Watson supercomputer bores of Jeopardy!, begins oncology career

Jeopardy! champ turned oncologist: The IBM Watson supercomputer. Watson uses 750 Power 7 servers, 2,800+ processing threads, and 16 TB of RAM.

Bored of playing Jeopardy!, forecasting the weather, teaching, and doing an array of other tasks, the Watson supercomputer has turned its attention to oncology, where it has recently proven its medical chops by expanding the treatment options for patients with cancer.

In a clinical trial with just over 1,000 cancer patients, Watson was able to identify additional treatment options for a full 32 percent of patients based on an extensive database of patient genomes and current medical literature. The results suggest that Watson can already outperform human “molecular
tumor boards” (MTBs) due to its ability to sift through and analyze vast quantities of data and rapidly synthesize appropriate treatment options. Essentially, Watson is tailoring precise treatment options for each individual patient based on what we know about their exact type of cancer down to the genome level. Pretty impressive stuff.

Authors of the study, published this week in the journal Oncology, were not shy to point out the limitations of human doctors in the area of cancer treatment, saying “Humans are poorly suited to tasks such as a comprehensive analysis of the enormous amount of new clinical information generated on a daily basis, and the present study shows that a cognitive computing approach such as WfG [Watson] can enhance care for a significant fraction of patients.”

Humans are still a critical part of the diagnosis and treatment system, yet, as they authors suggest “We have reached a juncture at which many of the tasks related to cancer care require a real-time analysis of large, continuously maintained datasets.”

It’s easy to see why: analyzing every possible treatment option can be an extremely time-consuming process when one considers the sheer volume of new clinical research findings and the huge stores of data produced from genomic sequencing. As stated in the article, in 2015 there were already 150,000 scholarly articles directly related to cancer, a number that’s growing at over 6 percent per year. At the time there were also 650 targeted therapies in development and thousands of clinical trials underway. The amount of data in all this is staggering.

Given the complexity of a disease like cancer — how it can manifest in so many different ways — there’s a desperate need to crunch numbers fast and accurately to ensure the best possible treatment across tens of thousands of cases. This is something Watson, and cognitive computing, can do better than we can. And unlike us puny humans, it’ll keep improving as new technologies and new data become available. We may just find that cognitive computing and AI are the best medical tools we’ve ever developed.


Leave a Reply

Your email address will not be published.