Method and System for Measuring, Predicting and Optimizing Human Alertness

Description (excerpt)

This patent application is a continuation application from U.S. patent application Ser. No. 17/603,854, filed on Oct. 14, 2021, which is a national stage application from PCT App. No. PCT/US2020/028559, filed on Apr. 16, 2020, which claims priority to U.S. Pat. App. No. 62/834,822 filed on Apr. 16, 2019 and U.S. Pat. App. No. 62/853,147 filed on May 27, 2019, which are all hereby incorporated by reference. I. FIELD OF THE INVENTION At least one embodiment of this invention is intended to optimize human cognitive performance and alertness. In at least one embodiment, the system includes a portable computing device, such as a smartphone, and software that allows users to measure their current level of cognitive performance and/or alertness based on alertness impairment, in a further embodiment, to make predictions about their future cognitive performance (using predicted alertness impairment as representative of cognitive performance), and, in a still further embodiment, to provide an optimized caffeine consumption and/or sleep-wake schedule to optimize alertness (or minimize alertness impairment) for a desired period(s) such as a work period or non-work period. In at least one embodiment, the cognitive performance and/or alertness is determined using a model with or without individualized parameters. In at least one embodiment, the model uses a sleep latency and/or sleep duration model with or without the above-mentioned embodiments. In at least one embodiment, the system includes a plurality of portable computing devices and at least one server that is configured to communicate with portable computing devices continually and/or intermittently. II. BACKGROUND OF THE INVENTION Irregular work-rest schedules often result in poor sleep hygiene and, consequently, cognitive performance impairments similar to those caused by consuming 3 or 4 alcoholic drinks. Individuals in certain occupations, such as healthcare workers, first responders, pilots, safety and security officers, and military members engaged in operations, are often on irregular work schedules that preclude them from obtaining adequate sleep. This leads to poor sleep hygiene, which is reflected in increased sleepiness and impaired measures of cognitive performance, including deficits in alertness and sustained attention (Akerstedt, 1988). For example, medical residents on a heavy call rotation (˜80 hours per week, including a shift of ˜35 consecutive hours) undergo alertness impairments similar to those induced by a blood alcohol concentration of 0.05% (Arnedt et al., 2005), which causes marked deterioration of driving skills (Council on Scientific Affairs of the American Medical Association, 1986). Poor sleep hygiene is also exacerbated by practical and social factors, as well as by sleep periods that occur at times of day when circadian rhythms make it difficult to sleep. For instance, after being awake for 24 hours, an individual might expect to recover by sleeping for 8 hours. However, if the sleep period were to start at around 0700, the individual may remain asleep for only about 4.5 hours (Akerstedt and Gillberg, 1981) and the recovery would be truncated. Therefore, planning effective sleep schedules that improve sleep hygiene requires knowing whether the schedule is physiologically favorable and feasible—that is, whether an individual can actually fall asleep at the proposed time and stay asleep for the desired duration, given their recent sleep history and time of day. To address these questions, a few computational models have been previously proposed. In a seminal study, Borbély (1982) proposed the two-process model, which postulates that sleep is regulated by 1) a homeostatic process S that depends on sleep history and 2) a sleep-independent circadian process C. Process Sis assumed to continuously increase while an individual is awake, leading to a monotonic increase in sleepiness. During subsequent sleep, process S monotonically decreases, resulting in full restoration when the individual follows a regular schedule. The level of sleepiness is also non-monotonically modulated by process C, depending on the time of day, where its effects become evident when an individual transitions from feeling drowsy in the afternoon to feeling refreshed in the evening. In subsequent work based on the two-process model, Borbély et al. (1989) proposed a theoretical modeling framework to predict how long it takes an individual to fall asleep (i.e., sleep latency). It was hypothesized that sleep latency was proportional to the difference between the homeostatic process S and an upper threshold H that accounted for circadian variation. They proposed a sinusoidal function to represent H and chose its amplitude and skewness so that the simulated sleep latencies qualitatively agreed with the temporal pattern observed during sleep deprivation: relatively long sleep latencies during the regular waking period followed by a steep reduction to short sleep latencies during sleep deprivation (Akerstedt and Gillberg, 1981). However, they did not estimate the parameters of H to quantitatively predict sleep latency. Subsequently, Akerstedt and Folkard (1995) proposed a model of alertness regulation that linked the combined levels of processes S and C to a subjective measure of alertness and, based on experimental observations, extended the model to predict sleep latency as an exponential function of the predicted alertness (Akerstedt and Folkard, 1