Premium
A bayesian approach for predicting judged hearing disability
Author(s) -
Phaneuf Richard,
Hétu Raymond,
Hanley James A.
Publication year - 1985
Publication title -
american journal of industrial medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.7
H-Index - 104
eISSN - 1097-0274
pISSN - 0271-3586
DOI - 10.1002/ajim.4700070410
Subject(s) - medicine , bayesian probability , audiology , hearing loss , physical medicine and rehabilitation , artificial intelligence , computer science
A method of determining the cutoff point for an administrative decision to award compensation is proposed. To construct the predictive system a Bayesian approach and discriminant analysis were employed. Judged hearing disability was used as the criterion with audiometric scores as the determining variables. The common law doctrine of the balance of probability was used as the criterion, namely the 50th centile, on which to propose a cutoff point. The highest precision in predicting judged hearing disability was obtained with an average audiometric score at 1,000, 2,000, 3,000, 4,000 Hz in the worse ear. Assuming that judged hearing disability is a valid predictor of handicap, the cutoff point based on the balance of probability (50th centile) was obtained at 25 dB. The study also confirmed results from previous studies: (1) hearing sensitivity in frequencies higher than 2,000 Hz is required to predict hearing disability and handicap, (2) judged hearing disability is better correlated with hearing sensitivity in the worst ear, and (3) the audiometric cutoff point for a medical legal definition of impairment should be lower than what certain technical groups have proposed in the past.