Sensitivity and Specificity of Oculometrics Under Fatigue Stress Compared to Performance and Subjective Measures.

摘要

We wished to compare the sensitivity of oculometric measures under fatigue stress to the sensitivities of performance and subjective measures, and to assess the specificity of oculometrics for predicting performance effects. We used data from the first night of sleep deprivation in a sleep aids study. Each of 13 subjects was represented four times in the final data set, with each of these four nights separated by at least a week. We focused on oculometric, simple cognitive task and subjective data and on two test periods within the night of sleep deprivation: at 21:00 and 03:00. All data were standardized as within-subject z scores across weeks. There were large, reliable differences in cognitive task performance and subjective assessments, in the expected directions, across the two measurement periods. There were smaller, less reliable differences in three of the four ocular measures, in the expected directions, across the two periods. We attempted to use hierarchical stepwise multiple linear regression as a first step in assessing the specificity of the three sensitive ocular measures for predicting the effects of the fatigue manipulation on cognitive performance. We selected simple response time and simple cognitive processing throughput as targets for prediction. Our strategy was to use the first 3 weeks of data to develop the regression equations, and then the last week of data for an assessment of specificity. The fatigue-related changes (from 21:00 to 03:00) in the three ocular measures were such poor predictors of the changes in logical reasoning throughput and simple response speed that we were unable to proceed with the specificity analysis. Overall, we were able to conclude that the individual ocular measures were less sensitive than the performance and subjective measures to this particular fatigue manipulation.