Initially thought up as an electronic challenger to the rigors of the television quiz show Jeopardy, IBM’s Watson has since moved on to more practical, commercial applications. The computer is being adapted for use in telecommunications, legal research, finance, and other industries that require analysis of large pools of data. Famous for its goal of being able to understand communication and language in human contexts, it now seems the computer system will be saving lives as well.
The Lineberger Comprehensive Cancer Center at the University of North Carolina is home to its own Watson supercomputer. Researchers at the facility required the system for a study of genome research. The study required sifting through large pools of data, ranging from genetic information databases as well as information from past studies. The ultimate goal of the program was to uncover treatment options for some 1,018 patients who presented with tumors that possessed specific types of gene abnormalities. After Watson sifted through data and made treatment suggestions, they were compared against care recommendations offered by a panel of doctors and cancer specialists.
703 of Watson’s recommendations matched those offered by the panel, confirming the system’s accuracy, More importantly, however, were the treatments overlooked by the panel. After the study, Watson was able to identify and offer possible treatment modes for a further 323 cases, 96 of which had not been identified by the panel of experts. The panel clarified that these cases presented with “actionable alterations” consisted of eight specific genes not initially deemed treatable by the board. For patients who presented with these genes, Watson primarily suggested medical trials, one of which had only begun a week before the demonstration.
While the system is far from perfect, it is a useful way for doctors to sort through large pools of data, as well as an extra means of offering treatment modes for patients. In other words, while the system won’t soon replace doctors as the main source of diagnosis, it can act as a useful supplement in some cases. William Kim, a practicing doctor and professor at the University of North Carolina, stated his belief that the system will one day help doctors evaluate potential treatments for patients in a more “systematic manner.” That said, he clarified that more data is needed before cognitive computing becomes a full-fledged part of precision cancer care, which customizes therapies on an individual basis. At the very least, the study is a useful proof-of-concept.