IBM's Watson being put to the test in healthcare

One of the leaders from the IBM Watson team discusses how the supercomputer could transform the way doctors make decisions.

SAN FRANCISCO -- IBM's Watson supercomputer might be best known by most people for winning Jeopardy, but the science behind the system is getting so much stronger that we could see the technology being implemented in various industries worldwide soon.

Rob High, vice President and chief technology officer for Watson Solutions within the IBM Software Group, cited healthcare as a prime example while speaking at the DataWeek 2012 Conference on Monday afternoon.

High admitted that IBM's philosophy and strategy with Watson is actually quite different from what is normally done at IBM. Typically, he continued, the approach is to maximize the distribution of risk.

But with Watson, IBM has taken a slightly different track by focusing on a much more narrow domain. For example, not only is IBM optimizing Watson for the healthcare industry, but it has narrowed the focus to oncology -- specifically breast cancer. High added that Watson will be put to work on research for blood and lung cancers after that.

The IBM Fellow asserted that there is "not a company in the world" at this point that denies the potential offered by analyzing big data for improving healthcare. High noted that insurance companies, in particular, see the tremendous advantage of increasing the quality of care and decreasing misdiagnoses, which he described as "expensive as hell."

Additionally, High said that treatments often don't work because the diagnosis was wrong, which is where Watson could come into play.

High explained that many of the algorithms on Watson are "substantially reused," but that a lot have been added in since the Jeopardy event. For instance, Watson now supports negation, which expands the supercomputer's power considerably.

"The difference between Jeopardy and healthcare was Jeopardy was a single sentence. In healthcare, we have 15-page electronic medical records we're plugging into this thing," High said while trying to explain how Watson's comprehension has evolved. Nevertheless, High acknowledged that the basic mechanisms are the same.

Thus, in healthcare, High said that Watson can take in a set of symptoms and diagnoses, but doctors could also add in family and patient histories, medications and test results, among other metrics.

"Now suddenly your whole diagnosis is going to shift," High posited.

For High personally, he said that it's not just the financial savings but also the humanitarian potential here is amazing.