Cognitive Computing with IBM’s Watson Tackles Oncology Data

Leading cancer institutes from across the country are collaborating with IBM Watson Health to use the power of cognitive computing to personalize cancer treatment, according to a statement by the company.

Leading cancer institutes from across the country are collaborating with IBM Watson Health to use the power of cognitive computing to personalize cancer treatment, according to a statement by the company.

The goal is to optimize the decision-making process that determines which patient therapy to administer based on genetic information. Typically a committee of doctors called a “tumor board” evaluates mutation data to arrive at a recommendation.

“The way molecular tumor boards report results is neither scalable nor perfect,” said Norman Sharpless, the director of the Lineberger Comprehensive Cancer Center at the University of North Carolina in Chapel Hill, NC, one of the participating cancer centers. “Its error prone and it won’t work as we do more. It’s going to require Watson or something like Watson.”

Advances in sequencing have facilitated a new era of molecular information in oncology research, and identified mutations linked to cancers continue to accumulate. This mountain of data, mostly in the form of documents or tables that are not easily searched or even understood by many clinicians, could be better utilized.

This is where cognitive computing by IBM’s Watson comes in. In 2011, Watson found fame on the game show “Jeopardy!” by beating two former champions at the trivia contest. Cognitive computing simulates human though processing using self-learning systems combined with data mining, pattern recognition, and natural language processing to mimic the way the human brain works.

IBM states that Watson can process up to 60 million pages of text per second, including prose, or “natural language." The vast majority of medical information is unstructured and includes physician notes, journal articles, and raw numbers in public health databases. Researchers have turned to Watson for its unique capability to process this information.

The longer Watson played Jeopardy, the better it got. Similarly, the computer must first learn to understand medical information, a training process that began with participating cancer centers--such as Memorial Sloan Kettering in New York--feeding clinical information extracted from real cases and teaching it how to interpret the data. The process of pulling relevant information and sometimes making sense of conflicting information in medical cases is more complex than a Jeopardy clue.

Nevertheless, Ari Caroline, Memorial Sloan Kettering’s director of quantitative analysis and strategic initiatives, predicts that Watson will prove valuable. As clinical options and subtleties around biomarkers expand, clinicians will, “need a tool like Watson because the complexity and scale of information will be such that a typical decision tool couldn’t possibly handle it all.

The tool is designed for oncologists anywhere to make the best decisions for individual patients. Watson arrives at several options and displays confidence levels for each, providing supporting evidence from guidelines, published research, and the breadth of knowledge input from the community of oncologists.

“When you’re dealing with cancer, it’s always a race,” said Lukas Wartman, MD, assistant director of cancer genomics at The McDonnell Genome Institute at Washington University in St. Louis. A cancer survivor himself, he understands the power of genomic information in cancer treatment. In his case, fellow researchers at the Genome Institute compared Dr. Wartman’s leukemia tumor cell genomes to normal genomes and found a number of mutations, but none were treatable with targeted therapies. RNA analysis revealed significant overexpression of FLT3. A drug-gene interaction database pointed to the drug sunitinib (Sutent), normally used in kidney cancer, as specific for FLT3. Treatment with sunitinib was successful in putting his cancer into remission.

“Unfortunately, translating cancer-sequencing results into potential treatment options often takes weeks with a team of experts to study just one patient’s tumor and provide results to guide treatment decisions. Watson appears to help dramatically reduce that timeline," said Dr. Wartman.

As cancer centers put Watson to use, results are predicted to continually improve. Among the first to participate in the program are Ann & Robert H. Lurie Children's Hospital of Chicago; BC Cancer Agency in British Columbia; City of Hope, in Duarte, California; Duke Cancer Institute in North Carolina; McDonnell Genome Institute at Washington University in St. Louis; New York Genome Center, Sanford Health in South Dakota; University of Kansas Cancer Center; University of Southern California Norris Comprehensive Cancer Center, Yale Cancer Center, and the University of Washington Medical Center.