A sleep health management method, wearable device and storage medium

Abstract

The application relates to the technical field of health management, and discloses a sleep health management method, a wearable device and a storage medium. The method comprises the following steps: monitoring and identifying emotional events of a user in a preset recent history in real time, accumulating current stress residual values of the emotional events, and obtaining a cumulative stress score at a current moment; estimating a DLMO reference point of the user based on historical sleep data of the user, determining a pre-sleep evaluation window period based on the DLMO reference point, calculating a sleep index according to environmental light exposure data, user behavior data and / or physiological stress data in the pre-sleep evaluation window period collected in real time and the cumulative stress score, and generating pre-sleep evaluation information and / or corresponding intervention suggestions according to the sleep index. Through daytime emotional accumulation quantification, personalized DLMO window period alignment and multi-source information fusion, the application achieves accurate and traceable pre-sleep evaluation and intervention suggestion generation, and changes sleep health management from post-event summary to proactive intervention.

Description

Technical Field

[0001] This application relates to the field of health management technology, and in particular to a sleep health management method, wearable device and storage medium. Background Technology

[0002] The inventors' research found that existing wearable devices have at least the following shortcomings in sleep health management:

[0003] Mood and sleep monitoring are isolated from each other and lack a causal relationship: Existing products (such as Huawei Watch) mostly use real-time point-in-time identification of instantaneous emotional states and can also provide sleep scores afterward. However, they treat the two as independent functions and do not use daytime mood fluctuations as a predictor of nighttime sleep quality.

[0004] Pre-sleep assessment is delayed and lacks forward-looking and causal guidance: Most devices focus on analyzing the state after "falling asleep" (sleep depth, sleep score), which is a "post-event summary." Current technologies generally lack real-time assessment before sleep onset for the key factor determining the speed of falling asleep—melatonin secretion and its influence by light, behavior, and physiological stress. In particular, although ambient light sensors are widely available, there is still no technology to apply the cumulative inhibitory effect of light on melatonin secretion to pre-sleep prompts.

[0005] Therefore, improvements to existing technologies are necessary.

[0006] The above information is provided as background information only to aid in understanding this application and does not constitute an assertion or admission that any of the above content can be used as prior art relative to this application. Summary of the Invention

[0007] This application provides a sleep health management method, wearable device, and storage medium to address the problems in the prior art, such as the lack of a causal relationship between emotions and sleep quality, and the lack of real-time assessment before falling asleep.

[0008] To achieve the above objectives, this application provides the following technical solution:

[0009] In a first aspect, embodiments of this application provide a sleep health management method, including:

[0010] Real-time monitoring and identification of users' emotional events in a preset recent historical period, summing up the current residual stress value of each emotional event to obtain the cumulative stress score at the current moment;

[0011] Based on the user's historical sleep data, the DLMO baseline point for the onset of melatonin secretion in dim light is estimated, and the pre-sleep assessment window is determined based on the DLMO baseline point.

[0012] The sleep index is calculated based on the ambient light exposure data, user behavior data, and / or physiological stress data collected in real time during the pre-sleep assessment window, as well as the cumulative stress score; the sleep index is used to characterize the user's tendency to fall asleep at the current moment.

[0013] Based on the sleep index, generate pre-sleep assessment information and / or corresponding intervention suggestions.

[0014] Optionally, the real-time monitoring and identification of the user's emotional events within a preset recent historical period, and the summing of the current residual stress values ​​of each emotional event to obtain the cumulative stress score at the current moment, includes:

[0015] Real-time acquisition of the user's physiological and motion signals;

[0016] When the physiological signal deviates from the preset personal resting baseline by more than a preset threshold, and the motion signal indicates that the user is in a non-motor state, a physiological arousal event is triggered and recorded;

[0017] When the physiological arousal event is triggered, the multimodal physiological features at the current moment are extracted, and the multimodal physiological features are input into a pre-trained emotion classifier to identify the emotion type, arousal level and / or effectiveness value corresponding to the physiological arousal event.

[0018] Based on the emotion type, the arousal level, and the effectiveness value, an initial impact value is assigned to the current physiological arousal event;

[0019] According to the corresponding preset half-life, calculate the decay of the initial impact value of each physiological arousal event from the time of occurrence to the current time, and obtain the current pressure residual value of each physiological arousal event at the current time.

[0020] The cumulative pressure score is obtained by summing the current residual pressure values ​​of all physiological arousal events that have been triggered within the preset recent historical period.

[0021] Optionally, the physiological signals include: heart rate (HR), heart rate variability (HRV), and / or skin conductance response (EDA);

[0022] The multimodal physiological characteristics include: HR rate of change, low-frequency to high-frequency power ratio of HRV and / or skin conductance level and its rate of change.

[0023] Optionally, the step of estimating the DLMO baseline point for the onset of melatonin secretion under dim light based on the user's historical sleep data, and determining the pre-sleep assessment window using the DLMO baseline point as a benchmark, includes:

[0024] Determine the habitual time to fall asleep based on the sleep time in the historical sleep data;

[0025] The DLMO reference point is obtained by subtracting a preset time offset from the habitual fall asleep time; the preset time offset is within the range of [1.5h, 3h].

[0026] Using the DLMO reference point as the center, extend forward and backward by a first preset duration to obtain the pre-sleep assessment window period.

[0027] Optionally, the step of calculating the sleep index based on the real-time collected ambient light exposure data, user behavior data, and / or physiological stress data during the pre-sleep assessment window, as well as the cumulative stress score, includes:

[0028] Obtain the time offset of the current moment relative to the DLMO reference point, and determine the basic rhythm score based on the time offset;

[0029] Based on the currently collected ambient light exposure data, user behavior data, physiological stress data, and cumulative stress score, light inhibition factor, behavioral regulation factor, physiological stress factor, and emotional stress factor are determined respectively.

[0030] The sleep index is estimated based on the baseline rhythm score, the light inhibition factor, the behavioral regulation factor, the physiological stress factor, and the emotional stress factor.

[0031] Optionally, the method for determining the basal rhythm score includes: obtaining a basal rhythm score function curve centered on the DLMO reference point and varying over time; obtaining the basal rhythm score corresponding to the current moment based on the function curve; wherein the function curve includes a first interval and a second interval located before and after the DLMO reference point; the basal rhythm score corresponding to the second interval is higher than the basal rhythm score corresponding to the first interval; within the second interval, the basal rhythm score rises positively over time, and the steepness of the rise is positively correlated with the user's sleep-wake regularity index, which is determined based on the fluctuation of the user's historical sleep-on time and wake-up time.

[0032] And / or,

[0033] The method for determining the light inhibition factor includes: during light exposure, collecting illuminance values ​​and blue light percentages at preset time units; integrating the weighted values ​​of the illuminance values ​​and blue light percentages unit by unit to calculate the light inhibition amount; after entering a dark environment, allowing the light inhibition amount to decay exponentially at a corresponding preset time constant; and calculating the light inhibition factor based on the light inhibition amount at the current moment, wherein the light inhibition factor is negatively correlated with the light inhibition amount.

[0034] And / or,

[0035] The method for determining the behavior regulation factor includes: when it is detected that the user's exercise intensity exceeds a preset vigorous exercise threshold within a preset time period before the pre-sleep assessment window, the behavior regulation factor is set to a value less than 1.0, and the behavior regulation factor is negatively correlated with the exercise intensity; when it is detected that the user remains in a horizontal posture for more than a preset time period, the behavior regulation factor is multiplied by a positive additive coefficient greater than 1.0.

[0036] And / or,

[0037] The method for determining the physiological stress factor includes: acquiring the user's short-term heart rate variability over a preset duration; comparing the short-term heart rate variability with the user's historical heart rate variability during the same period at rest; and setting the physiological stress factor to a value less than 1.0 when the short-term heart rate variability is lower than the historical heart rate variability during the same period by more than a preset standard deviation threshold.

[0038] And / or,

[0039] The method for determining the emotional stress factor includes: using a preset stress threshold as the center, mapping the cumulative stress score to an emotional stress factor between 0.5 and 1.0 through an S-shaped function; wherein, at the preset stress threshold, the rate of change of the emotional stress factor reaches its maximum value; as the cumulative stress score gradually deviates from the preset stress threshold, the rate of change of the emotional stress factor gradually decreases.

[0040] Optionally, the step of generating pre-sleep assessment information and / or corresponding intervention suggestions based on the sleep index includes:

[0041] The preset score range in which the sleep index falls is determined, the preset score range includes a first score range, a second score range, a third score range and a fourth score range with corresponding scores decreasing sequentially;

[0042] If the sleep index is within the first score range, it is determined that all factors are within the preset ideal range, indicating that the user is currently in a good sleep state.

[0043] If the sleep index is within the second score range, identify the main contributing factors that are causing the sleep index to decrease, and generate second assessment information and / or second intervention suggestions corresponding to the main contributing factors.

[0044] If the sleep index is within the third score range, identify multiple factors that cause the current decrease in the sleep index to form a superimposed inhibition, and generate third assessment information and / or third intervention suggestions corresponding to the multiple factors;

[0045] If the sleep index falls within the fourth score range, corresponding fourth assessment information and / or fourth intervention recommendations are generated.

[0046] Optionally, the sleep health management method further includes:

[0047] The direction of the user's circadian rhythm shift is determined based on the difference between the user's weekday sleep midpoint and rest day sleep midpoint.

[0048] Based on the real-time perceived multimodal scene information of the user, the corresponding real-time scene label of the user is determined; the current multimodal scene information includes: ambient light data, location and semantic information, behavioral and physiological state and / or time and schedule information;

[0049] Based on the rhythm offset direction and the real-time scene label, a target intervention strategy is matched from a preset cross-scene intervention strategy library and output; wherein, the cross-scene intervention strategy library contains multiple intervention rules, and each intervention rule is labeled with the applicable rhythm offset direction and the applicable scene label.

[0050] Optionally, the sleep health management method further includes:

[0051] Obtain the user's most recent actual sleep latency, which is the duration from the moment of lying down to the moment of falling asleep; based on the actual sleep latency, automatically calibrate the DLMO reference point, the preset half-life of the physiological arousal event, and / or the weights of each factor;

[0052] And / or,

[0053] The system continuously monitors the changes in the user's actual sleep latency and sleep midpoint shift index after intervention, and automatically adjusts the strategy type and / or intervention intensity of the intervention recommendations based on the changes in these trends. The sleep midpoint shift index is calculated by the difference between the user's weekday sleep midpoint and rest day sleep midpoint.

[0054] And / or,

[0055] Based on the most recently updated historical sleep data, the individual resting baseline, the DLMO baseline, and / or the sleep midpoint offset index are updated in a sliding window manner.

[0056] Secondly, embodiments of this application provide a wearable device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the sleep health management method described above.

[0057] Thirdly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions thereon, which are executed by a computer processor to implement the sleep health management method described above.

[0058] Compared with the prior art, this application has the following beneficial effects:

[0059] This application achieves precise pre-sleep assessment based on the accumulation of daytime emotions: It quantifies each daytime emotional event into a residual stress value with decay characteristics. The initial impact value assigned when each emotional event occurs gradually decays over time according to the half-life, but it does not disappear when it decays to zero. Instead, it continues to accumulate in a residual manner. The residual values ​​of all events are added together to form a cumulative stress score, which reflects the dynamic superposition and dissipation of daytime emotional impacts over time, rather than the instantaneous emotional state at a certain moment. This cumulative stress score is used as an input factor for calculating the sleep index, enabling the pre-sleep assessment to reflect the continuous impact of the cumulative effect of daytime emotions on nighttime sleep. At the same time, by determining the pre-sleep assessment window based on the user's DLMO benchmark, the sleep index calculation is triggered before the user falls asleep, thereby generating corresponding pre-sleep assessment information and / or intervention suggestions. This ensures that the assessment and intervention occur within the critical physiological transition range from weak to strong sleep drive. Therefore, the embodiments of this application overcome the shortcomings of the prior art, which suffers from insufficient assessment accuracy due to the isolation of the two, and can only conduct post-event summaries and cannot provide real-time guidance before falling asleep, by incorporating the accumulation of daytime emotions into the pre-sleep assessment.

[0060] This application achieves precise assessment triggering based on individual biological clocks: By estimating a user's personalized DLMO baseline based on their own historical sleep data, the pre-sleep assessment window automatically adapts to the user's individual rhythm, ensuring that assessment and intervention occur within the critical physiological transition range from weak to strong sleep drive, thus avoiding the problem that fixed-time triggering assessments cannot cover users with different rhythm types.

[0061] It achieves accurate and traceable assessment through multi-source information fusion: By integrating four types of information—ambient light exposure data, user behavior data, physiological stress data, and cumulative stress scores—during the pre-sleep assessment window, it comprehensively assesses sleep tendency from four dimensions: environment, behavior, physiology, and psychology. Compared with single-dimensional assessment, it has higher accuracy and causal interpretability, providing multi-dimensional data support for generating intervention recommendations with causal logic.

[0062] This application has other features and advantages that will be apparent from or will be set forth in detail in the accompanying drawings and following detailed description, which together serve to explain the particular principles of this application. Attached Figure Description

[0063] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0064] Figure 1 This is a flowchart of the sleep health management method provided in the embodiments of this application;

[0065] Figure 2 This is a schematic diagram of a "five-stage pipeline-style" sleep health management engine provided in the embodiments of this application. Detailed Implementation

[0066] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0067] To overcome the shortcomings of existing technologies, such as the lack of a causal link between emotion and sleep quality, and the lack of real-time assessment before falling asleep, please refer to [link to relevant documentation]. Figure 1 This application provides a sleep health management method, including:

[0068] S1. Real-time monitoring and identification of the user's emotional events in the preset recent historical period, summing the current residual stress value of each emotional event to obtain the cumulative stress score at the current moment.

[0069] Among them, the preset recent history period refers to the backtracking time period set in advance by the system to trace the user's emotional events. It is usually a preset time period to trace back from the current moment (such as from 0:00 on the day to the current moment, that is, within the day), so as to ensure that the accumulated stress score can reflect the user's accumulated emotional effect within the day.

[0070] Emotional events refer to discrete physiological arousal events triggered by fluctuations in a user's physiological signals and assigned specific emotional labels (such as anger, anxiety, excitement, or calmness) after being identified by an emotion classifier. Each emotional event includes attributes such as the time of occurrence, the initial impact value, and the preset half-life.

[0071] The current residual stress value refers to the remaining value after the initial impact value of any emotional event has decayed from the time of occurrence to the current time according to the corresponding preset half-life, reflecting the intensity of the physiological impact of the event at the current time.

[0072] The cumulative stress score is the sum of the current residual stress values ​​of all emotional events that have occurred within a preset recent historical period, used to quantify the total amount of stress accumulated by a user during the day.

[0073] S2. Based on the user's historical sleep data, estimate the user's DLMO (Dim Light Melatonin Onset) baseline, and determine the pre-sleep assessment window based on the DLMO baseline.

[0074] The DLMO baseline, which is the starting point of melatonin secretion in dim light, can be selected as the time estimated by subtracting a preset offset from the user's habitual fall-off time in historical sleep data. This time marks the starting point of melatonin secretion in the user's body.

[0075] The pre-sleep assessment window can be selected as a time window centered on the DLMO baseline and extended before and after it by a first preset duration. Within this window, the system assesses the user's current tendency to fall asleep in real time.

[0076] It's important to explain that this step uses the DLMO baseline as the center to determine the pre-sleep assessment window because DLMO is a reliable single physiological marker of the body's endogenous circadian rhythm phase, marking the starting point of the physiological transition from "daytime wakefulness mode" to "nighttime sleep preparation mode." Using the DLMO baseline to determine the pre-sleep assessment window essentially aligns the sleep tendency assessment with the user's own circadian rhythm phase, ensuring that both assessment and intervention occur within the critical physiological transition range from weak to strong sleep drive, avoiding assessment failure due to assessment too early (before sleep drive is formed) or too late (after the adjustment window has closed). Furthermore, since there are significant individual differences in the DLMO baseline among different users, estimating the user's DLMO baseline based on their own historical sleep data allows the pre-sleep assessment window to automatically adapt to the individual user's rhythm, balancing group adaptability with individual accuracy.

[0077] S3. Calculate the sleep index based on the real-time collected ambient light exposure data, user behavior data, and / or physiological stress data, as well as the cumulative stress score during the pre-sleep assessment window. The sleep index is used to characterize the user's tendency to fall asleep at the current moment.

[0078] Multi-source information refers to the collective term for four types of data collected from multiple dimensions and used in the calculation of the sleep index during the pre-sleep assessment window. These mainly include ambient light exposure data, user behavior data, physiological stress data, and cumulative stress scores.

[0079] Ambient light exposure data can be collected in real time by ambient light sensors deployed on devices, such as light intensity (illuminance value) and the proportion of blue light. This data reflects the degree to which the user's current environment physiologically inhibits melatonin secretion: the higher the light intensity and the greater the proportion of blue light, the stronger the inhibition of melatonin secretion, and the lower the drive to fall asleep.

[0080] User behavior data, such as intensity of physical activity, movement status, and body posture, can be collected in real time using accelerometers and gyroscopes deployed on devices. This data reflects the user's current level of physical activity and rest readiness: vigorous exercise can raise core body temperature, thus delaying sleep onset, while a sustained horizontal posture indicates that the body has entered a state of physical relaxation.

[0081] Physiological stress data can be collected in real time by heart rate sensors deployed on devices, which collect heart rate and heart rate variability signals, and extract the heart rate variability characteristics within the current time period. This data reflects the current balance of the user's autonomic nervous system: low heart rate variability indicates that the sympathetic nervous system is dominant, the body is in a state of stress, and it is not conducive to falling asleep.

[0082] The cumulative stress score is the sum of the residual stress values ​​of all emotional events that have occurred at the current moment, reflecting the total amount of physiological and psychological stress accumulated by the user during the day up to the current moment.

[0083] The above four types of information describe the user's current sleep preparation status from four dimensions: environment, behavior, physiology and psychology. This embodiment integrates the four types of information into a single sleep index through a multi-factor weighted model to achieve a comprehensive quantitative assessment of the user's sleep tendency.

[0084] S4. Based on the sleep onset index, generate pre-sleep assessment information and / or corresponding intervention suggestions.

[0085] Pre-sleep assessment information refers to descriptive information generated based on the sleep index during the pre-sleep assessment window to provide users with feedback on their current sleep readiness status.

[0086] Intervention suggestions refer to actionable prompts generated based on the sleep onset index, which guide users to adjust their behavior or environmental factors to improve sleep conditions during the bedtime window.

[0087] For example, assessment information tells users "what their current state is and where the problem is," while intervention suggestions tell users "what to do." The two complement each other, allowing users to understand their sleep readiness status and receive actionable guidance.

[0088] It is understood that the embodiments of this application have the following technical effects:

[0089] First, it enables accurate pre-sleep assessment based on accumulated daytime emotions.

[0090] This application quantifies daily emotional events into residual stress values ​​with decay characteristics. The initial impact value assigned when each emotional event occurs gradually decays over time according to its half-life, but it does not disappear upon reaching zero; instead, it continuously accumulates in a residual manner. The cumulative stress score is the sum of the residual values ​​of all events, reflecting the dynamic superposition and dissipation of daily emotional impacts over time, rather than the instantaneous emotional state at a particular moment. This cumulative stress score is used as an input factor in calculating the sleep onset index, enabling the pre-sleep assessment to reflect the continuous impact of the cumulative effect of daytime emotions on nighttime sleep. Simultaneously, by determining the pre-sleep assessment window based on the user's DLMO baseline, the sleep onset index is calculated before the user falls asleep, thereby generating corresponding pre-sleep assessment information and / or intervention suggestions. This ensures that the assessment and intervention occur within the critical physiological transition range from weak to strong sleep drive. Therefore, this application's embodiments overcome the shortcomings of existing technologies, such as insufficient assessment accuracy due to the isolation of daytime emotional accumulation and the inability to provide real-time guidance before sleep, by introducing daytime emotional accumulation into pre-sleep assessment.

[0091] Second, it enables precise assessment and triggering based on individual biological clocks.

[0092] This application embodiment estimates a personalized DLMO baseline based on the user's own historical sleep data, so that the pre-sleep assessment window automatically adapts to the user's individual rhythm, ensuring that the assessment and intervention occur within the critical physiological transition range from weak to strong sleep drive, avoiding the problem that fixed-time triggering assessment cannot cover users with different rhythm types.

[0093] Third, it enables accurate and traceable assessment through the fusion of multi-source information.

[0094] By integrating four types of information—ambient light exposure data, user behavior data, physiological stress data, and cumulative stress scores—during the pre-sleep assessment window, sleep tendency is comprehensively assessed from four dimensions: environment, behavior, physiology, and psychology. Compared with single-dimensional assessment, this method has higher accuracy and causal interpretability, providing multi-dimensional data support for generating intervention recommendations with causal logic.

[0095] In addition, the inventors also discovered the following shortcomings in existing wearable devices for sleep health management:

[0096] Intervention suggestions are homogenized and lack direction and context-specific adaptability: Existing solutions for regulating sleep generally "provide direction, but not a map." They are mostly general and static suggestions (such as "wake up at a fixed time" or "get more sun"), which neither adopt the opposite strategy based on whether the user's biological clock is "phase-delayed" or "phase-advanced," nor change the content of the suggestions and the timing of the push based on whether the user is "in a meeting in the office" or "taking a walk outdoors."

[0097] To overcome various shortcomings, this application also provides another method for sleep health management. For example... Figure 2 As shown, this embodiment constructs a "five-stage pipeline-style" sleep health management engine, where stages one through four are a forward data pipeline (diagnosis → monitoring → assessment → intervention); stage five is a reverse feedback loop (actual sleep data is fed back, while all model parameters from stages one through four are optimized, forming a closed loop). The specific implementation is as follows:

[0098] Phase 1: Establishing a personal circadian rhythm baseline (continuous operation).

[0099] Step 1.1: Calculate the physiological sleep midpoint offset index S_index.

[0100] The system can monitor a user's sleep data for at least 7 days, automatically distinguish between weekdays and rest days, record the time of falling asleep and waking up, and calculate the sleep midpoint time using the following formula: S_index = T_free - T_work;

[0101] Here, T_free is the midpoint of sleep on the most recent rest day, representing the natural phase preferred by the internal biological clock (unconstrained by work / school); T_work is the average midpoint of sleep over the most recent N workdays, representing the actual phase forced by the social clock. The difference between T_free and T_work quantifies "social jet lag"—the degree of misalignment between the user's biological clock and the social timetable.

[0102] If S_index > 0, it is determined to be "phase-delayed" (night owls forced to wake up early). "Phase-delayed" means that the user's internal biological clock is later than the social timetable. For example, on a weekday, the user is forced to get up at 7 am to go to work, but their body really wants to sleep until 9 am; therefore, the midpoint of sleep on a weekday is 3:30 am (sleep at 0:00 am - wake up at 7:00 am), and the midpoint of sleep on a rest day is 5:00 am (sleep at 1:00 am - wake up at 9:00 am). In this case, S_index = 5:00 - 3:30 = +1.5 hours.

[0103] If S_index < 0, it is determined to be "phase-advanced type" (early birds forced to stay up late). "Phase-advanced type" means that the user's internal biological clock is earlier than the social timetable. For example, on weekdays, the user is forced to go to bed at 11 pm, but the body is actually sleepy at 9 pm; therefore, the midpoint of sleep on weekdays is 3:30 am (sleep at 11 pm - wake up at 7 am), and the midpoint of sleep on rest days is 2:00 am (sleep at 9 pm - wake up at 5 am). In this case, S_index = 2:00 - 3:30 = -1.5 hours.

[0104] Offset severity classification: |S_index| < 1 hour → Slight; 1-2 hours → Moderate; > 2 hours → Severe.

[0105] Step 1.2: Establish DLMO reference points.

[0106] Analyze historical sleep data to calculate the user's habitual fall-off and wake-up times, determine the sleep midpoint, and estimate the DLMO baseline.

[0107] Based on extensive experimental research, the inventors discovered that melatonin secretion typically begins 1.5 to 3 hours before the usual time to fall asleep in dim light. Optionally, a default value of 2 hours is used, which is a relatively robust population average estimate.

[0108] Based on this, the DLMO baseline can be estimated by subtracting a specific offset from the habitual fall-off time, such as DLMO ≈ habitual fall-off time - 2h. It can then be further dynamically corrected based on real fall-off data through the self-calibration module in stage five.

[0109] Simultaneously, the standard deviation of the user's long-term resting heart rate rhythm and sleep-wake cycle regularity indicators is recorded. The smaller the standard deviation, the more regular the sleep-wake cycle; the larger the standard deviation, the more disordered the biological clock, which will affect the steepness of the subsequent basal rhythm score rise curve.

[0110] It's important to explain the relationship between steps 1.2 and 1.1: S_index reflects the "degree of misalignment" between the user's biological clock and the social timetable (reflecting the severity of social jet lag), while the DLMO baseline reflects the "intrinsic phase" of the user's biological clock (reflecting the position of the natural sleep window). Both characterize the circadian rhythm from different dimensions.

[0111] (1) S_index determines the "direction" of the four-stage intervention plan (phase delay type vs. phase advance type, different directions adopt opposite intervention logic);

[0112] (2) The DLMO baseline determines the “triggering time” (when to start the assessment during the DLMO window) and the starting point for calculating the baseline rhythm score in the third stage of sleep assessment.

[0113] Step 1.3: Establish your personal resting baseline.

[0114] Establish personal resting baselines for physiological indicators such as heart rate (HR), heart rate variability (HRV), and electrodermal activity (EDA) to accurately identify deviation events in the future.

[0115] Heart rate refers to the number of times the heart beats per minute. Heart rate variability refers to the minute fluctuations in the time interval between successive heartbeats; a higher variability indicates an active parasympathetic nervous system (relaxation system), while a lower variability indicates a dominant sympathetic nervous system (stress system). Skin conductance refers to the subtle changes in skin resistance / conductance; it is influenced by emotional arousal, such as tension, anxiety, and excitement, which increase sweating and thus conductance.

[0116] Personal resting baseline: the normal range of various physiological indicators of a user in a quiet, awake, and still state (which can be expressed as mean ± standard deviation).

[0117] Phase Two: Monitoring of Daytime Emotional Events and Quantification of Stress Accumulation (to be carried out throughout the day).

[0118] Step 2.1: Real-time monitoring and emotional event identification.

[0119] It collects heart rate, heart rate variability, skin conductance, and accelerometer data in real time. It employs a two-level recognition mechanism of "threshold triggering + classification judgment."

[0120] Level 1 – Deviation Detection: When the short-term sliding window mean of any physiological signal (HR / HRV / EDA) deviates from the personal resting baseline established in step 1.3 by more than N standard deviations (N is 1.5 by default), and no motion state is detected in the acceleration data at the same time (MET of exercise intensity < 3.0), a "physiological arousal event" is triggered.

[0121] This step will only trigger the "physiological arousal event" when two conditions are met simultaneously. This is because: the heart rate spikes during running and the HRV decreases after exercise, which are normal physiological reactions and are unrelated to emotions; the accelerometer data can identify that "the user is exercising," thus eliminating these interferences.

[0122] Physiological arousal events refer to brief responses in which physiological signals (HR / HRV / EDA) deviate significantly from an individual's resting baseline when the user is not moving, reflecting activation of the autonomic nervous system by some internal or external stimulus.

[0123] Level 2 – Emotion Classification: The multimodal physiological feature vector at the trigger moment (including HR rate of change, HRV frequency domain feature LF / HF ratio, EDA skin conductance level SCL and its first derivative) is input into a pre-trained emotion classifier, and the output is:

[0124] Type tags (anger, anxiety, excitement, calm);

[0125] Arousal level (high / medium / low, representing emotional intensity);

[0126] Effect value (positive / negative, representing pleasant or unpleasant).

[0127] HR rate of change characterizes the severity of heart rate changes. LF / HF ratio characterizes the balance between sympathetic and parasympathetic nervous systems (a high ratio indicates tension / anxiety). SCL characterizes baseline skin conductance, reflecting overall arousal. EDA first derivative characterizes transient changes in skin conductance, reflecting the response to specific stimuli.

[0128] It should be noted that the specific implementation of the classifier in this application embodiment is not limited, and various existing methods can be used, such as lightweight neural networks, SVM, or rule-based fuzzy inference systems.

[0129] Step 2.2: Calculate the cumulative emotional stress pool.

[0130] The intuitive principle of "one event, one score, individual decay, and final summation" is adopted:

[0131] Step 2.2.1 – Assign an “initial impact score” to each emotional event:

[0132] Each emotional event is assigned an initial score based on its type and intensity. Negative, high-arousal emotions receive higher scores, while positive, low-arousal emotions receive lower scores or even negative scores. For example: intense argument = 30 points, mild anxiety = 10 points, hearty laughter = -5 points (to offset stress), calm and relaxation = 0 points.

[0133] Step 2.2.2 – Calculate the “residual value” of each emotional event up to the current moment:

[0134] The impact of each emotion fades over time, and the rate of fading is determined by the "emotional half-life": high-arousal negative emotions fade slowly (e.g., anger has a half-life of 4 hours), while low-arousal emotions fade quickly (e.g., mild anxiety has a half-life of 1 hour).

[0135] Calculation formula: Residual value of an event = Initial impact score × (Hourly residual rate)^(Number of hours elapsed);

[0136] Wherein, the hourly residual rate = (0.5)^(1 / half-life).

[0137] For example: 30 minutes of anger, half-life of 4 hours, after 2 hours, the residual value = 30 × (0.84)² ≈ 21 minutes.

[0138] Step 3.2.3 – Add up all residual values:

[0139] At any given moment, the current residual values ​​of all historical emotional events of today are summed to obtain the "Cumulative Pressure Score" (CPS).

[0140] Phase 3: Multifactorial assessment of nighttime sleep onset tendency (pre-sleep trigger).

[0141] Step 3.1: Collect real-time impact factors.

[0142] During the DLMO window period before the user falls asleep (exemplarily defined as 1 hour before and after the DLMO baseline point, i.e., the time period from DLMO-1h to DLMO+1h daily), the system collects three types of data in real time:

[0143] Ambient light exposure data: Monitoring light intensity and blue light content using an ambient light sensor;

[0144] Behavioral data: Acquired through accelerometers, including intensity of movement, motion state, and body posture;

[0145] Physiological stress data: The cardiac relaxation index was analyzed using HRV to assess the body's relaxation state.

[0146] Step 3.2: Calculate the Sleep Propensity Index (SPI).

[0147] Input the above data into a multi-factor weighted model, integrate the daily cumulative stress score, and calculate a quantitative value of 0-100:

[0148] SPI = Basal Rhythm Score × Light Inhibition Factor × Behavioral Regulation Factor × Physiological Stress Factor × Emotional Stress Factor

[0149] Explanation of each factor:

[0150] Basal rhythm score: A function that changes over time, centered on the DLMO baseline. The score is extremely low (0-10 points) before DLMO is activated, and rapidly rises to 80 points after activation. The steepness of the rise curve is regulated by "routine regularity": the more regular the routine, the steeper the curve, and the faster sleepiness occurs.

[0151] Light-induced inhibitory factors: A dynamic integral model of "cumulative-decaying light-induced inhibition" was established to simulate the physiological inhibitory effect of light exposure on melatonin secretion. This model incorporates two symmetrical processes:

[0152] Integral sub-phase (during light exposure): Define a light suppression factor L_supp, and integrate minute by minute using the weighted average of light intensity and blue light percentage within the time window:

[0153] L_supp(t) = Σ[minute i] [ α × Lux(i) × (1 + β × BlueRatio(i)) ].

[0154] Where Lux(i) is the illuminance value (lux) at minute i, BlueRatio(i) is the proportion of blue light in the total illuminance, α is the illuminance normalization coefficient, and β is the blue light weighting coefficient.

[0155] Light inhibition factor = exp(-γ × L_supp), where γ is the inhibition sensitivity coefficient.

[0156] For example, 30 minutes of continuous mobile phone use (typical screen illuminance of about 150 lux, with a high proportion of blue light) can cause this factor to drop to about 0.7.

[0157] Attenuation phase (after entering a dark environment): L_supp decays exponentially at a fixed time constant τ.

[0158] L_supp(t) = L_supp(t0) × e^(-Δt / τ), where τ is taken as 30 minutes by default.

[0159] The lower limit of the factor was set at 0.3 (the human body retains a basic sleep drive even with extreme light exposure).

[0160] Behavioral modulatory factor: assesses the impact of exercise and posture on core body temperature and sleep readiness. When the accelerometer detects an exercise intensity exceeding MET ≥ 6.0 (metabolic equivalent, corresponding to the threshold for vigorous exercise) and this occurs within 1 hour before the DLMO window, this factor < 1.0 (the specific attenuation coefficient is positively correlated with exercise intensity); when the gyroscope + accelerometer continuously detects a horizontal posture (angle with the direction of gravity < 30°) for more than 15 minutes, a slight positive bonus (×1.05) is given, reflecting that the body has entered a physical rest readiness state.

[0161] Physiological stress factor: Compare the current 30-minute short-term HRV with the individual's historical HRV during the same period at rest. If the current HRV is significantly low (below 1 standard deviation), it indicates a state of stress, and the factor is adjusted down to 0.8-0.9.

[0162] Emotional Stress Factor: The current cumulative stress score (CPS) calculated in Stage 2 is mapped to a moderating factor of 0.5-1.0. The higher the CPS, the lower the factor. The specific mapping relationship is achieved through an S-shaped function, enabling fine-grained differentiation of the intermediate stress range and gradual saturation at both ends.

[0163] Emotional stress factor = 1.0 - 0.5 / [1 + e^(-(CPS - CPS_threshold) / k)];

[0164] in:

[0165] CPS_threshold is the stress threshold, with a default value of 50 (according to laboratory calibration, this value corresponds to the inflection point where the sleep latency begins to prolong significantly). It will be automatically adjusted by the self-calibration module in stage five based on individual physiological characteristics.

[0166] k is the curve steepness coefficient, with a default value of 10, which controls the width of the pressure-sensitive region. This parameter is also optimized by the Phase 5 self-calibration.

[0167] Step 3.3: Generate tiered early warnings and causal tracing prompts.

[0168] Based on the sleep index range and the reverse tracing of the main contributing factors, action suggestions with causal logic are generated, as exemplified by:

[0169] 80-100 points (sleep green light zone): The system identifies that all factors are close to the ideal value.

[0170] Tip: "You're in a great state to prepare for sleep, and melatonin is being secreted smoothly. Put down your phone and enjoy a good night's sleep."

[0171] 50-79 points (delayed secretion zone): The main reason for the decrease in score due to system retrospection.

[0172] If your light-inhibiting factor is low → "Current light levels may delay your sleepiness. It is recommended to go to a dimly lit room to help melatonin secretion."

[0173] If your emotional stress factor is low → "Your emotional stress has accumulated to a high level today, which may make it harder to fall asleep than usual. It is recommended to do some diaphragmatic breathing before bed to relax your nerves."

[0174] If behavioral regulation factors are low → "You just exercised, and your body temperature is still high. Wait for your body to cool down naturally before trying to fall asleep."

[0175] 20-49 points (secretion inhibition zone): The system identifies multiple factors that inhibit secretion.

[0176] Tip: "Your body is still in 'daytime mode,' with melatonin secretion suppressed in many ways. We recommend dimming all lights, staying away from screens, and listening to some soft music to help you relax."

[0177] < 20 points (severe inhibition zone): The system triggers emergency intervention.

[0178] Tip: "Today's overall condition may lead to severe difficulty falling asleep. Guided meditation is strongly recommended to give yourself a period of complete relaxation."

[0179] Explanation: The division of labor between Phase Three and Phase Four: Phase Three focuses on providing immediate assessment and prompts during the "bedtime window" (defined as one hour before and after the DLMO baseline, i.e., the period from DLMO-1h to DLMO+1h daily), answering questions like "Will I be able to sleep tonight? Why? What should I do now?" Phase Four, on the other hand, spans the entire day, answering questions like "How can I fundamentally adjust my sleep schedule? What should I do today?" based on the user's circadian rhythm shift profile. The two phases complement each other, jointly covering all levels of intervention from immediate emergency response to long-term conditioning.

[0180] Phase 4: Generation of cross-scene intervention solutions based on offset profiles and real-time scenes (instant push).

[0181] Step 4.1: Real-time perception of the user's current multimodal scene information.

[0182] The system integrates the following multi-source information to tag users with real-time scene information:

[0183] Ambient light data: Distinguish between indoor low light / strong light / outdoor natural light using an ambient light sensor;

[0184] Location and semantic information: By combining GPS / Wi-Fi positioning with map semantics, it identifies locations such as "home", "office", and "on the way to work";

[0185] Behavioral and physiological states: Identifying motion states such as resting / walking / driving through accelerometers; identifying resting / active states through heart rate;

[0186] Time and schedule information: current time, day type (weekday / restday).

[0187] Step 4.2: Build a cross-scenario intervention strategy library.

[0188] Pre-defined multi-dimensional intervention rules, with each intervention clearly labeled with "applicable offset direction" and "suitable scenario" attributes:

[0189] 1) Light exposure intervention.

[0190] Phase-delay type → Focus on "strong morning light exposure";

[0191] Phase advance type → Focus on "strong light exposure in the evening" and "strict light avoidance at night".

[0192] 2) Exercise and thermoregulation strategies.

[0193] Phase-delayed type → It is recommended to engage in moderate-intensity exercise 4-6 hours before the target bedtime;

[0194] For those with an advanced phase, it is recommended to engage in light activities in the evening to stabilize the rhythm.

[0195] 3) Diet and caffeine time window.

[0196] All offset types → Calculate personalized caffeine intake cutoff time based on offset direction and degree;

[0197] Establish a fixed eating window and avoid eating at night, which is subjectively related to your biological clock.

[0198] 4) Rhythm education information prompts.

[0199] Relevant scientific research knowledge cards are pushed based on the direction of the offset.

[0200] Step 4.3: Dynamically match and generate personalized intervention plans.

[0201] When a user meets the trigger conditions, the strategy library is cross-queried between the "offset profile" and the "real-time scenario" to generate a push card that includes immediate actions and follow-up strategies for today.

[0202] Example 1 (Office scenario - Phase delay type user, 10 AM):

[0203] The notification reads: "Your biological clock is a bit out of sync. Getting more sunlight today will help adjust it. Now you can: stand up and walk to the window to get 5-10 minutes of natural light. Today's suggestion: Stop consuming caffeine after 2 PM. Exercise reminder: Schedule 30 minutes of aerobic exercise after get off work; it will help you fall asleep more easily tonight."

[0204] Example 2 (Home relaxation scenario - same user, 10 PM):

[0205] The push notification reads: "Your melatonin secretion window is approaching, so it's time to give it a boost. What you can do now: Immediately dim the lights in your room and put your phone in night mode. Avoid: Checking work messages and staying away from any high-intensity screens. Recommendation: Take a warm bath to help cool your body and fall asleep."

[0206] Phase 5: Fully Closed-Loop Personalized Self-Calibration (Continuous Operation).

[0207] Step 5.1: Sleep latency feedback calibration.

[0208] The model parameters are automatically adjusted based on the user's actual sleep latency (time from lying down to falling asleep) as a feedback signal.

[0209] If a user's sleep latency remains too long even with a high sleep onset index (e.g., 80 points) for several consecutive days, the user's personal DLMO baseline will be automatically adjusted.

[0210] If the actual decay rate of a certain emotion type deviates systematically from the default half-life, the personalized half-life value will be automatically adjusted.

[0211] If the contribution of certain factors in the actual prediction does not match the model's preset weights, the factor weights will be automatically adjusted.

[0212] Step 5.2: Tracking the effectiveness of the intervention program.

[0213] The system continuously monitors the changes in the user's sleep latency and S_index offset after intervention. If the user's morning behavior has been continuously corrected but the sleep latency has not improved, the intervention engine automatically adjusts the intensity of subsequent interventions or changes the priority strategy type.

[0214] Step 5.3: Dynamic maintenance of the baseline.

[0215] The system uses a sliding window mechanism to continuously update the baseline parameters:

[0216] Daily routine regularity indicators and resting heart rate baseline: A sliding window using data from the most recent 7 days is updated daily to ensure that the baseline always reflects the user's latest physiological state;

[0217] DLMO baseline: After accumulating at least 7 days of valid sleep data, a weighted average is used for dynamic correction (with more recent data having higher weight);

[0218] S_index offset index and degree classification: automatically recalculated for each complete rest day cycle (i.e., after each rest day), capturing the trend of rhythm changes.

[0219] Step 5.4: Feedback signal quality filtering.

[0220] Before using sleep latency as a feedback signal, the system performs data quality checks:

[0221] (a) When a user actively takes medication, drinks alcohol, or travels across time zones, the data is marked as not participating in the model update;

[0222] (b) Outlier detection was performed on the sleep latency (data exceeding three times the standard deviation of an individual's history were removed) to prevent accidental events from contaminating the model parameters.

[0223] In summary, the embodiments of this application have the following advantages:

[0224] I. Comprehensive Coverage, No Blind Spots: For the first time, a complete closed loop of "daytime emotion accumulation → circadian rhythm diagnosis → pre-sleep tendency assessment → cross-scenario intervention execution → effect feedback calibration" is achieved in the same system, covering users' sleep health management around the clock.

[0225] Second, it is forward-looking and the intervention window has been significantly moved forward: it has changed from the traditional "telling you how you slept last night afterward" to "telling you during the day what consequences your current emotional state may bring tonight" and "telling you before bed how your current behavior is affecting melatonin secretion", expanding the intervention window from the single time before bed to the whole day.

[0226] Third, the causal logic is clear and the behavior guidance is strong: the "emotional stress pool" establishes an intuitive causal chain from daytime emotions to difficulty falling asleep at night, the "light inhibition accumulation-attenuation model" establishes a causal chain from behavior to melatonin secretion, and the "offset profile × real-time scene" establishes a causal chain from diagnosis to intervention. Each prompt tells the user "why" and "how to do it specifically".

[0227] IV. Highly Personalized and Adapted to Time and Place: It is the first to separate and then integrate "biological clock diagnosis" with "real-time execution scenario" to ensure that every suggestion appears at the right time in a currently executable manner; it innovatively designs completely opposite intervention logics for phase delay and phase advance respectively to achieve truly precise reverse adjustment.

[0228] VI. Pure software implementation with zero hardware cost: It fully utilizes the common sensors (accelerometers, heart rate monitors, ambient light sensors, etc.) and computing power of existing wearable devices and other equipment, and can achieve high-value functions through algorithm innovation alone, making it easy to deploy quickly and upgrade over-the-air (OTA).

[0229] Secondly, embodiments of this application provide a wearable device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the sleep health management method described in any embodiment. This device can execute the method provided in any embodiment of this application, possessing the corresponding functional modules and beneficial effects for executing the method, which will not be elaborated further here.

[0230] Thirdly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions thereon, which, when executed by a processor, implement the sleep health management method provided in all embodiments of this application.

[0231] Any combination of one or more computer-readable media may be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0232] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0233] The program code contained on a computer-readable medium may be transmitted using any suitable medium, including—but not limited to—wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0234] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0235] Finally, it should be noted that although the above embodiments have been described in the text and drawings of this application, this should not limit the scope of patent protection of this application. Any technical solutions that are based on the essential concept of this application and utilize the content described in the text and drawings of this application, resulting in equivalent structural or procedural substitutions or modifications, as well as the direct or indirect application of the technical solutions of the above embodiments to other related technical fields, are all included within the scope of patent protection of this application.

Claims

1. A sleep health management method, characterized in that, include: Real-time monitoring and identification of users' emotional events in a preset recent historical period, summing up the current residual stress value of each emotional event to obtain the cumulative stress score at the current moment; Based on the user's historical sleep data, the DLMO baseline point for the onset of melatonin secretion in dim light is estimated, and the pre-sleep assessment window is determined based on the DLMO baseline point. The sleep index is calculated based on the ambient light exposure data, user behavior data, and / or physiological stress data collected in real time during the pre-sleep assessment window, as well as the cumulative stress score; the sleep index is used to characterize the user's tendency to fall asleep at the current moment. Based on the sleep index, generate pre-sleep assessment information and / or corresponding intervention suggestions.

2. The sleep health management method according to claim 1, characterized in that, The real-time monitoring and identification of users' emotional events within a preset recent historical period, and the summing of the current residual stress values ​​of each emotional event to obtain the cumulative stress score at the current moment, includes: Real-time acquisition of the user's physiological and motion signals; When the physiological signal deviates from the preset personal resting baseline by more than a preset threshold, and the motion signal indicates that the user is in a non-motor state, a physiological arousal event is triggered and recorded; When the physiological arousal event is triggered, the multimodal physiological features at the current moment are extracted, and the multimodal physiological features are input into a pre-trained emotion classifier to identify the emotion type, arousal level and / or effectiveness value corresponding to the physiological arousal event. Based on the emotion type, the arousal level, and the effectiveness value, an initial impact value is assigned to the current physiological arousal event; According to the corresponding preset half-life, calculate the decay of the initial impact value of each physiological arousal event from the time of occurrence to the current time, and obtain the current pressure residual value of each physiological arousal event at the current time. The cumulative pressure score is obtained by summing the current residual pressure values ​​of all physiological arousal events that have been triggered within the preset recent historical period.

3. The sleep health management method according to claim 2, characterized in that, The physiological signals include: heart rate (HR), heart rate variability (HRV), and / or skin conductance response (EDA). The multimodal physiological characteristics include: HR rate of change, low-frequency to high-frequency power ratio of HRV and / or skin conductance level and its rate of change.

4. The sleep health management method according to claim 2, characterized in that, The step of estimating the DLMO baseline point for the onset of melatonin secretion in dim light based on the user's historical sleep data, and determining the pre-sleep assessment window using the DLMO baseline point as a benchmark, includes: Determine the habitual time to fall asleep based on the sleep time in the historical sleep data; The DLMO reference point is obtained by subtracting a preset time offset from the habitual fall asleep time; the preset time offset is within the range of [1.5h, 3h]. Using the DLMO reference point as the center, extend forward and backward by a first preset duration to obtain the pre-sleep assessment window period.

5. The sleep health management method according to claim 4, characterized in that, The sleep index is calculated based on real-time collected ambient light exposure data, user behavior data, and / or physiological stress data within the pre-sleep assessment window, as well as the cumulative stress score. Obtain the time offset of the current moment relative to the DLMO reference point, and determine the basic rhythm score based on the time offset; Based on the currently collected ambient light exposure data, user behavior data, physiological stress data, and cumulative stress score, light inhibition factor, behavioral regulation factor, physiological stress factor, and emotional stress factor are determined respectively. The sleep index is estimated based on the baseline rhythm score, the light inhibition factor, the behavioral regulation factor, the physiological stress factor, and the emotional stress factor.

6. The sleep health management method according to claim 5, characterized in that, The method for determining the basic rhythm score includes: obtaining a basic rhythm score function curve centered on the DLMO reference point and varying over time; obtaining the basic rhythm score corresponding to the current moment based on the function curve; wherein, the function curve includes a first interval and a second interval located before and after the DLMO reference point; the basic rhythm score corresponding to the second interval is higher than the basic rhythm score corresponding to the first interval; within the second interval, the basic rhythm score increases positively over time, and the steepness of the increase is positively correlated with the user's sleep-wake regularity index, which is determined based on the fluctuation of the user's historical sleep-on time and wake-up time; And / or, The method for determining the light inhibition factor includes: during light exposure, collecting illuminance values ​​and blue light percentages at preset time units; integrating the weighted values ​​of the illuminance values ​​and blue light percentages unit by unit to calculate the light inhibition amount; after entering a dark environment, allowing the light inhibition amount to decay exponentially at a corresponding preset time constant; and calculating the light inhibition factor based on the light inhibition amount at the current moment, wherein the light inhibition factor is negatively correlated with the light inhibition amount. And / or, The method for determining the behavior regulation factor includes: when it is detected that the user's exercise intensity exceeds a preset vigorous exercise threshold within a preset time period before the pre-sleep assessment window, the behavior regulation factor is set to a value less than 1.0, and the behavior regulation factor is negatively correlated with the exercise intensity; when it is detected that the user remains in a horizontal posture for more than a preset time period, the behavior regulation factor is multiplied by a positive additive coefficient greater than 1.

0. And / or, The method for determining the physiological stress factor includes: acquiring the user's short-term heart rate variability over a preset duration; comparing the short-term heart rate variability with the user's historical heart rate variability during the same period at rest; and setting the physiological stress factor to a value less than 1.0 when the short-term heart rate variability is lower than the historical heart rate variability during the same period by more than a preset standard deviation threshold. And / or, The method for determining the emotional stress factor includes: using a preset stress threshold as the center, mapping the cumulative stress score to an emotional stress factor between 0.5 and 1.0 through an S-shaped function; wherein, at the preset stress threshold, the rate of change of the emotional stress factor reaches its maximum value; as the cumulative stress score gradually deviates from the preset stress threshold, the rate of change of the emotional stress factor gradually decreases.

7. The sleep health management method according to claim 6, characterized in that, The step of generating pre-sleep assessment information and / or corresponding intervention suggestions based on the sleep index includes: The preset score range in which the sleep index falls is determined, the preset score range includes a first score range, a second score range, a third score range and a fourth score range with corresponding scores decreasing sequentially; If the sleep index is within the first score range, it is determined that all factors are within the preset ideal range, indicating that the user is currently in a good sleep state. If the sleep index is within the second score range, identify the main contributing factors that are causing the sleep index to decrease, and generate second assessment information and / or second intervention suggestions corresponding to the main contributing factors. If the sleep index is within the third score range, identify multiple factors that cause the current decrease in the sleep index to form a superimposed inhibition, and generate third assessment information and / or third intervention suggestions corresponding to the multiple factors; If the sleep index falls within the fourth score range, corresponding fourth assessment information and / or fourth intervention recommendations are generated.

8. The sleep health management method according to claim 1, characterized in that, The sleep health management method also includes: The direction of the user's circadian rhythm shift is determined based on the difference between the user's weekday sleep midpoint and rest day sleep midpoint. Based on the real-time perceived multimodal scene information of the user, the corresponding real-time scene label of the user is determined; the current multimodal scene information includes: ambient light data, location and semantic information, behavioral and physiological state and / or time and schedule information; Based on the rhythm offset direction and the real-time scene label, a target intervention strategy is matched from a preset cross-scene intervention strategy library and output; wherein, the cross-scene intervention strategy library contains multiple intervention rules, and each intervention rule is labeled with the applicable rhythm offset direction and the applicable scene label.

9. The sleep health management method according to claim 7, characterized in that, The sleep health management method also includes: Obtain the user's most recent actual sleep latency, which is the duration from the moment of lying down to the moment of falling asleep; based on the actual sleep latency, automatically calibrate the DLMO reference point, the preset half-life of the physiological arousal event, and / or the weights of each factor; And / or, The system continuously monitors the changes in the user's actual sleep latency and sleep midpoint shift index after intervention, and automatically adjusts the strategy type and / or intervention intensity of the intervention recommendations based on the changes in these trends. The sleep midpoint shift index is calculated by the difference between the user's weekday sleep midpoint and rest day sleep midpoint. And / or, Based on the most recently updated historical sleep data, the individual resting baseline, the DLMO baseline, and / or the sleep midpoint offset index are updated in a sliding window manner.

10. A wearable device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the sleep health management method as described in any one of claims 1-9.

11. A computer-readable storage medium having computer-executable instructions stored thereon, characterized in that, The computer-executable instructions are executed by a computer processor to implement the sleep health management method as described in any one of claims 1-9.

Images