Researchers say new computer models aim to better classify injuries
Engineers from Purdue University and researchers from the Liberty Mutual Research Institute for Safety are creating technology to comb through thousands of injury reports in large administrative medical datasets or insurance claims data to automatically classify them based on specific words or phrases. The reports -- usually filled out by employers, health care professionals, or the claimants themselves -- are currently classified by manual coders. These coders are hired by users such as the National Center for Health Statistics, hospital staff, or insurance industry handlers who review thousands of "injury narratives" included in reports, a process that can be extremely labor-intensive.
Mark Lehto, associate professor at Purdue University's School of Industrial Engineering, said researchers assigned codes to injury reports from workers' compensation claims using two different models developed with a technique called "Bayesian methods." The models calculated the probability that reports would be classified by human coders in specific categories. One model, called "naive," reviewed individual words, and the other, called "fuzzy," looked at sequences of words and phrases in the narratives, such as "fell off a ladder." The researchers used a database of 14,000 claim cases.
"The predictions were quite good," he said. "The results were comparable to the human coders. The accuracy is surprising considering all of the misspellings, run-on words, abbreviations and inconsistent or missing punctuations seen in these workers' compensation claim narratives."
Lehto said the new models might lead to programs that code reports as they are being filed. These models, he said, can be easily updated to deal with new types of accidents.
The findings were detailed in a recent issue of Injury Prevention journal.
Read more at the WORKERSCOMP ForumTM homepage.
October 1, 2009
Copyright 2009© LRP Publications