The accuracy of artificial neural networks in predicting long-term outcome after traumatic brain injury

Short Title:
The accuracy of artificial neural networks in predicting long-term outcome after traumatic brain injury
Model System:
TBI
Reference Type:
Journal Article
Accession No.:
J50964.
Journal:
Journal of Head Trauma Rehabilitation
Year, Volume, Issue, Page(s):
2006, vol. 21, issue 4, pp 298-314
Publication Website:
Abstract:
Study compared the accuracy of artificial neural networks (ANN), multiple regression, and classification and regression trees (CART) in predicting the outcomes of 1,644 patients 1 year after traumatic brain injury (TBI). The ANN procedure models complex, non-linear processes using nodes or layers, and is sometimes compared to the human brain's ability to process complex information. Variables from rehabilitation admission were used to predict discharge scores on the Functional Independence Measure, the Disability Rating Scale, and the Community Integration Questionnaire. Results showed no significant difference between ANN and multiple regression in predicting long-term outcome after TBI; both of these methods outperformed CART. However, because of the sophisticated form of multiple regression that was used, firm conclusions about the relative accuracy of ANN are limited.
Author(s):
Segal, Mary E.; Goodman, Philip H.; Goldstein, Richard; Hauck, Walter; Whyte, John; Graham, John W.; Polansky, Marcia; Hammond, Flora M.
Author Address(es):

Some items may be available for document delivery from the National Rehabilitation Information Center. Make a note of the title and any accession number and contact NARIC at 800-346-2742 to request a copy. There is a charge for document delivery from the NARIC collection.