Predicting the Timing and Duration of Sleep in an Operational Setting Using Social Factors

Abstract

In recent years, there has been increasing interest in the use of bio-mathematical models to predict alertness, performance, and/or fatigue in operational settings. Current models use only biological factors to make their estimations, which can be limited in operational settings where social and geo-physical factors also dictate when sleep occurs. The interaction between social and biological factors that help determine the timing and duration of sleep during layover periods have been investigated in order to create and initially validate a mathematical model that may better predict sleep in the field. Participants were 32 male transmeridian airline pilots (17 captains, 10 first officers, and 5 second officers) flying the Sydney-Bangkok-London-Singapore-Sydney (SYD-LHR) pattern. Participants continued their regular schedule while wearing activity monitors and completing sleep and work diaries. The theoretical sleep timing model underpinning this analysis consists of separate formulations for short (<32 h) and long (>32 h) break periods. Longer break periods are split into three distinct phases-recovery (break start until first local night), personal (first local night until last local night), and preparation phases (last local night until break end)-in order to exploit potential differences specific to each. Furthermore, an iterative procedure combining prediction and retrodiction (i.e., using future duty timing information to predict current sleep timing) was developed to optimize predictive ability. Analysis found an interaction between the social and circadian sleep pressures that changed over the break period. Correlation analysis indicated a strong relationship between the actual sleep and new model's predictions (r = 0.7-0.9), a significant improvement when compared to existing models (r = 0.1-0.4). Social and circadian pressures play important roles in regulating sleep for international flight crews. An initial model has been developed in order to regulate sleep in these crews. The initial results have shown promise when applied to small sets of data; however, more rigorous validation must be carried out.

Modelling fatigue and the use of fatigue models in work settings

“…Black boxes indicate periods of wakefulness. This figure is reproduced from Kandelaars (2006). at both the group and individual level.…”

“…Employee demographics, cultural obligations and personal lifestyle choices along with work factors can all influence sleep-wake behaviour. Thus, field-based data (Gander et al, 1998;Kandelaars et al, 2006a;Lowden and Åkerstedt, 1998) have clearly demonstrated that psychosocial factors can drive the timing and duration of sleep.…”

“…Field-based investigations of aviation pilots show a strong and consistent trend for pilots to adjust the timing of sleep to coincide with the day/night and social zeitgebers of the local time zone (Gander et al, 1998;Graeber et al, 1986;Lowden and Åkerstedt, 1998). These studies reveal that for aviation pilots the timing and duration of sleep is strongly influenced by non-biological factors, and that pilots can voluntarily over-ride the circadian propensity to sleep during the biological night in order to socialize at socially appropriate times and attempt (often successfully) to sleep at biologically inappropriate times when there are no social opportunities available (Kandelaars et al, 2006a). As a consequence, the timing of sleep in workplace settings can be a complex interaction of volitional psychosocial factors as well as biological factors related to prior sleep-wake and circadian time.…”

“…Comparison of one and two-step approaches using data obtained from airline pilots (n = 21) flying from Australia to Bangkok, London and Singapore. The figure (from Kandelaars, 2006) includes a comparison of both the alertness levels obtained using the one-step approach (with 95% CI) with the two-step approach output and self-rated Samn-Perelli fatigue ratings completed pre-and post-duty periods. wake behaviour.…”

Fatigue and Performance Modeling

“…43 However, this model does not take into account social and other nonbiological constraints on sleep. 44 For this reason, biomathematical models applied to work schedules may make use of an alternative method for estimating potential sleep given the available sleep opportunities.…”