Accession Number:



Can We Predict Cognitive Performance Decrements Due to Sleep Loss and the Recuperative Effects of Caffeine

Descriptive Note:

Conference Paper

Corporate Author:

U.S. Army Medical Research and Materiel Command Fort Detrick United States

Report Date:


Pagination or Media Count:



Warfighters are often subjected to challenging sleepwake schedules that hinder their cognitive performance. Countermeasures, such as timely short naps and caffeine, are often used to mitigate the effects of sleep loss on performance. However, the timing, duration, and dosage of these countermeasures should be optimal or near optimal to maintain high levels of performance at the desired times. To this end, mathematical models that can accurately predict the detrimental effects of sleep loss and the restorative effects of different dosages of caffeine on performance could be of great utility. Here, we present a mathematical model that predicts individualized cognitive performance for subjects exposed to the continuum of sleep loss from no sleep to partial sleep while considering the recuperative effects of caffeine. In particular, we developed and validated both group-average and individual-specific models on performance data obtained from four different studies. Results from the first two studies showed that a group-average model developed on one study could accurately predict the temporal dynamics of both total and partial sleep loss in another study, with 75 of the predictions falling within 2 standard errors of the measured data. The results also showed that, on average, individual-specific models provided 30 improved prediction accuracy compared with the group-average models. Importantly, we showed that once the model had been customized to an individual under total sleep loss, it could be directly applied to predict the same individuals performance under partial sleep loss, and vice versa. Results from the third and fourth studies showed that the group-average model that accounts for caffeine effects on performance can provide up to 90 improved accuracy in capturing the effects of a range of caffeine doses 50-300 mg over a model that does not account for caffeine.


Subject Categories:

Distribution Statement: