Programmers Develop AI Solutions to Read, Classify Injury Records
NIOSH has announced the winners of an international programming competition to develop an algorithm that best uses artificial intelligence, or AI, to automatically read injury records and classify them in occupational health and safety surveillance systems. When an employee is injured at work, someone—or, soon, something—must write explanations of how the injury occurred, read all the narratives, and assign codes to classify injuries. According to NIOSH, this process takes time and is open to human error. The agency worked with the National Aeronautics and Space Administration’s Tournament Lab and vendor Topcoder to host the recent online competition, which was focused on exploring the use of AI to automate data processing in OHS surveillance systems. First place went to Raymond van Venetië, a doctoral student in Numerical Mathematics at the University of Amsterdam in the Netherlands, whose submission improved NIOSH’s ability to classify worker injuries from the agency’s baseline of 82 percent accuracy to nearly 90 percent accuracy. NIOSH will use van Venetië’s programming solution to build a web tool that OHS professionals can use to classify injury narratives.
Competitors from around the world, including Poland, Kenya, Macedonia, Romania, Bangladesh, Russia, China, India, and the United States, submitted 961 entries. More than half of the submissions came from the latter two countries, with 21 percent of submissions coming from India and 32 percent from the U.S.
Other top-placing participants include a senior data scientist at Sherbank AI lab in Russia; a developer and data scientist from China; a biostatistician at the School of Medicine at Emory University in Atlanta, Ga.; and a full stack engineer from Bangalore, India.
This was the first competition of its kind from NIOSH and its parent organization, CDC. An overview of the competition is available from NIOSH’s vendor Topcoder, a company with an open global community of designers, developers, data scientists, and competitive programmers.