Musculoskeletal disorders are common in prolonged computer use. The dynamics of the relationship between musculoskeletal outcomes and duration of computer use is not easy to model by common statistical methods due to the nonlinearity of the data. The use of inappropriate statistical models increases the likelihood of drawing wrong conclusions. In this paper we present a new mathematical methodology for modelling such a dynamic data set. Data on simultaneous direct measures of computer usage and questionnaire diaries of discomfort ratings were analysed using singular value decomposition as a means to track dominant dynamic trends, which presented the common characteristic of the relationship between computer-related workload and musculoskeletal outcomes over time. The relationship was constructed explicitly as a dose-response functional relationship that was parametrized by body region parameters. Standard statistical software was employed to quantify the variability of the estimates due to the complicated feature involved in the data. A deterministic population model was then developed to simulate the dynamics of computer-related workload and musculoskeletal outcomes. A stochastic model was also proposed to describe the stochastic nature of dynamics for individual subjects. The model was verified by comparison with real data for forecasting purposes. Possibilities for extending the model to accommodate more complicated medical data are discussed. The mathematical model can be served as both a deterministic population model that can project the demographic consequences and a stochastic model that can describe the whole quantity evolution of the study subjects. The proposed methodology is a flexible and broadly applicable one, which can be utilized by a variety of medical research.
Copyright (c) 2008 John Wiley & Sons, Ltd.