IBM has been working with healthcare providers to develop commercial applications for its industry for almost five years. Two particularly interesting applications are helping transform how patients are diagnosed and treated through the use of individualized, evidence-based medicine. Their development offers insight into the extreme processing and analysis capabilities of cognitive computing and the complexities involved in implementing and training cognitive applications.
Developed in concert with healthcare company WellPoint, Inc. (now known as Anthem, Inc.) and Memorial Sloan Kettering Cancer Center, these applications are designed to apply Watson’s advanced analytics, natural language processing (NLP), and machine learning (ML), along with Clinical Language Understanding (CLU) technology licensed from Nuance Communications to assist healthcare decision makers — including physicians and nurses — with the diagnosis and treatment of patients.
The applications harness Watson’s ability to analyze the meaning and context of human language, rapidly processing information to find precise answers by uncovering critical knowledge and facts buried under huge volumes of medical information, and offering answers human experts may not have considered to help validate their own ideas or hypotheses.
To develop these applications, IBM worked separately with WellPoint and Memorial Sloan Kettering for more than a year to train Watson in the areas of oncology and utilization management. Clinicians and technology experts spent thousands of hours teaching Watson how to process, analyze, and interpret the meaning of complex clinical information using NLP in an effort to help improve healthcare quality and efficiency.
Training involved feeding Watson more than 600,000 pieces of medical evidence and 2 million pages of text from 42 medical journals and clinical trials in the area of oncology research. To get an idea of just how powerful Watson’s NLP and analytical capabilities are, consider that Watson can sift through 1.5 million patient records (i.e., decades of cancer treatment history), such as medical records and patient outcomes, and generate evidence-based treatment options for physicians in seconds; a feat no human researcher could ever hope to achieve.
It took Memorial Sloan Kettering almost a year to immerse Watson in the complexities of cancer and genetic research to develop applications that offer the promise of many cancer patients receiving highly specialized treatments based on their personal genetic tumor type.
Initially utilizing 1,500 lung cancer cases, Memorial Sloan Kettering clinicians and analysts began training Watson to extract and interpret physician notes, lab results, and clinical research, while sharing its expertise and experience in treating hundreds of thousands of patients with cancer.
The first two adopters of Watson for cancer fighting were the Maine Center for Cancer Medicine and WESTMED Medical Group. Oncologists from both facilities tested the product and provided feedback to WellPoint, IBM, and Memorial Sloan Kettering to improve its usability.
During WellPoint’s utilization management pilot, Watson was fed more than 25,000 test case scenarios and 1,500 real-life cases. In this way, it acquired the ability to interpret 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. Additionally, nurses spent more than 14,700 hours meticulously training Watson. Watson’s education continues while it learns on the job, working with the same nurses who originally provided its training.
Watson started processing common medical procedure requests by providers for members in WellPoint-affiliated health plans in December 2012. This was expanded to include five additional provider offices in the Midwestern US, where Watson is helping to speed the review process between a patient’s physician and the patient’s health plan.
