Personalized sleep pattern providing system

Abstract

The present invention provides a personalized sleep pattern providing system comprising: a data collection unit configured to collect sleep-arousal data including sleep pattern information about a user and work-arousal data including information about the degree of arousal during a period of arousal of the user; an analysis unit configured to estimate a homeostatic sleep pressure and a biological sleep threshold on the basis of the sleep-arousal state data and the work-arousal data and calculate a predicted degree of arousal on the basis of the homeostatic sleep pressure and the biological sleep threshold; and a sleep pattern providing unit configured to generate a personalized sleep pattern on the basis of the predicted degree of arousal, wherein the personalized sleep pattern includes adaptive biologically-segmented sleep including required main sleep and sleep immediately before work.

Description

A system that provides personalized sleep patterns

[0001] The present invention relates to a system for providing a personalized sleep pattern, and more particularly, to a system for providing a personalized sleep pattern using a mathematical model based on user data collected by a wearable device.

[0002] As modern 24-hour society demands high levels of alertness and productivity outside of traditional working hours, individuals struggle to maintain wakefulness beyond the natural alignment of their circadian rhythms with daylight.

[0003] When our body's internal biological clock is influenced by external cues, such as the light-dark cycle, we can achieve high performance during the day and recover from nighttime sleep. However, in today's 24-hour society, approximately 20% of the global workforce works outside of traditional daytime hours due to constant labor demands. Furthermore, with the widespread exposure to artificial light, more than 80% of the general population engages in shift work-related lifestyles. This disrupts the normal sleep-wake cycle, leading to excessive sleepiness during work hours and insufficient sleep and sleep disorders outside of work hours. Irregular sleep-wake cycles and insufficient sleep alter physiological and psychological defense mechanisms, such as cardiometabolic stress, altered immune function, and cellular stress, which can increase the risk of negative health problems such as type 2 diabetes, cardiovascular disease, and cancer. Furthermore, fatigue and decreased attention, key consequences of sleep deprivation, increase the risk of mistakes and accidents during and after work, especially while driving.

[0004] An effective intervention to alleviate attention deficits in shift workers is to schedule adequate sleep during off-duty hours. The National Institute for Occupational Safety and Health (NIOSH) recommends that night shift nurses get as much sleep as possible and take preventive naps before 2 p.m. Several studies have shown that napping before a night shift can improve attention during the workday. Conversely, recent research suggests that sleeping as late as possible before a night shift may lead to better adjustment. However, this approach is impractical because shift workers must force themselves to stay awake before this delayed sleep. The optimal sleep-wake pattern for preventing attention deficits during shift work has not yet been identified. Importantly, there may not be a single, universally applicable intervention, regardless of an individual's sleep pattern.

[0005] Recent advances in wearable devices that track individuals' sleep history in real time have enabled personalized sleep interventions. One promising approach to analyzing sleep history and identifying effective personalized interventions involves using mathematical models to predict wakefulness based on two key processes: sleep processes, which reflect the body's physiological need for sleep, and circadian processes, which represent the regulation of sleep and wakefulness based on the body's internal biological clock. For example, Åkerstedt and Vital-Lopez et al. used mathematical models to identify feasible (i.e., practical and achievable) optimal sleep patterns that maximize wakefulness during work hours while avoiding long periods of low wakefulness during non-work hours. However, these models assume static circadian processes, which may be inappropriate for shift workers, whose circadian rhythms are significantly altered by reversed light exposure patterns. To address this limitation, recently developed physiologically based models incorporate the impact of light exposure on circadian processes to predict wakefulness. However, these models may have limited predictive accuracy because they predict wakefulness based on model-predicted sleep duration rather than actual sleep duration.

[0006] Recent advances in wearable devices have enabled real-time tracking of an individual's sleep history, enabling personalized sleep interventions to address these issues. Leveraging this technology, we developed a practical and flexible sleep intervention framework that accurately predicts real-time wakefulness and recommends a sleep schedule that maximizes and maintains wakefulness during target periods. Furthermore, we designed an application that implements this framework in a real-world setting, aiming to increase productivity and reduce safety risks outside of typical daytime activities.

[0007] The problem to be solved by the present invention is to provide a system that accurately predicts a user's real-time arousal level using a mathematical model and, accordingly, provides a customized sleep pattern to maintain a high state of arousal during shift work.

[0008] One embodiment for solving the above-described problem includes a data collection unit configured to collect sleep-wake data including user's sleep pattern information and work-wake data including information on the user's level of alertness during the user's wake-up time, an analysis unit configured to estimate homeostatic sleep pressure and a biological sleep threshold based on the sleep-wake state data and the work-wake data, and to calculate a predicted level of alertness based on the homeostatic sleep pressure and the biological sleep threshold, and a sleep pattern providing unit configured to generate a personalized sleep pattern based on the predicted level of alertness, wherein the personalized sleep pattern includes a personalized sleep pattern providing system including adaptive biological segmented sleep including required main sleep and sleep immediately before work.

[0009] One embodiment for solving the above-described problem includes a step of collecting, by a data collection unit, sleep-wake data including user's sleep pattern information and work-wake data including information on the user's level of alertness during the user's wake-up time, a step of estimating, by an analysis unit, homeostatic sleep pressure and a biological sleep threshold based on the sleep-wake state data and the work-wake data, and calculating a predicted level of alertness based on the homeostatic sleep pressure and the biological sleep threshold, and a step of generating, by a sleep pattern providing unit, a personalized sleep pattern based on the predicted level of alertness, wherein the personalized sleep pattern includes an operating method of a personalized sleep pattern providing system including adaptive biological segmented sleep including necessary main sleep and sleep immediately before work.

[0010] The problem to be solved by the present invention is to provide a system that accurately predicts a user's real-time level of arousal using a mathematical model and, accordingly, provides a customized sleep pattern to maintain a high state of arousal during shift work.

[0011] Figure 1A is a diagram showing data collected on shift work nurses.

[0012] Figure 1B is a graph showing that there is no correlation between total sleep time before shift work and KSS score.

[0013] FIG. 2 is a diagram showing the configuration of a personalized sleep pattern providing system according to one embodiment of the present invention.

[0014] Fig. 3 is a flowchart showing the operation method of the personalized sleep pattern provision system of Fig. 2.

[0015] Figure 4A is a drawing for explaining the operation of calculating the predicted arousal level in the analysis unit of Figure 3.

[0016] FIG. 4B is a diagram showing the relationship between the predicted arousal level and the actual arousal level calculated by the analysis unit according to an embodiment of the present invention.

[0017] Figure 4C is a drawing showing a comparison between the mathematical model of the present invention and three process models.

[0018] Figure 4D is a drawing showing a comparison between the mathematical model of the present invention and the post-nova model.

[0019] Figures 5A and 5B are drawings for explaining the effect of the adaptive bio-segmentation sleep of the present invention.

[0020] Figure 6 is a diagram illustrating providing a personalized sleep pattern based on the user's sleep history.

[0021] A drawing showing the best mode for carrying out the present invention is Fig. 2.

[0022] As used herein, the term “device” or “part” refers to any combination of software, firmware, and / or hardware configured to provide the functionality described herein. For example, software may be implemented as a software package, code, and / or instruction set or instructions, and for example, hardware may include a single or any combination or assembly of hardwired circuitry, programmable circuitry, state machine circuitry, and / or firmware storing instructions executed by programmable circuitry.

[0023] Hereinafter, embodiments of the present invention will be described clearly and in detail to the extent that a person having ordinary skill in the art can easily practice the present invention.

[0024]

[0025] Figure 1A is a diagram showing data collected on shift work nurses.

[0026] Referring to Figure 1A, various data can be collected through a wearable device (110) and sleep diary (120) worn by a user. For example, the user may be a nurse working shifts at a hospital.

[0027] In one embodiment, sleep-wake state data may be measured through a wearable device (110) worn by a user. For example, the sleep-wake state data may include information on the activity level (black graph) and sleep amount (or sleep pattern) (gray area) of a shift nurse.

[0028] In one embodiment, shift-wake data may be collected through a sleep diary (120) maintained by a shift nurse. For example, the shift-wake data may include information on the shift nurse's work hours (hatched area) and wakefulness at the start (black arrow) and end (white arrow) of the shift.

[0029] In this specification, alertness refers to the Karolinska Sleepiness Scale (KSS). For example, the KSS is a single-item questionnaire ranging from 1 (very awake) to 9 (very sleepy and struggling to fall asleep), with scores of 7 to 9 indicating a high level of sleepiness.

[0030] Figure 1A presents data on multiple sleep patterns of shift-shift nurses before the start of their shifts. If the shift nurse's work hours fall during the day (between 6:00 AM and 6:00 PM (graph 18)), this indicates a day shift. If the shift nurse's work hours fall during the night (between 6:00 PM and 6:00 AM), this indicates a night shift.

[0031] Figure 1B is a graph showing that there is no correlation between total sleep time before shift work and KSS score.

[0032] Referring to Figure 1B, the solid black line represents the least-squares fit line, and r and p represent the Pearson correlation coefficient and the p-value of the Pearson correlation test, respectively. Total sleep time before shift work was not significantly correlated with the average KSS score during shift work, indicating that longer sleep duration does not necessarily guarantee higher alertness.

[0033]

[0034] FIG. 2 is a diagram showing the configuration of a personalized sleep pattern providing system according to one embodiment of the present invention.

[0035] Referring to FIG. 2, a personalized sleep pattern provision system (200) may include a data collection unit (210), an analysis unit (220), and a sleep pattern provision unit (230).

[0036] The data collection unit (210) may be configured to collect status data of individual system users. For example, the data collection unit (210) may include at least one of a wearable device (110) attached to the user in FIG. 1 and a sleep diary (120) entered by the user. For example, the wearable device may include a wrist actigraphy device, the Readiband.

[0037] Status data may include sleep-wake data and work-wake data. For example, sleep-wake status data may include information about a shift nurse's activity level, sleep patterns, etc. For example, work-wake data may include information about a shift nurse's working hours, level of alertness during wake time, etc.

[0038] The analysis unit (220) may be configured to estimate homeostatic sleep pressure and circadian rhythms in real time based on sleep-wake state data and work-wake data. For example, the analysis unit (220) may estimate dynamic circadian sleep thresholds in real time based on irregular sleep-wake data and irregular light exposure.

[0039] The analysis unit (220) may be configured to calculate a predicted arousal based on the homeostatic sleep pressure and biological sleep threshold estimated in real time. In one embodiment, the analysis unit (220) may calculate the predicted arousal as the difference value between the homeostatic sleep pressure and the biological sleep threshold. In one embodiment, the analysis unit (220) may calculate the predicted arousal through a linear mixed effects model based on the arousal obtained at the beginning and end of each work period.

[0040] The sleep pattern provider (230) can be configured to generate a personalized sleep pattern based on the calculated predicted arousal level and the user's available sleep time.

[0041] In one embodiment, the personalized sleep pattern may be Adaptive Circadian Split Sleep (ACSS). Adaptive circadian split sleep may include required primary sleep and pre-work sleep. The required primary sleep may include a period during non-work hours when homeostatic pressure is below the circadian sleep threshold, and the pre-work sleep may include a period following the required primary sleep and prior to the start of work.

[0042] The sleep pattern provision unit (230) can be configured to provide a personalized sleep pattern generated by the analysis unit (220) to the user in real time.

[0043] Fig. 3 is a flowchart showing the operation method of the personalized sleep pattern provision system of Fig. 2.

[0044] Referring to FIG. 3, in step S310, the data collection unit (210) may collect sleep-wake state data and work-wake data from shift nurses. For example, sleep-wake state data may include information on the shift nurses' activity level, sleep patterns, etc. For example, work-wake data may include information on the shift nurses' working hours, level of alertness during wake time, etc.

[0045] In step S320, the analysis unit (220) can estimate homeostatic sleep pressure and circadian rhythms in real time based on sleep-wake state data and work-wake data, and can calculate a predicted wakefulness level based on the homeostatic sleep pressure and circadian sleep threshold estimated in real time.

[0046] In step S330, the sleep pattern providing unit (230) can generate a personalized sleep pattern based on the calculated predicted arousal level and the user's available sleep time, and the sleep pattern providing unit (230) can provide the personalized sleep pattern generated by the analysis unit (220) to the user in real time.

[0047] Figure 4A is a drawing for explaining the operation of calculating the predicted arousal level in the analysis unit of Figure 3.

[0048] Referring to FIG. 4A, the analysis unit (220) can simulate homeostatic sleep pressure (black solid line) at the biological sleep threshold (black dotted line) using a physiologically based mathematical model that describes the human sleep-wake cycle to predict the level of wakefulness based on sleep patterns.

[0049] In one embodiment, a physiologically based mathematical model can simulate two major processes that influence sleep: homeostatic sleep pressure (black solid line), which increases while an individual is awake and decreases while sleeping, and circadian sleep threshold (black dashed line), which fluctuates according to internal circadian rhythms.

[0050] In one embodiment, the mathematical model may include a model that integrates forced wakefulness (FW) and forced sleep (FS), as disclosed in patent document KR 2409990 B1. Forced wakefulness may refer to a state in which the predicted wakefulness is less than 0, and forced sleep may refer to a state in which the predicted wakefulness is greater than 0.

[0051] For example, the amplitude of the circadian sleep threshold is altered by individual light exposure, which influences the internal circadian clock. If a mathematical model predicts that an individual should be asleep because their sleep pressure is above the sleep threshold, but they are actually awake, the model simulates forced awakening to prevent a decrease in sleep pressure. Similarly, if the model predicts that an individual should be awake but they are actually asleep, the model simulates forced sleep to prevent an increase in sleep pressure. Consequently, the model can simulate two processes by utilizing actual sleep-wake states rather than predicted natural states. As the difference between the circadian sleep threshold and homeostatic sleep pressure increases, the individual is expected to be in a more natural state of wakefulness and thus more aroused. This difference is called predicted arousal and can be used to predict actual arousal. For example, if the predicted arousal is negative, the individual is in a state of forced wakefulness, which indicates sleep deprivation and low arousal.

[0052] The analysis unit (220) can calculate the difference in homeostatic sleep pressure from the simulated biological sleep threshold as a predicted arousal (red and blue arrows indicate positive and negative predicted arousal, respectively).

[0053] FIG. 4B is a diagram showing the relationship between the predicted arousal level and the actual arousal level calculated by the analysis unit according to an embodiment of the present invention.

[0054] Referring to Figure 4B, a scatterplot (black and white circles) of actual alertness (KSS scores) and the mixed-effects model predictions (black solid lines) between them were obtained from four representative individuals with shift schedules that included a continuous 12-hour night shift (2N12), a continuous 8-hour night shift (2N8 or 3N8), and a continuous 3-hour night shift (3N8 only). In our case, the mixed-effects model accurately predicted nurses' KSS scores regardless of shift schedule. The mean absolute error was ~1, and the predictions reached 75% within ±1 range of the measured scores on both day and night shifts.

[0055] Figure 4C is a drawing showing a comparison between the mathematical model of the present invention and three process models.

[0056] Referring to FIG. 4C, the mathematical model used in the analysis unit (220) of the present invention estimates a dynamic biological sleep threshold to predict the level of arousal, but in contrast, the three process model assumes a static biological process.

[0057] It is known that the light exposure patterns of shift workers at night can significantly alter their biological clocks. Therefore, estimating dynamic biological rhythms, as described in the present invention, can more accurately predict the alertness of shift workers.

[0058] To test this, the predicted KSS score from the analysis unit (220) calculated using the dynamic sleep threshold was compared with the predicted score based on the arousal level predicted by the three process models. The comparison results show that the predicted KSS score from the mathematical model of the present invention is more accurate than the predicted results from the three process models. In particular, the prediction results of the present invention become more accurate as the phase difference between static biological processes increases (especially in the case of night shifts). This shows that estimating dynamic biological rhythms can more accurately predict the arousal level of shift workers during night shifts than assuming static biological rhythms.

[0059] Figure 4D is a drawing showing a comparison between the mathematical model of the present invention and the post-nova model.

[0060] Referring to Figure 4D, while the mathematical model utilized in the present invention utilizes both forced sleep and forced wakefulness, the postnova model simulates only forced wakefulness. Consequently, the sleep time simulated in the postnova model is shorter than the actual sleep time, resulting in a sleep mismatch.

[0061] Shift workers, in particular, frequently experience sleep inconsistency due to their daytime sleep patterns, often requiring forced sleep simulations. Sleep inconsistency can reduce the accuracy of wakefulness predictions.

[0062] To test this, we compared the KSS score predictions with those predicted by a post-nova model that only incorporates forced awakenings. The simulated sleep duration (black line) does not exactly match the actual sleep duration (hatched area), resulting in sleep mismatch (black arrow). In the present invention, compared to the post-nova model, the ±1% difference (±1% improvement rate) between our model and the post-nova model increases as the sleep mismatch ratio among the total data increases.

[0063] For individuals with high rates of sleep mismatch, the predicted wakefulness of the Post-Nova model is less accurate than that predicted by the mathematical model of the present invention. This highlights the importance of utilizing actual sleep patterns to accurately predict the wakefulness of shift workers.

[0064]

[0065] Figures 5A and 5B are drawings for explaining the effect of the adaptive bio-segmentation sleep of the present invention.

[0066] Referring first to Fig. 5A, applying the mathematical model of the present invention to the nurse's original sleep pattern predicted a significant decrease in the biological sleep threshold during the night shift (Fig. 5A (i) left). This resulted in a significant decrease in the predicted arousal level, resulting in a long period of forced wakefulness with a predicted arousal level less than 0 (Fig. 5A (i) right).

[0067] Previous studies have shown that a delayed sleep pattern, which delays sleep onset as close to the start of night work as possible, can effectively increase alertness during night work. Indeed, model simulations based on a delayed sleep pattern with the same sleep duration (Fig. 5A (ii) left) predict an effective reduction in forced wakefulness during night work (Fig. 5A (ii) right, hatched area). However, this delayed sleep pattern requires a long forced wakefulness period even during non-work hours before the night shift (Fig. 5A (ii) right, white area), making it difficult for shift workers to adhere to it. This explains why a previous study found a low rate of 26.8% of shift workers adhering to a delayed sleep pattern before the night shift.

[0068] Adaptive circadian sleep in the present invention includes necessary primary sleep, which lasts until homeostatic sleep pressure falls below the circadian sleep threshold, and a late nap taken before the start of shift work (Fig. 5A (iii) left). It is expected that necessary primary sleep and late nap can effectively reduce the period of forced wakefulness during both working and non-working hours (Fig. 5A (iii) right).

[0069] Referring to Figure 5B, the forced wake period is presented after applying adaptive circadian split sleep and the original sleep pattern while maintaining the same sleep time as before work each night. Compared to the original sleep pattern, adaptive circadian split sleep reduced the forced wake period during working hours (Figure 5B (i)), and the forced wake period during off-work hours barely increased (Figure 5B (ii)).

[0070] When adaptive circadian rhythm sleep was compared with delayed sleep patterns, the forced wake period during working hours was similar (Fig. 5B (iii)), but the forced wake period during off-work hours was significantly reduced (Fig. 5B (iv)).

[0071] The adaptive circadian sleep pattern of the present invention is a more suitable sleep-wake pattern for individuals than the delayed sleep pattern, while maintaining a low forced arousal state during night shifts. Furthermore, adaptive circadian sleep effectively induces a low forced arousal state even during evening shifts, demonstrating its applicability to various work schedules.

[0072]

[0073] Figure 6 is a diagram illustrating providing a personalized sleep pattern based on a user's sleep history. Hereinafter, providing a personalized sleep pattern to a user will be described with reference to Figure 2.

[0074] Referring to FIGS. 2 and 6, the data collection unit (210) may be configured to collect and store the user's sleep history. For example, the sleep history may include multiple sleep-wake data and multiple work-wake data of the user.

[0075] The analysis unit (220) may be configured to calculate a predicted arousal level in real time based on sleep history. In one embodiment, the analysis unit (220) may estimate homeostatic sleep pressure and dynamic biological sleep threshold in real time using a physiologically based mathematical model that describes the human sleep-wake cycle, and calculate a predicted arousal level in real time based on the difference between the homeostatic sleep pressure and dynamic biological sleep threshold.

[0076] The sleep pattern provision unit (230) can be configured to provide a personalized sleep pattern to the user in real time based on the predicted arousal level generated by the analysis unit (220).

[0077] In one embodiment, the sleep pattern provider (230) may be configured to receive the user's sleep start time from a user input. Based on the user's sleep start time, the sleep pattern provider (230) may be configured to generate an adaptive biological split sleep pattern that includes necessary main sleep and sleep immediately before work, which are necessary to reduce napping during work hours. The sleep pattern provider (230) may display the generated adaptive biological split sleep pattern on the user's terminal display.

[0078]

[0079] The above-described embodiments are specific examples for practicing the present invention. The present invention will encompass not only the embodiments described above, but also embodiments that can be easily modified or modified. Furthermore, the present invention will encompass techniques that can be easily modified and implemented using the embodiments described above. Therefore, the scope of the present invention should not be limited to the above-described embodiments, but should be defined not only by the claims set forth below, but also by equivalents of the claims of the present invention.

[0080] The present invention relates to a system for providing personalized sleep patterns. More specifically, it can be used in a system that provides personalized sleep patterns using a mathematical model based on user data collected through a wearable device.

Claims

1. A data collection unit configured to collect sleep-wake data including information on the user's sleep pattern and work-wake data including information on the user's level of alertness during the user's wake-up time; An analysis unit configured to estimate homeostatic sleep pressure and biological sleep threshold based on the sleep-wake state data and the work-wake data, and to calculate a predicted wakefulness level based on the homeostatic sleep pressure and the biological sleep threshold; A sleep pattern providing unit configured to generate a personalized sleep pattern based on the predicted arousal level, The above-mentioned personalized sleep pattern is a personalized sleep pattern providing system that includes adaptive biological split sleep that includes necessary main sleep and sleep immediately before work.

2. In paragraph 1, The above data collection unit is a personalized sleep pattern providing system including a wearable device attached to the user.

3. In paragraph 1, A personalized sleep pattern providing system configured to estimate the homeostatic sleep pressure and the biological sleep threshold using a physiologically based mathematical model that describes the human sleep-wake cycle.

4. In paragraph 3, A personalized sleep pattern providing system, wherein the analysis unit is configured to calculate the predicted arousal level based on the difference between the homeostatic sleep pressure and the biological sleep threshold.

5. In paragraph 1, The above required sleep period includes the period during which the homeostatic numerical pressure is less than the biological sleep threshold during the user's non-working hours, The above-mentioned sleep immediately before work is a system that provides a personalized sleep pattern that includes the period before the user starts work after the above-mentioned required main sleep.

6. A step of collecting sleep-wake data including information on the user's sleep pattern and work-wake data including information on the user's level of alertness during the user's wake-up time in the data collection unit; A step of estimating homeostatic sleep pressure and biological sleep threshold based on the sleep-wake state data and the work-wake data in the analysis unit, and calculating a predicted wakefulness based on the homeostatic sleep pressure and the biological sleep threshold; A step of generating a personalized sleep pattern based on the predicted arousal level in the sleep pattern providing unit is included. The above personalized sleep pattern is a method of operation of a personalized sleep pattern providing system including adaptive biological split sleep including necessary main sleep and sleep immediately before work.

7. In paragraph 6, The above data collection unit is an operating method of a personalized sleep pattern providing system including a wearable device attached to the user.

8. In paragraph 6, An operating method of a personalized sleep pattern providing system further comprising a step of estimating the homeostatic sleep pressure and the biological sleep threshold using a physiologically based mathematical model that describes the human sleep-wake cycle in the above analysis unit.

9. In paragraph 8, An operating method of a personalized sleep pattern providing system, comprising a step of calculating the predicted arousal level based on the difference between the homeostatic sleep pressure and the biological sleep threshold in the above analysis unit.

10. In paragraph 6, The above required sleep period includes the period during which the homeostatic numerical pressure is less than the biological sleep threshold during the user's non-working hours, The method of operation of a system for providing a personalized sleep pattern, wherein the sleep immediately before the above-mentioned work is a period prior to the start of the user's work after the above-mentioned required main sleep.

Images