How specifically have they trained Watson?
A: For more than a year, IBM partnered separately with WellPoint and Memorial Sloan-Kettering to train Watson in the areas of oncology and utilization management. During this time, clinicians and technology experts spent thousands of hours "teaching" Watson how to process, analyze and interpret the meaning of complex clinical information using natural language processing, all with the goal of helping to improve healthcare quality and efficiency. To date, Watson has ingested more than 600,000 pieces of medical evidence, two million pages of text from 42 medical journals and clinical trials in the area of oncology research.

Watson has the power to sift through 1.5 million patient records representing decades of cancer treatment history, such as medical records and patient outcomes, and provide to physicians evidence-supported treatment options all in a matter of seconds. Starting with 1,500 lung cancer cases, MSK clinicians and analysts are training Watson to extract and interpret physician notes, lab results and clinical research, while sharing its profound expertise and experiences in treating hundreds of thousands of patients with cancer.

Throughout WellPoint's utilization management pilot, Watson absorbed more than 25,000 test case scenarios and 1,500 real-life cases, and gained the ability to interpret the meaning and analyze queries in the context of complex medical data and human and natural language, including doctors notes, patient records, medical annotations and clinical feedback. In addition, more than 14,700 hours of hands-on training was spent by nurses who meticulously trained Watson.

Watson continues to learn while on the job, much like a medical resident, while working with the WellPoint nurses who originally conducted its training. Watson started processing common, medical procedure requests by providers for members in WellPoint affiliated health plans in December, and was expanded to include five provider offices in the Midwest. Watson will serve as a powerful tool to accelerate the review process between a patient's physician and their health plan.

About the author of this Post

Dr. Martin Kohn, Chief Medical Scientist for IBM Research
Martin Kohn, MD, a graduate of Harvard Medical School, is currently Chief Medical Scientist for IBM Research. During a 30-year career in clinical practice, Dr. Kohn has specialized in emergency medicine. He is certified by the American Board of Emergency Medicine, and has practiced in New York, Ohio and California. He is the author of numerous articles published on clinical subjects. He has devoted his career to improving health care and clinical processes.
Sponsor Content

Join the Discussion

The Atlantic does not moderate these reader comments, except to the same extent comments are moderated pursuant to the Terms & Conditions generally applicable to all content on The Atlantic's sites. blog comments powered by Disqus