The use of the logistic model in space motion sickness prediction

The one-equation and the two-equation logistic models were used to predict tested subjects' susceptibility to motion sickness in KC-135 parabolic flights using data from other ground-based motion sickness tests. A data set containing data from 6 provocative tests, 2 vestibular function tests, and 1 motion sickness experience questionnaire from 162 subjects was used in this study. The prediction results from the logistic models were compared with those from the previously-used Bayes linear discriminant analysis procedures. The results based on this data set show that the logistic models correctly predicted substantially more cases (an average of 13%) in the data subset used for model building. In the data subset used for model cross-validation, the logistic models correctly predicted 4% and 5% more cases in the prediction of vomit or nonvomit, and of degree of susceptibility, respectively. Overall, the logistic models ranged from 53 to 65% predictions of the three endpoint parameters, whereas the Bayes linear discriminant procedure ranged from 48 to 65% correct for the cross validation sample.