Having learned the nuances of human language, improving its ability to search for and use data to support new procedures such as personalized oncology treatments, a new version of the IBM Watson supercomputer is taking its computing capabilities to the next level.
IBM Watson Discovery Advisor, now available, understands the languages of chemistry, biology, legal and intellectual property, according to the company, and early adopters that tested the new capabilities have reaped fast rewards.
At Baylor College of Medicine, researchers using Watson analyzed 23 million abstracts of scientific literature and identified 70,000 articles on “p53,” a tumor suppressing protein. They then analyzed the studies to predict other proteins that turn on or off activity in p53. While scientists have averaged one targeted protein discovery annually, Baylor researchers took just weeks to identify six potential proteins for new research.
Not only has Watson learned languages of science and law, but medical molecular compounds and structures, as well as the ability to analyze images, says Rob Merkel, vice president and healthcare leader at the IBM Watson Group. “Imagine a long chemical chain and Watson understands the composition and can compare it to other compositions.” Further, Watson has a form of conversational capability to maintain context of the information being analyzed.
Earlier forms of Watson were about getting to an answer, Merkel says. Discover Advisor is an exploratory process through intellectual property with tools to go through results quickly and offering researchers a new dimension of insight. Johnson & Johnson, for instance, is teaching Watson to read and understand scientific papers, a process that often can take a three-person team nearly a year just to collect and prepare data for analysis. The goal is to have Watson synthesize information directly from literature, enabling researchers to start asking questions about the data immediately, according to IBM.
For instance, pharmaceutical firm Sanofi is using Watson to discover alternative applications for existing drugs, which could significantly cut the time and cost of drug development.
Originally published by Health Data Management.