AsianScientist (Sep. 5, 2017) – When Jeopardy! champion Mr. Ken Jennings realized that he had been beaten by IBM’s Watson supercomputer in a 2011 match, he bowed out with a touch of gallows humor.
“I, for one, welcome our new computer overlords,” he wrote by way of surrender, turning video screen into white flag.
Six years on, Watson has yet to achieve world domination. But its advanced machine learning capabilities, natural language processing skills and formidable computing power are being channeled towards a higher purpose: improving healthcare.
Watson can, for example, help oncologists find the best treatment options for cancer patients based on patient information, clinical obser vations and data from laboratory tests. If this sounds like magic, it’s not. With help from doctors at the Memorial Sloan Kettering Cancer Center in New York, Watson ‘trained’ by ingesting nearly 15 million pages of medical literature, Ms. Farhana Nakhooda, Director of Healthcare and Social Services at IBM Asia Pacific, tells Supercomputing Asia.
Based in the Watson Center in Singapore, Nakhooda draws from her training in biomedical research and her knowledge of IBM’s portfolio to help the company’s regional teams put together end-to-end solutions for their clients.
The oncologist’s right-hand man
While Watson the Jeopardy! champion ran on servers, the Watson of today is a cognitive computing system that runs in the cloud. Human experts use machine learning techniques to train it to interpret and apply the huge volumes of information it ingests; once that is done, Watson continues to learn through interactions with users.
Rather than being a black box, Watson shows the user which journal articles it based its treatment recommendations on. In addition, unlike an internet search, it will request more information if it doesn’t have enough to make a decision with sufficiently high confidence, said Nakhooda.
The physician, therefore, retains the final say. “It’s a decision support system—if the doctor wants to do something different, it’s entirely up to him or her,” said Nakhooda.
Another Watson service that helps oncologists factor genetic data into their decisions is also being used in Asia, says Nakhooda. While DNA sequencing is increasingly being used to detect mutations that can then be targeted with specific treatments, analyzing terabytes of genomic data remains onerous and timeconsuming. Watson automates this process, running the data against its accumulated knowledge, and coming back with treatment options.