Scholarly journals publish staggering amounts of medical research. The National Library of Medicine at the National Institute of Health processes up to 4,000 references each day, resulting in more than 805,000 new citations in 2015 alone. Yet this still represents only a fraction of all medical research published each year.
How can busy clinicians -- who have barely enough time to even see patients -- keep pace with this constant stream of information? One possible solution stems from research being conducted by a team of students and professors at Carnegie Mellon University. In 2010, they launched the Never Ending Language Learner, or NELL, a machine that teaches itself to read. In the same way humans read to retrieve information, NELL scans websites for text patterns that it uses to learn discrete pieces of information (so-called “beliefs”). Since its inception, NELL has learned more than 2.5 million beliefs with a high level of confidence.
If NELL can learn so much from such a vast diversity of text, we believe we can design a machine to read and learn from the enormous, ever-evolving body of medical research, and make the findings and implications of every published study easily searchable. This is our goal in developing NELL Med – arming physicians with the latest research so they can make the best clinical decisions for their patients