IBM’s Watson for Oncology, a computer that helps oncologists make treatment decisions for individual patients, made the same recommendations 90% of the time as a team of specialist oncologists, according to a new study presented on Friday, December 9, at the San Antonio Breast Cancer Symposium.
-----
Related Content
Primary care docs participating in breast cancer decisions
Early anti-HER2 therapy may help tailor treatment in breast cancer
-----
Watson for Oncology is a computing system that assists oncologists with making treatment decisions for individual patients. Watson bases its treatment decisions on individualized patients’ attributes and medical history as well as a wealth of published medical literature. Within a few seconds, physicians receive a recommendation on whether the standard treatment should be followed, considered, or abandoned.
In a double-blind study involving 638 breast cancer cases, Watson was pitted against the Manipal Comprehensive Cancer Center’s (Bengaluru, India) molecular tumor board, a group of doctors that meets weekly to evaluate cancer cases. Overall, Watson made the same recommendations as doctors 90% of the time (n=574).
When broken down into breast cancer subgroups, Watson had lower rates of agreement for more complex cases: the system agreed with the molecular tumor board on 80% of non-metastatic disease cases, 45% of metastatic disease cases, and 35% of HER2/neu-negative disease cases. These lower rates of concordance were expected, the researchers said, because there are many more treatments, variables, and factors to consider in such cases. “This increases the demands on human thinking capacity. More complicated cases lead to more divergent opinions on the recommended treatment,” said S P Somashekhar, MS, MCh, chairman of the cancer center.
Dr Andrew Norden, Watson Health deputy chief, concluded that the 90% concordance rate validates the Watsom system as useful. IBM currently has other concordance studies underway to support these results. Perhaps the largest benefit Watson offers, the researchers said, is alleviating the burden on doctors to research treatments, allowing them to interact more with the patients and learn more about their values and goals.
