BIC

The cognitive performance of humans during civilian and military activities is affected by multiple factors, including limited and disrupted sleep, excessive information load, and physical and emotional stress. In particular, sleep deprivation affects the performance in armed-forces personnel involved in sustained military operations and civilians involved in shift work in a number of industrial sectors, such as medical care, emergency response, and ground/air transportation. The adverse impact of sleep disruption on human performance is well documented and is corroborated by various control studies on healthy subjects. It is necessary to develop biomathematical models which possess the capability to quantify the risk of fatigue and performance degradation at specified future times. These models can be used to determine the precise time and amount of countermeasures (e.g., naps and caffeine) that should be applied to yield performance peaks at a desired time and that can safely prolong peak performance.

Existing research in this area has focused on predicting a group-average performance impairment of a set of individuals having similar prior sleep/wake pattern and exposed to similar external conditions. Even if accurate, group-average models are inadequate for predicting performance of a specific individual, owing to substantial inter-individual differences in performance impairment due to sleep loss. This circumstance could be critical in both military and civilian settings when an emphasis is on predicting performance of one or two key individuals that may show performance deterioration dynamics significantly different than that of a group.

To address this problem, we have developed adaptive, individual-specific biomathematical models capable of predicting performance impairment of individuals subjected to total sleep deprivation. The physiologic underpinning of our model is Borbély's two-process model of sleep regulation, which postulates that performance impairment is a combination of the underlying sleep homeostatic and circadian processes. Numerical simulations performed on individuals exposed to sleep deprivation studies reveal that individualized models developed by our method considerably outperform group-average models in prediction accuracy (note closer fit of green [individualized] rather than red [group] line to blue performance measures in figure).

Our future efforts include: (1) the development of a performance model for individuals subjected to chronic, partial sleep restriction, (2) allow the ability to model the effects of countermeasures of sleep loss, such as naps, caffeine and pharmaceuticals, on performance impairment, and (3) represent the reliability of the model predictions by generating statistical error bounds on the predictions.

In summary, our work is focused on advancing the capability of existing biomathematical models by shifting the focus away from traditional, group-average performance models to more accurate, individual-specific models.