Fuzzy linear regression models for assessing risks of cumulative trauma disorders

Abstract

Cumulative trauma disorders (CTDs) are a major problem in industry and today's office environments, especially for keyboard users. Since comprehensive knowledge about individual or a combination of risk factors contributing to CTDs of the hand and forearm is still lacking, developing methodologies to quantify relationships of risk factors is greatly needed. The purpose of this research was to build fuzzy linear regression models to reveal the relationship of CTD risk factors, to predict the injuries, and to evaluate risk levels of individuals. Twenty-seven keyboard users (twenty-two for model building and five for model validation) and three CTD experts participated in the model building. Four fuzzy models were built corresponding to four risk categories, and a final fuzzy linear model was established using AHP pairwise comparisons. Multicolinearity effects were addressed and a partial standard deviation scaling method was used to eliminate the effects. From a methodological point of view, fuzzy linear regression models provide useful insight into risk factors-CTD relationship.