NIOSHTIC-2 Publications Search
Commentary on a model to predict work-related fatigue based on hours of work.
Aviat Space Environ Med 2004 Mar; 75(3)(1):A72-A73
THE SUBJECT OF THIS COMMENTARY, the Fatigue Audit InterDyne (FAID) predictive model, is designed for use in operational settings to provide quantitative estimates of fatigue based on an individual's work schedule. The model is straightforward and simple to implement as it is based on two inputs to estimate fatigue-duration of work (e.g., hours on shift) and time of day. From these inputs, a fatigue score for a given time in a schedule is calculated in arbitrary units and assigned a degree of severity (e.g., low, medium, high) relative to the benchmark schedule of 40 daytime hours per week (the modal work schedule in industrialized countries). This information can then be used to estimate the relative fatigue risk of a work schedule so that new schedules can be designed, or so that fatigue-sensitive work tasks can be timed to avoid the hours of high fatigue. Perhaps the greatest strength of the model is its simplicity and the minimal amount of readily available data input needed for the fatigue estimates. The required input data often can be obtained from records kept by the worker or the organization. To maintain this simplicity, and thus accessibility by non-technical users, detailed information about actual work/rest patterns or work demands and workload are not obtained and the model is not adjusted for these variables. Components of the model are quite similar to other circadian rhythm and sleep/wake cycle models discussed in the Workshop. These models calculate alertness/ sleepiness/fatigue based on time awake (or time at work) and an estimate of the circadian rhythm of arousal with a peak in the late afternoon and a trough in the overnight/early morning hours. Unlike other models, however, the FAID does not estimate circadian arousal levels using a sinusoidal function (symmetrical or otherwise) that is explicitly included in the model. Rather, the circadian component appears to be based on the likelihood of being asleep at a given time of day. That likelihood appears to be based on sleep propensity functions developed in laboratory studies (e.g., 2) and on 24-h sleep frequency estimates derived from the authors' surveys of actual workers. This approach, and the emphasis in the writings toward accessibility by the general user, leave the impression that a precise pointestimation of fatigue is less important to these modelers than relative comparisons of fatigue that can be applied easily by those in the operational environment. For the intended users, the sinusoidal approach to point estimation probably is not necessary and might even discourage operational users from adopting the model. In future writings, however, considerably more detail about the derivation of the circadian estimates would be valuable to researchers or other mathematically-oriented readers. Some of the discussion at the modeling workshop implied that this derivation is proprietary information that is not readily available to the public. It would be helpful if the proprietary aspects of the model were clarified further by the authors. The FAID approach cannot be assumed to be inferior to the sinusoidal models as it remains to be seen whether the sinusoidal models are more or less successful than the FAID at predicting fatigue outcomes. It is hoped, of course, that the modeling workshop will demonstrate the utility of the various models in this regard. An attractive aspect of the FAID is the minimum amount of categorical output derived from the model (as in high, medium, or low fatigue). With sufficient validation, such simple metrics could be useful for broad-based epidemiological studies aimed at population estimates of fatigue risk associated with demanding work schedules. A gap in the FAID and other models presented at the workshop is their applicability to more physically demanding tasks, as opposed to the more perceptual, cognitive, or attention-based tasks typically used to validate these models. One wonders whether different relative weights for time on task versus circadian influences would be observed when the worker is confronted with tasks requiring more muscular than cognitive effort (e.g., manual materials handling versus monitoring natural gas allocation at a public utility). It was demonstrated some time ago, for example, that deficits in typical perceptual and cognitive tasks used in sleep deprivation studies could be produced in military personnel, who were not sleep deprived, by vigorous marching with a full backpack (1). In future developments of the FAID, demonstrations of how the model might be fine-tuned with more detailed information on sleep/wakecycles would be welcome, as would direct tests of whether modeling circadian adaptation to night work adds significantly to estimates of fatigue or risk from fatigue. In addition, large-sample demonstrations of the model's utility against measures of actual job performance, or improvements in practical safety indices, would be embraced enthusiastically.
Work-intervals; Work-practices; Workplace-studies; Fatigue-properties; Shift-work; Shift-workers; Models; Circadian-rhythms; Sleep-deprivation; Rest-periods
Roger R. Rosa, PhD, Senior Scientist, National Institute for Occupational Safety & Health, Office of the Director, Room 715H, Hubert H. Humphrey Building, 200 Independence Ave, SW, Washington DC 20201
Issue of Publication
Aviation, Space, and Environmental Medicine
Page last reviewed: May 5, 2020
Content source: National Institute for Occupational Safety and Health Education and Information Division