Sleep monitoring data processing method and apparatus

Abstract

The application discloses a sleep monitoring data processing method and device. The method comprises the following steps: obtaining sleep monitoring data of a to-be-monitored object in a preset time period; the sleep monitoring data comprises a blood oxygen saturation time sequence; identifying a respiratory event in the blood oxygen saturation time sequence, and obtaining at least one respiratory event and a corresponding time window of the respiratory event; calculating an event-related hypoxia load in the preset time period according to blood oxygen saturation in the time window corresponding to each respiratory event; calculating an event duration and an event time distribution feature in the preset time period according to the time window corresponding to each respiratory event; and fusing the event-related hypoxia load, the event duration and the event time distribution feature to obtain a cardiovascular event load index of the to-be-monitored object. According to the embodiment of the application, the accuracy of the cardiovascular event load index can be improved.

Description

Technical Field

[0001] This application belongs to the field of sleep monitoring data processing technology, and in particular relates to a sleep monitoring data processing method and apparatus. Background Technology

[0002] Sleep-disordered breathing, especially obstructive sleep apnea, is a very common sleep-related respiratory disease. Its core pathological feature is the repeated collapse of the upper airway during sleep, leading to apnea or hypoventilation, which in turn triggers a series of pathophysiological changes such as intermittent hypoxia, dramatic fluctuations in intrathoracic pressure, and sleep structure disturbances. Numerous clinical studies have confirmed that the above pathophysiological processes are independent risk factors for various cardiovascular diseases, such as hypertension, coronary heart disease, heart failure, arrhythmia, and even sudden cardiac death. Therefore, in sleep monitoring, how to accurately quantify the cardiovascular risk of patients based on the collected physiological signals has become a key aspect of interdisciplinary research in sleep medicine and cardiovascular fields.

[0003] Currently, for cardiovascular risk assessment in patients with sleep-disordered breathing, polysomnography or portable sleep monitoring devices are commonly used in clinical practice to collect multi-channel physiological signals such as airflow, blood oxygen saturation, and chest and abdominal movements. Based on these signals, the apnea-hypopnea index is routinely calculated clinically, which is the total number of apnea and hypopnea events per hour during sleep, and is used as a core quantitative indicator to assess the severity of the disease. Meanwhile, some protocols further incorporate indicators such as the oxygen depletion index, attempting to supplement the description of the patient's hypoxia exposure level from a single dimension such as the depth of blood oxygen decline.

[0004] However, in clinical practice, there is often a lack of good correlation between respiratory event frequency indicators and cardiovascular event burden indicators. Some patients, although assessed as having mild or moderate sleep apnea, may actually have a much higher cardiovascular physiological burden than the indicators suggest, and vice versa. This assessment bias directly makes it difficult for clinicians to accurately stratify patients' cardiovascular risk based on existing indicators, thus affecting the reliability of subsequent intervention strategies and efficacy assessments. Therefore, how to calculate cardiovascular event burden indicators more accurately and comprehensively based on sleep monitoring data to truly reflect the cardiovascular physiological burden caused by sleep-disordered breathing has become an urgent technical problem to be solved. Summary of the Invention

[0005] This application provides a method and apparatus for processing sleep monitoring data, which can improve the accuracy of cardiovascular event load indicators.

[0006] On one hand, embodiments of this application provide a method for processing sleep monitoring data, the method comprising: Acquire sleep monitoring data of the subject to be monitored within a preset time period; the sleep monitoring data includes a time series of blood oxygen saturation. Identify respiratory events in a blood oxygen saturation time series and obtain at least one respiratory event and its corresponding time window; Based on the blood oxygen saturation within the time window corresponding to each respiratory event, calculate the event-related hypoxia load within the preset time period. Based on the time window corresponding to each respiratory event, calculate the event duration and event time distribution characteristics within the preset time period; By integrating event-related hypoxia load, event duration, and event time distribution characteristics, a cardiovascular event load index for the subjects to be monitored is obtained.

[0007] On the other hand, embodiments of this application provide a sleep monitoring data processing device, the device comprising: The data acquisition module is used to acquire sleep monitoring data of the subject to be monitored within a preset time period; the sleep monitoring data includes blood oxygen saturation time series. The event recognition module is used to identify respiratory events in the blood oxygen saturation time series and obtain at least one respiratory event and its corresponding time window. The hypoxia calculation module is used to calculate the event-related hypoxia load within a preset time period based on the blood oxygen saturation within the time window corresponding to each respiratory event. The event time module is used to calculate the event duration and event time distribution characteristics within a preset time period based on the time window corresponding to each respiratory event; The data fusion module is used to fuse event-related hypoxia load, event duration, and event time distribution characteristics to obtain cardiovascular event load indicators for the subjects to be monitored.

[0008] The sleep monitoring data processing method and apparatus of this application acquire sleep monitoring data including blood oxygen saturation time series, and identify various respiratory events and their corresponding time windows, thereby providing a data foundation for subsequent multidimensional feature extraction. Based on this, the method quantifies physiological load from three dimensions: First, it calculates event-related hypoxia load based on blood oxygen saturation within each respiratory event time window. This indicator is no longer limited to the lowest blood oxygen decrease or only counting the number of decreases, but also includes the magnitude and duration of blood oxygen decrease in each event, thus more comprehensively reflecting the cumulative effect of hypoxia on the body. Second, it calculates the event duration based on the respiratory event time window, incorporating the duration of respiratory events—a factor reflecting the degree of physiological stress—into the evaluation, reducing the load distortion caused by simply relying on event counts while ignoring duration differences. Third, it calculates the event time distribution characteristics, incorporating the temporal organization pattern of respiratory events throughout the night into the evaluation system. Finally, the method fuses the information from the above three dimensions to obtain a comprehensive cardiovascular event load index. This fusion mechanism enables the final output index to more comprehensively reflect the cardiovascular physiological burden caused by sleep-disordered breathing, thereby improving the accuracy of the cardiovascular event load index. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic flowchart of a sleep monitoring data processing method provided in one embodiment of this application; Figure 2 This is a flowchart illustrating a sleep monitoring data processing method provided in another embodiment of this application; Figure 3 This is a flowchart illustrating a sleep monitoring data processing method provided in another embodiment of this application; Figure 4 This is a flowchart illustrating a sleep monitoring data processing method provided in another embodiment of this application; Figure 5 This is a schematic diagram of the structure of a sleep monitoring data processing device provided in another embodiment of this application; Figure 6 This is a schematic diagram of the structure of an electronic device provided in another embodiment of this application. Detailed Implementation

[0011] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0012] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0013] It should be noted that the acquisition, storage, use, and processing of data in this application embodiment all comply with the relevant provisions of national laws and regulations.

[0014] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.

[0015] In practical applications, various indicators output by polysomnography or portable sleep monitoring devices are difficult to accurately reflect the cardiovascular event burden caused by sleep-disordered breathing in patients. This makes it difficult for clinicians to reliably stratify patients' cardiovascular risk based on existing indicators, which in turn affects the formulation of subsequent intervention decisions.

[0016] The problem stems from the lack of integrated assessment across multiple dimensions. Specifically, the apnea-hypopnea index, a core quantitative indicator, simply adds apnea and hypopnea events together, resulting in a uniform count. However, these two types of events differ fundamentally in the degree of airflow obstruction, leading to different physiological impacts on blood oxygen levels and the cardiovascular system. This simplistic equivalence prevents the event frequency index from accurately reflecting the differences in workload. Secondly, existing hypoxia-related indicators are all single-dimensional statistics. The lowest blood oxygen saturation only describes the extreme degree of blood oxygen decline, the oxygen depletion index only describes the frequency of blood oxygen decline, and the percentage of time blood oxygen is below a certain threshold only describes the duration of hypoxia. Each of these indicators represents a specific aspect of hypoxia exposure, lacking a comprehensive measurement method that simultaneously considers the depth, duration, and cumulative effects of hypoxia. Furthermore, the current assessment framework does not consider the temporal organization patterns of respiratory events throughout the night. Different distribution patterns result in different physiological stresses on the cardiovascular system; ignoring this dimension makes it impossible to effectively distinguish between patients with uniform and concentrated distribution patterns. Because the aforementioned multidimensional information lacks an integration mechanism in the existing indicator system, there is a significant discrepancy between the calculated cardiovascular event burden index and the actual physiological burden.

[0017] The sleep monitoring data processing method of this application acquires sleep monitoring data including blood oxygen saturation time series and identifies various respiratory events and their corresponding time windows, thus providing a data foundation for subsequent multidimensional feature extraction. Based on this, the method quantifies physiological load from three dimensions: First, it calculates event-related hypoxia load based on blood oxygen saturation within each respiratory event time window. This indicator is no longer limited to the lowest blood oxygen decrease or simply counting the number of decreases, but incorporates both the magnitude and duration of the blood oxygen decrease in each event, thus more comprehensively reflecting the cumulative impact of hypoxia on the body. Second, it calculates the event duration based on the respiratory event time window, incorporating the duration of respiratory events—a factor reflecting the degree of physiological stress—into the assessment, reducing the load distortion caused by simply relying on event counts while ignoring duration differences. Third, it calculates the event temporal distribution characteristics, incorporating the temporal organization pattern of respiratory events throughout the night into the assessment system. Finally, the method fuses the information from the above three dimensions to obtain a comprehensive cardiovascular event load index. This fusion mechanism enables the final output index to more comprehensively reflect the cardiovascular physiological burden caused by sleep-disordered breathing, thereby improving the accuracy of the cardiovascular event load index.

[0018] To address the problems of the prior art, embodiments of this application provide a method, apparatus, device, computer storage medium, and computer program product for processing sleep monitoring data. The sleep monitoring data processing method provided in this application embodiment will be described first below.

[0019] Figure 1A schematic flowchart of a sleep monitoring data processing method according to an embodiment of this application is shown. Figure 1 As shown, the method includes the following steps S101-S105.

[0020] S101, acquire sleep monitoring data of the subject to be monitored within a preset time period; the sleep monitoring data includes blood oxygen saturation time series.

[0021] As an example, the preset time period can refer to the total sleep time recorded during overnight sleep monitoring of the subject, typically measured in hours. Sleep monitoring data serves as the foundational information source for all subsequent analyses and calculations. This data must at least include a blood oxygen saturation time series, which is a continuous sequence of blood oxygen saturation values ​​recorded over time during sleep using pulse oximetry. This sequence can be recorded at equal intervals, typically generating one blood oxygen saturation reading per second or every few seconds, reflecting the dynamic changes in blood oxygen levels throughout the night. In addition to the blood oxygen saturation time series, sleep monitoring data may also include other physiological signals during sleep, such as airflow signals, chest and abdominal movement signals, respiratory effort signals, body position signals, sleep stage information, wakefulness information, or heart rate signals. These auxiliary signals can play a supplementary role in subsequent analyses.

[0022] Specifically, as an example, raw physiological signal records of the subject within a preset time period can be obtained through polysomnography, portable sleep monitoring devices, wearable sleep monitoring devices, bedside monitoring devices, or cloud-based sleep monitoring platforms. These raw records are then preliminarily organized and extracted to form a structured sleep monitoring dataset. This step provides the entire evaluation process with collected and preliminarily organized raw physiological data, ensuring that subsequent signal preprocessing and respiratory event recognition steps have complete data input.

[0023] S102, identify respiratory events in the blood oxygen saturation time series, and obtain at least one respiratory event and its corresponding time window.

[0024] As an example, a respiratory event can refer to an apnea or hypopnea event that occurs during sleep in the subject being monitored. Apnea can refer to a state where airflow stops or nearly stops, while hypopnea can refer to a state where airflow is significantly reduced but not completely stopped. Based on the differences in their mechanisms, respiratory events can be further subdivided into obstructive apnea, central apnea, mixed apnea, and multiple grades of hypopnea events. The grade of a hypopnea event can be determined by one or more of the following: the magnitude of the decrease in airflow, the magnitude of the decrease in blood oxygen saturation, the duration of the event, and whether it is accompanied by awakening. A time window can refer to the time interval encompassed by each respiratory event from its start to its end, with a clearly defined start and end time point; its length is the duration of the respiratory event.

[0025] Specifically, as an example, this step can be implemented using manual annotation or an automatic recognition algorithm. In manual annotation, a sleep technician annotates respiratory events segment by segment based on airflow signals, changes in blood oxygen saturation, and chest and abdominal movement signals from polysomnography recordings, following standard interpretation rules and recording their start and end times. In automatic recognition, rule-based or model-based algorithms can be used. Rule-based algorithms identify apnea and hypoventilation events based on the magnitude and duration of airflow decline, the magnitude of blood oxygen decline, changes in chest and abdominal movement, or arousal information, and output the event type, start and end times, and duration. Model-based algorithms use machine learning or deep learning models, taking airflow signals, blood oxygen saturation sequences, chest and abdominal movement signals, snoring signals, body position signals, or heart rate signals as input, to automatically segment, detect, and classify respiratory events during sleep. The automatic recognition results can be used in conjunction with the manual annotation results and, after quality control, are incorporated into the subsequent comprehensive event load calculation process.

[0026] This step extracts clinically significant event location and type information from the original blood oxygen saturation time series to form a structured respiratory event sequence, providing basic data for subsequent calculations of event-related hypoxia load, event duration, and event temporal distribution characteristics.

[0027] S103, calculate the event-related hypoxia load within a preset time period based on the blood oxygen saturation within the time window corresponding to each respiratory event.

[0028] As an example, event-related hypoxia load can refer to the sum of the areas of the blood oxygen drop curves caused by all respiratory events within a preset time period, and then the value obtained after standardization according to the total sleep time of the preset time period. Its unit can be the area of ​​drop per minute, which is used to measure the cumulative degree of hypoxia exposure per unit sleep time.

[0029] Specifically, as an example, for each identified respiratory event, the blood oxygen saturation level over a period prior to the event is first determined as a reference baseline. This baseline can be the mean or median of blood oxygen saturation within a fixed time window before the event. Then, within the time window corresponding to the current respiratory event, the difference between the actual blood oxygen saturation and the reference baseline is calculated point by point. These differences are then integrated or summed within the time window to obtain the area under the blood oxygen saturation decline curve corresponding to the respiratory event. This area includes both the magnitude and duration of the decline. The areas under the decline curves for all respiratory events are summed to obtain the total area of ​​decline for all events. This total area is then divided by the total sleep time over a preset period to obtain the event-related hypoxia load. This step transforms the scattered hypoxia information in traditional assessments into a unified cumulative metric, making the hypoxia exposure levels comparable between different patients and between different times for the same patient.

[0030] S104, calculate the duration of events and the time distribution characteristics of events within a preset time period based on the time window corresponding to each respiratory event.

[0031] As an example, event duration refers to the weighted sum of the durations of respiratory events within a unit of sleep time, reflecting the degree of physiological stress on the cardiovascular system caused by respiratory events in the time dimension. Event temporal distribution characteristics refer to the degree of dispersion or concentration reflected in the arrangement of all respiratory events on the nighttime sleep time axis, used to distinguish between two different patterns: events evenly distributed throughout the night and events concentrated in local time periods.

[0032] Specifically, as an example, the start and end points of the time window for each respiratory event are extracted. The duration of the event is obtained by subtracting the start point from the end point. Then, the duration of all respiratory events is standardized according to the duration of a preset time period to obtain the average or weighted duration per unit sleep time. The calculation method for the event temporal distribution characteristics is as follows: all respiratory events are sorted from morning to night according to the start time of each respiratory event to form a respiratory event sequence. Then, the time interval between each two adjacent respiratory events in the sequence is calculated, and the dispersion or variability of all time intervals is statistically analyzed. The magnitude of the dispersion directly reflects the distribution state of the events. A smaller dispersion indicates that the events are more evenly distributed throughout the night, while a larger dispersion indicates that the events occur concentratedly in some time periods and sparsely in other time periods. This step extracts information from two dimensions: event duration and event temporal organization pattern, which makes up for the shortcomings of simply relying on event counts and hypoxia amplitude, which cannot reflect the differences in duration and distribution patterns.

[0033] S105 integrates event-related hypoxia load, event duration, and event time distribution characteristics to obtain the cardiovascular event load index of the subjects to be monitored.

[0034] As an example, fusion refers to integrating the above three dimensions of indicators through specific mathematical operations to form a single numerical value that comprehensively reflects information from multiple dimensions. The cardiovascular event burden index is the final comprehensive quantitative result, and the value of this index directly corresponds to the magnitude of the cardiovascular physiological burden related to sleep-disordered breathing in the monitored subject.

[0035] Specifically, as an example, the three indicators of event-related hypoxia load, event duration, and event temporal distribution characteristics can be normalized separately. Normalization aims to eliminate incomparability caused by differences in dimensions and numerical ranges between different indicators, ensuring that the three indicators are within a unified numerical scale. Then, the three normalized indicators are combined using either multiplicative fusion or weighted summation fusion. Multiplicative fusion multiplies the three normalized values ​​to obtain a comprehensive indicator; weighted summation fusion assigns a weight coefficient to each dimension, multiplies the three values ​​by their respective weights, and then adds them together to obtain the comprehensive indicator. When using multiplicative fusion, a lower value in any dimension will lower the final comprehensive indicator value, reflecting the synergistic effect of each dimension on load assessment; when using weighted summation, the contribution of each dimension can be flexibly adjusted by the weight coefficients. This step integrates information originally scattered across three independent dimensions into a single intuitive comprehensive indicator, allowing clinical users to obtain an overall assessment of the patient's cardiovascular load without having to interpret multiple separate indicators.

[0036] The sleep monitoring data processing method of this application acquires sleep monitoring data including blood oxygen saturation time series and identifies various respiratory events and their corresponding time windows, thus providing a data foundation for subsequent multidimensional feature extraction. Based on this, the method quantifies physiological load from three dimensions: First, it calculates event-related hypoxia load based on blood oxygen saturation within each respiratory event time window. This indicator is no longer limited to the lowest blood oxygen decrease or simply counting the number of decreases, but incorporates both the magnitude and duration of the blood oxygen decrease in each event, thus more comprehensively reflecting the cumulative impact of hypoxia on the body. Second, it calculates the event duration based on the respiratory event time window, incorporating the duration of respiratory events—a factor reflecting the degree of physiological stress—into the assessment, reducing the load distortion caused by simply relying on event counts while ignoring duration differences. Third, it calculates the event temporal distribution characteristics, incorporating the temporal organization pattern of respiratory events throughout the night into the assessment system. Finally, the method fuses the information from the above three dimensions to obtain a comprehensive cardiovascular event load index. This fusion mechanism enables the final output index to more comprehensively reflect the cardiovascular physiological burden caused by sleep-disordered breathing, thereby improving the accuracy of the cardiovascular event load index.

[0037] As another implementation of this application, such as Figure 2 As shown, S102 may also include the following steps S1021-S1023.

[0038] S1021, For each first respiratory event, determine the reference value of blood oxygen saturation for the first respiratory event based on the blood oxygen saturation before the first respiratory event; the first respiratory event is any respiratory event corresponding to the blood oxygen saturation time series.

[0039] As an example, the blood oxygen saturation reference value is a baseline blood oxygen level used to measure the degree of blood oxygen decline caused by the current respiratory event. This reference value represents the normal blood oxygen level that the subject under monitoring should have had before the current respiratory event occurred. Establishing this baseline allows for a more accurate calculation of the magnitude of the blood oxygen decline caused by the current event.

[0040] Specifically, as an example, for each identified respiratory event, a period of blood oxygen saturation data prior to its start time is selected as a reference window. The blood oxygen saturation values ​​within this reference window are then statistically analyzed, and the statistical value is used as the reference value for blood oxygen saturation in that respiratory event. The length of the reference window can be adaptively adjusted according to the interval between adjacent events. When the event interval is long, the mean or median blood oxygen saturation within a fixed time window before the event can be used as the reference value. When the event interval is short or blood oxygen has not fully recovered after the previous event, the maximum blood oxygen saturation value within the window before the event can be used as the reference value. The blood oxygen saturation reference value can also be calculated within a fixed time window using the mean, median, maximum value, or higher percentile, etc. The method can be flexibly selected based on the actual signal quality.

[0041] This step establishes an individualized blood oxygenation baseline level for each respiratory event, enabling the subsequent calculation of the area of ​​decrease to accurately reflect the degree of blood oxygenation deviation caused by the event itself, and reducing calculation biases introduced by insufficient blood oxygenation recovery between events.

[0042] S1022, within the time window corresponding to the first respiratory event, calculate the area under the curve of the decrease in blood oxygen saturation relative to the blood oxygen saturation reference value.

[0043] As an example, the area under the curve (AUC) can refer to the area enclosed by the gradual decrease in actual blood oxygen saturation relative to the reference blood oxygen saturation value within the time window corresponding to a respiratory event. This area contains information in two dimensions: the depth of the blood oxygen decrease and the duration of the blood oxygen decrease, reflecting the total hypoxic exposure caused by the respiratory event.

[0044] Specifically, as an example, from the start to the end of the current respiratory event, the difference between the reference value and the actual oxygen saturation reading is calculated point by point within the time window according to the sampling points of the oxygen saturation time series. When the actual oxygen saturation is lower than the reference value, the difference is positive, representing the magnitude of the decrease in oxygen saturation. Multiplying this difference by the duration of the difference yields a component of the area of ​​decrease within that time interval. In the case of continuous sampling, integrating or summing the decrease differences at all sampling points over time intervals yields the total area of ​​decrease within the time window corresponding to the respiratory event.

[0045] This step quantifies the magnitude and duration of the blood oxygen drop caused by a single respiratory event into a single area value, reducing the information loss caused by using the lowest blood oxygen value alone, which only reflects the instantaneous extreme value, or using the duration alone, which only reflects the duration and ignores the depth.

[0046] S1023. Based on the total area of ​​the decline curves corresponding to all respiratory events and the duration of the preset time period, determine the area of ​​the decline curve of blood oxygen saturation per unit sleep time to obtain the event-related hypoxia load.

[0047] As an example, event-related hypoxia load can be defined as the average area of ​​hypoxia per unit sleep time obtained by summing the areas of hypoxia corresponding to all respiratory events and dividing by the total sleep duration of a preset time period. A higher value indicates a heavier burden of blood oxygen saturation experienced by the monitored subject per unit sleep time due to respiratory events.

[0048] Specifically, as an example, the areas under the curves calculated for each respiratory event can be summed to obtain the total area under the curves corresponding to all respiratory events throughout the night. This total area is then divided by the total sleep time over a preset period, expressed in hours or minutes, to obtain the area under the curve per unit sleep time. This standardized process eliminates the impact of varying total sleep durations on the total hypoxia load. This step aggregates the scattered hypoxia accumulation information throughout the night into a standardized comprehensive metric, providing a unified input for the subsequent fusion of multidimensional indicators.

[0049] The sleep monitoring data processing method of this application determines an independent blood oxygen saturation reference value for each respiratory event, ensuring that the calculation of the decrease in blood oxygen saturation for each event is based on the individual's blood oxygen level before the event, thus reducing interference between different events due to varying degrees of blood oxygen recovery. Furthermore, the decrease in blood oxygen saturation and its duration within each event's time window are quantified into a single area value. This area reflects both the depth of blood oxygen saturation decrease and the duration of hypoxia in a single event, providing more information than using only the lowest blood oxygen value or the duration of the event alone. Finally, the areas of decrease curves for all events are summed and divided by the total sleep time, resulting in a decrease area per unit sleep time that reduces the impact of differences in total sleep duration. These steps, considering both the characteristics of hypoxia in a single event and the overall cumulative burden of hypoxia load, ensure that the final calculated event-related hypoxia load accurately reflects the total blood oxygen saturation burden borne by the monitored subject due to respiratory events within a unit sleep time.

[0050] As another implementation of this application, S1021 may also include the following steps S10211-S10213.

[0051] S10211, calculate the time interval between the end time of the second respiratory event and the start time of the first respiratory event; the second respiratory event is the preceding respiratory event of the first respiratory event.

[0052] As an example, the time interval can refer to the time distance between the first respiratory event for which the blood oxygen saturation reference value needs to be calculated and the most recent adjacent respiratory event that occurred before it, i.e., the second respiratory event. Specifically, the start point of this time interval is the end time of the second respiratory event, the end point is the start time of the first respiratory event, and the duration between the two points is the numerical value of the time interval.

[0053] Specifically, as an example, for a sequence of identified and chronologically ordered respiratory events, the first respiratory event requiring processing is located, and its start time is read from its event record. Then, one event is traced back to its preceding adjacent respiratory event, the second respiratory event, and its end time is read from that record. The start time of the first respiratory event is subtracted from the end time of the second respiratory event, yielding the difference between the two times; this difference is the time interval. This step provides the basic data for subsequent determination of whether the previous respiratory event still affects the blood oxygen level of the current event.

[0054] S10212, based on the blood oxygen saturation reference value of the second respiratory event, calculate the degree of recovery of blood oxygen saturation within the time interval.

[0055] As an example, the degree of recovery can refer to the relative extent to which blood oxygen saturation recovers from the decline caused by the previous event to near normal levels within the time interval following the end of the second respiratory event. This indicator is used to quantitatively assess the extent to which the blood oxygen level of the monitored subject has recovered during the period between the end of the second respiratory event and the beginning of the first respiratory event.

[0056] Specifically, as an example, a reference value for blood oxygen saturation is obtained for the second respiratory event. This reference value represents the normal blood oxygen level of the subject before the second respiratory event, which is the theoretically target level to which blood oxygen should recover. Then, the actual blood oxygen saturation value reached within this time interval is determined. This value can be the actual blood oxygen saturation reading at the end of the time interval or within the time interval, or the maximum blood oxygen saturation value within the time interval. The ratio or difference between the actual blood oxygen saturation value and the reference value for blood oxygen saturation in the second respiratory event is used as a quantitative result of the degree of recovery. The closer the actual blood oxygen level is to the reference value, the higher the degree of recovery; conversely, the farther the actual blood oxygen level is from the reference value, the lower the degree of recovery. This step provides a quantitative basis for determining whether blood oxygen has fully recovered from the impact of the previous event. This degree of recovery, together with the time interval, constitutes the two core conditions for determining how to subsequently determine the reference value for blood oxygen saturation in the current event.

[0057] S10213, Determine the reference value of blood oxygen saturation corresponding to the first respiratory event based on the time interval and the degree of recovery.

[0058] Specifically, as an example, by comprehensively considering the time interval and the degree of recovery, the actual state of blood oxygen levels before the first respiratory event is comprehensively judged, and the most suitable reference value determination method for the current scenario is selected accordingly. When the time interval is long and the degree of recovery is high, it indicates that the impact of the previous respiratory event has largely subsided, and the blood oxygen level before the first respiratory event is in a stable state. In this case, the mean or median blood oxygen saturation within a stable window before the event can be used as the reference value. When the time interval is short or the degree of recovery is low, it indicates that the impact of the previous respiratory event has not completely subsided, and the blood oxygen level before the event is still suppressed. In this case, using the mean or median will lead to an underestimation of the reference value. Therefore, it is necessary to use the maximum blood oxygen saturation value or a higher percentile value within the window before the event as the reference value to reduce the underestimation caused by the previous event. In addition, the level that blood oxygen may reach when it is fully recovered can be estimated by using trend extrapolation based on the length of the time interval and the degree of recovery as a reference value.

[0059] This step adaptively selects the most appropriate reference value determination method based on the actual interval between two events and the blood oxygen recovery status, so that the blood oxygen saturation reference value for each respiratory event can truly reflect the baseline blood oxygen level that should have been present before the event, thereby ensuring the accuracy of the subsequent calculation of the area under the decline curve.

[0060] The sleep monitoring data processing method in this application uses a time interval to reflect the length of time available for blood oxygen recovery after the previous event, and the degree of recovery directly quantifies the actual recovery of blood oxygen within that time interval. Both assess whether the impact of the previous event on the current blood oxygen level has subsided from both time and effect perspectives. The above steps achieve adaptive adjustment of blood oxygen reference values ​​in scenarios of continuous or intense respiratory events, ensuring that the blood oxygen saturation reference value for each event accurately reflects the baseline blood oxygen level that should have occurred before the event, thereby guaranteeing the accuracy and stability of hypoxic area calculation in scenarios of intense events.

[0061] As another implementation of this application, S10213 may also include the following steps S102131-S102132.

[0062] S102131, when the time interval is greater than the preset time interval and the recovery degree is greater than the preset recovery threshold, the blood oxygen saturation characteristic value within the preset duration before the start time of the first respiratory event is used as the blood oxygen protection reference value of the first respiratory event; the blood oxygen saturation characteristic value includes the mean or median of blood oxygen saturation within the preset duration before the start time of the second respiratory event.

[0063] As an example, the preset time interval can be a pre-defined time length threshold used to determine whether the interval between two adjacent respiratory events is long enough. The preset recovery threshold can be a pre-defined recovery threshold used to determine whether blood oxygen saturation has recovered to near-normal levels after the previous event. Blood oxygen saturation characteristic values ​​are representative values ​​extracted from a set of blood oxygen saturation data; in this step, the mean or median can be used.

[0064] Specifically, as an example, when the time interval between the end of the second respiratory event and the start of the first respiratory event is greater than a preset time interval, and the recovery of blood oxygen saturation within this time interval is greater than a preset recovery threshold, it indicates that the hypobaric effect of the previous respiratory event has subsided, and the blood oxygen before the current first respiratory event is in a stable and undisturbed state. In this case, multiple blood oxygen saturation readings within a preset time period before the start of the first respiratory event are selected, and the mean or median of these readings is calculated. This value is used as the reference value for blood oxygen saturation during the first respiratory event. Using the mean or median can reduce the impact of individual abnormal readings on the reference value, obtaining a stable value that represents the concentrated level of blood oxygen during this period. When the two respiratory events are sufficiently far apart and blood oxygen has fully recovered, the statistic reflecting the concentration trend is used as the blood oxygen benchmark, thus providing a stable and reliable blood oxygen reference starting point for subsequent calculation of the area of ​​decline.

[0065] S102132, when the time interval is less than or equal to the preset time interval or the recovery degree is less than or equal to the preset recovery threshold, the maximum blood oxygen saturation value within the preset time before the start of the first respiratory event is used as the blood oxygen protection reference value of the first respiratory event.

[0066] As an example, the preset duration refers to a fixed time interval before the start of the first respiratory event. This window is used to search for the maximum value of blood oxygen saturation or to obtain blood oxygen saturation characteristic values ​​within this interval.

[0067] Specifically, as an example, when the time interval between the end of the second respiratory event and the start of the first respiratory event is less than or equal to a preset time interval, or when the recovery of blood oxygen saturation within that time interval is less than or equal to a preset recovery threshold, it indicates that the hypoxic effect of the previous respiratory event has not completely subsided, and the blood oxygen level before the start of the current first respiratory event is still suppressed due to the influence of the previous event. In this case, if the mean or median is used, the resulting value will be lower than the true baseline blood oxygen level, thus underestimating the area of ​​decline in the first respiratory event and failing to accurately reflect the degree of blood oxygen decline in that event. Therefore, the maximum value of blood oxygen saturation within a preset time window before the start of the first respiratory event is selected as the reference value. This maximum value represents the blood oxygen level at the point within that window where the influence of the previous event is least pronounced, and can more accurately reflect the blood oxygen baseline that should have existed before the current event.

[0068] When the interval between two respiratory events is short or the blood oxygen recovery after the previous event is insufficient, the maximum value is used to reduce the interference of the depression effect of the previous event on the reference value, so as to ensure that the area of ​​blood oxygen drop in each event can be accurately calculated in the scenario of dense respiratory events.

[0069] The sleep monitoring data processing method of this application uses the mean or median as a reference value when the interval between adjacent events is long enough and blood oxygenation is fully restored. This method can obtain a stable and reliable blood oxygenation benchmark, unaffected by single-point abnormal readings. When the interval between adjacent events is short or blood oxygenation is insufficient, the maximum value is used as the reference value. This method can reduce the interference of the previous event on the reference value and reduce the underestimation of the hypoxia area due to the underestimation of the reference value. The above mechanism ensures that, under the two different respiratory event arrangement patterns of sparse and dense event intervals, the blood oxygen saturation reference value of each respiratory event can truly reflect the baseline blood oxygen level that should have occurred before the event, thereby ensuring the consistency and accuracy of hypoxia area calculation throughout the sleep recording period.

[0070] As another implementation of this application, such as Figure 3 As shown, S104 may also include the following steps S104A1-S104A3.

[0071] S104A1, determine the weight of each respiratory event according to its type.

[0072] As an example, the weights of each respiratory event are used to quantify the differential contribution of different types of respiratory events to cardiovascular physiological load.

[0073] Specifically, as an example, a base weight value is assigned to each type of respiratory event. Apnea events, due to complete or near-complete airflow blockage, have a more severe impact on blood oxygen levels and the cardiovascular system, and therefore are assigned a higher weight value; hypoventilation events, with only partial airflow reduction, have a relatively milder physiological impact, and therefore are assigned a lower weight value. This step ensures that the contribution of different types of respiratory events to the final load index matches their actual physiological impact, reducing load distortion caused by treating different types of events equally.

[0074] S104A2, based on the weight of each respiratory event, the duration of each respiratory event is weighted and summed to obtain the weighted duration of the respiratory event.

[0075] As an example, the weighted duration of respiratory events can be calculated by multiplying the actual duration of each respiratory event during the night's sleep by its corresponding weighting factor, and then summing the results for all events. This summation reflects information on three aspects: the number of events, the weight differences among various types of events, and the duration of each event.

[0076] Specifically, as an example, for each identified breathing event, the event type and its corresponding weight coefficient are obtained, along with the event duration, which is equal to the end time of the event's time window minus the start time. The weight coefficient is multiplied by the duration to obtain a weighted duration value for the event. A value greater than the actual duration indicates that the event type is assigned a higher weight, while a value less than the actual duration indicates that the event type is assigned a lower weight. Then, the weighted duration values ​​of all events are summed to obtain the total weighted duration for the entire night. This step integrates information from the three dimensions of the number of breathing events, type differences, and duration into a summary value, serving as the basis for the final event duration calculation.

[0077] S104A3 determines the weighted duration within a unit of sleep time based on the weighted duration of the respiratory event and the duration of a preset time period, thus obtaining the event duration.

[0078] As an example, event duration can refer to the average weighted duration of respiratory events per unit of sleep time.

[0079] Specifically, as an example, the calculated sum of weighted durations of respiratory events is divided by the total sleep duration over a preset time period. Total sleep duration refers to the total time the subject is actually asleep throughout the night, typically measured in hours. Dividing the two yields the weighted duration per unit of sleep time. This step transforms the summed weighted duration into a standardized metric, providing a unified input for the subsequent fusion of multidimensional indicators.

[0080] The sleep monitoring data processing method of this application assigns different weights to different types of respiratory events, allowing apnea and hypopnea, two fundamentally different events, to make differentiated contributions to the final workload index, thus reducing the workload distortion caused by traditional methods that count all events equally. Based on the determined weights, the weight coefficient of each event is multiplied by its actual duration and summed. This process incorporates information from three aspects: the total number of events, the weight differences between different types of events, and the duration of each event, providing richer physiological workload information than simple event counting. Finally, the summed weighted duration is divided by the total sleep time, making the final event duration a standardized metric. These steps ensure that the final event duration accurately reflects the comprehensive physiological burden borne by the monitored subject per unit sleep time due to different types and durations of respiratory events.

[0081] As another implementation of this application, the sleep monitoring data also includes sleep state parameters, and S104A1 further includes the following steps S104A11-S104A13.

[0082] S104A11 assigns different base weights to different types of respiratory events based on their type.

[0083] As an example, the base weight can refer to a benchmark weight coefficient that is pre-set according to the physiological impact level corresponding to different respiratory event types.

[0084] Specifically, as an example, a correspondence table between respiratory event types and baseline weights can be established. Since apnea events involve complete or near-complete airflow blockage, the drop in blood oxygen saturation is usually greater, resulting in a more severe instantaneous impact on the cardiovascular system; therefore, apnea events are assigned higher baseline weight values. Hypoventilation events involve partial airflow reduction rather than complete blockage, with a relatively smaller drop in blood oxygen saturation; therefore, hypoventilation events are assigned lower baseline weight values. Within the subcategories of apnea, obstructive apnea, accompanied by sustained respiratory effort and dramatic fluctuations in intrathoracic pressure, can have a higher baseline weight than central apnea. Within the subcategories of hypoventilation, hypoventilation events with a greater drop in airflow can be assigned relatively higher baseline weights.

[0085] This step establishes the preliminary contribution of different respiratory events to physiological load from the perspective of event type, providing a basis for subsequent personalized weighting adjustments based on individual characteristics.

[0086] S104A12, determine the feature correction factor based on the preset risk coefficients corresponding to the sleep state parameters and the physiological characteristic parameters of the subject to be monitored.

[0087] As an example, sleep state parameters can refer to the specific physiological state information at the time of a respiratory event, including the sleep stage, body position, chest and abdominal movement amplitude, respiratory effort, and whether the event was accompanied by awakening. Physiological characteristic parameters can refer to the basic information and health status of the subject, including age, sex, body mass index, history of cardiovascular disease, hypertension, diabetes, heart failure, arrhythmia, and medication use. Preset risk coefficients are numerical coefficients pre-set for each sleep state parameter and each physiological characteristic parameter value; these coefficients reflect the degree of influence of that parameter on cardiovascular risk. Characteristic correction factors are single numerical adjustment coefficients obtained by combining multiple preset risk coefficients.

[0088] Specifically, as an example, for a respiratory event currently being processed, information such as sleep stage, body position, chest and abdominal movement amplitude, respiratory effort, and whether awakening was present at the time of the event is obtained. The risk coefficient value corresponding to each state parameter is then found by referring to a pre-set risk coefficient table. Simultaneously, information such as the subject's age, gender, body mass index, history of cardiovascular disease, hypertension, diabetes, heart failure, arrhythmia, and medication status is obtained, and the risk coefficient value corresponding to each physiological characteristic parameter is found by referring to a pre-set risk coefficient table. All the obtained risk coefficient values ​​are then multiplied or weighted and summed to form a single value, which is the feature correction factor.

[0089] S104A13 determines the weight of each respiratory event by correcting the base weights through a feature correction factor.

[0090] This step combines the base weight with the feature correction factor to generate a final weight value for each respiratory event that reflects both the inherent physiological impact of the event type and the individual characteristics of the monitored subject and the specific physiological state at the time of the event.

[0091] Specifically, as an example, the base weight corresponding to the event type is multiplied by the calculated feature correction factor. The product of the base weight and the feature correction factor is the final weight of the respiratory event. When the feature correction factor is greater than 1, the same type of event in the monitored subject or physiological state is assigned a weight higher than the base level; when the feature correction factor is less than 1, it is assigned a weight lower than the base level; when the feature correction factor is equal to 1, the base weight remains unchanged. In this way, each respiratory event obtains a personalized adjusted weight value, which reflects the comprehensive influence of three levels: event type, individual characteristics of the monitored subject, and physiological state at the time of the event. The final weight is then used for a weighted calculation of the event's duration.

[0092] The sleep monitoring data processing method of this application sets differentiated base weights according to different respiratory event types, so that apnea and hypopnea, two fundamentally different events, have different contribution benchmarks in subsequent load calculations. Based on this, preset risk coefficients corresponding to sleep state parameters and physiological characteristic parameters are introduced, and information such as the specific sleep stage, body position, respiratory effort, age, and underlying medical history of the monitored subject at the time of the event is integrated into a feature correction factor. Finally, the base weights are individually adjusted using this correction factor, so that the final weight of each respiratory event not only reflects the inherent physiological impact of the event type but also reflects the influence of the individual characteristics of the monitored subject and the specific physiological context at the time of the event, thereby achieving a more accurate quantification of event load than the fixed-weight method.

[0093] As another implementation of this application, such as Figure 4 As shown, S104 may also include the following steps S104B1-S104B3.

[0094] S104B1, according to the start time of each respiratory event, sort all respiratory events to form a respiratory event sequence.

[0095] As an example, a respiratory event sequence can refer to an ordered list formed by arranging all respiratory events identified throughout the night in chronological order of their occurrence. This sequence reflects the complete sequence in which respiratory events occur throughout the sleep period.

[0096] Specifically, as an example, the start time of each breathing event is obtained. This start time is the initial point of the breathing event's time window, accurate to the second or even smaller. All breathing events are then sorted in ascending order of their start times, from earliest to latest. This sorted order represents the actual chronological order of the breathing events. After sorting, each breathing event has a unique position number in the sequence, and adjacent events are two breathing events that occur sequentially. This step organizes the originally scattered breathing events into an ordered structure along the time dimension, establishing the chronological relationship between events and laying the data structure foundation for subsequent calculations of the time interval between adjacent events and the extraction of time distribution features.

[0097] S104B2 calculates the time interval between any two adjacent respiratory events in a respiratory event sequence.

[0098] As an example, a time interval can refer to the temporal distance between any two adjacent respiratory events in a respiratory event sequence, that is, the duration between the end of the previous respiratory event and the start of the next respiratory event. This interval reflects the temporal proximity between two respiratory events.

[0099] Specifically, as an example, the respiratory event sequence can be traversed, starting from the first event in the sequence, and then processing the current event and its next adjacent event in sequence. For each pair of adjacent events, the end time of the preceding event and the start time of the following event are obtained. The start time of the following event is subtracted from the end time of the preceding event, and the difference is the time interval between the pair of adjacent events. This time interval can be in seconds or minutes. The above calculation is repeated for all pairs of adjacent events in the sequence, resulting in a set containing multiple time interval values. The length of this set is equal to the total number of respiratory events minus one. This step transforms the respiratory event sequence into a time interval sequence, quantifying the temporal density relationship between adjacent events in the form of interval values, providing basic data for subsequent calculations of dispersion.

[0100] S104B3 calculates the event time distribution characteristics based on the dispersion of all time intervals.

[0101] As an example, dispersion can refer to the degree of difference between the values ​​of all time intervals. When the values ​​of each time interval are close to each other, the dispersion is small, indicating that respiratory events are relatively evenly distributed throughout the night's sleep; when the values ​​of each time interval differ significantly, the dispersion is large, indicating that respiratory events occur densely in some periods and sparsely in others. The event temporal distribution characteristics are numerical indicators obtained by quantifying this dispersion.

[0102] Specifically, as an example, all the calculated time interval values ​​are treated as a dataset, and the coefficient of variation (COP) of this dataset is calculated. The COP equals the standard deviation of the time interval values ​​divided by the mean of the time interval values. This coefficient reduces the influence of the absolute value of the time interval on the judgment of dispersion. A larger COP indicates a greater dispersion of the time intervals, a more uneven distribution of respiratory events throughout the night, and obvious clustering; a smaller COP indicates a smaller dispersion of the time intervals, and a more uniform distribution of respiratory events throughout the night. This step quantifies the temporal organization pattern of respiratory events into a numerical index, enabling effective numerical differentiation between uniform and clustered distributions, providing quantitative input in the temporal dimension for subsequent multidimensional fusion.

[0103] The sleep monitoring data processing method of this application establishes a clear temporal sequence between events by sorting all breathing events according to their start times, allowing for accurate definition of the time distance between adjacent events. Based on this, the time interval between each pair of adjacent events is calculated, with each value reflecting the density between two adjacent events. Finally, by statistically analyzing the dispersion of these time interval values, they are transformed into a numerical index that can distinguish two different distribution states: close time intervals indicate a uniform event distribution, while significant differences in time intervals indicate event clustering. These steps extract the temporal distribution pattern of breathing events throughout the night from the raw time-point information into a quantifiable feature value. This feature value reflects the uniformity or clustering of event distribution, providing a quantitative temporal input for subsequent multi-dimensional fusion.

[0104] As another implementation of this application, S104B3 may also include the following steps S104B31-S104B34.

[0105] S104B31 divides a preset time period into multiple sub-time periods.

[0106] As an example, a sub-time period can refer to dividing a preset time period of overnight sleep monitoring into several consecutive and non-overlapping time intervals in chronological order. Each sub-time period covers a local segment of the entire night's sleep, such as the division between the first half and the second half of the night, or multiple time periods divided by hours.

[0107] Specifically, as an example, the total sleep duration can be divided equally into multiple sub-segments of equal length. For instance, if the total sleep duration is 480 minutes, it can be divided into 8 sub-segments, each 60 minutes long. Alternatively, a non-equal division method can be used, such as dividing by sleep stage or by the first and second halves of the night. For each sub-segment, its start and end times, as well as its position within the entire preset time period, need to be clearly defined. This step decomposes the entire night's sleep into multiple local time periods, providing analytical units for subsequent analysis of the distribution differences of respiratory events in different time periods, allowing the trend of workload changes throughout the night to be quantitatively captured.

[0108] S104B32, for each sub-time period, the duration of each respiratory event within the sub-time period is weighted and summed with the preset weight of each respiratory event to obtain the weighted duration of the sub-time period.

[0109] As an example, the weighted duration of a sub-time period refers to the sum of the actual duration of each respiratory event within that sub-time period multiplied by its corresponding weight coefficient. This value reflects the aggregated respiratory event load within that sub-time period.

[0110] Specifically, as an example, for each sub-time period, the start and end times of that sub-time period are determined, and respiratory events whose start times fall within the range of that sub-time period are selected from all respiratory events. For each selected respiratory event, its actual duration and weight coefficient are obtained. The weight coefficient is multiplied by the duration to obtain the weighted duration value of the event. Then, the weighted duration values ​​of all events within that sub-time period are summed.

[0111] S104B33, determine the difference index of duration based on the weighted duration of each sub-time period.

[0112] As an example, the duration difference index can refer to a numerical value that quantifies the degree of difference between the weighted durations of all sub-time periods. This index reflects whether the respiratory event load is evenly distributed across different periods of sleep throughout the night.

[0113] Specifically, as an example, the weighted duration of all calculated sub-time periods is used as a dataset, and the dispersion of this dataset is calculated. Dispersion can be calculated using methods such as standard deviation or range. Standard deviation reflects the degree of deviation of each sub-time period's value from the mean, while range reflects the difference between the maximum and minimum values. When the weighted duration of each sub-time period is close to each other, the difference index is small, indicating that the respiratory event load is relatively evenly distributed throughout the night. When the weighted duration of some sub-time periods is significantly higher than that of other sub-time periods, the difference index is large, indicating a significant periodic concentration of the load throughout the night. This step quantifies the evenness of the respiratory event load distribution across different time periods throughout the night into a numerical index, providing dimensional information reflecting inter-time period differences for calculating the event temporal distribution characteristics.

[0114] S104B34 calculates the event time distribution characteristics based on the dispersion and duration differences of all time intervals.

[0115] Specifically, as an example, the calculated time interval dispersion and duration difference indices can be normalized separately to eliminate differences in numerical range and units. After normalization, the two indices are fused using a weighted summation method to obtain a composite event time distribution characteristic. The weights of the two indices can be equal or adjusted according to clinical needs during fusion. The event time distribution characteristic obtained after fusion reflects two aspects of information: the time interval dispersion reflects the local density variation of adjacent events on the time axis, and the duration difference index reflects the overall distribution balance of the workload across different time periods.

[0116] This step integrates the distribution information from two complementary dimensions into a comprehensive index, which includes both discrete information on local event intervals and balanced information on global time-period load, thus reflecting the distribution pattern of respiratory events more comprehensively.

[0117] The sleep monitoring data processing method of this application divides a preset time period into multiple sub-time periods, decomposing the entire night's sleep into comparable analysis units. Based on this, the weighted duration of respiratory events within each sub-time period is calculated. The duration and weight information of events scattered across various local time periods are aggregated into comparable values, providing a quantitative basis for comparing the workload levels between different time periods. Then, a difference index is calculated based on the weighted duration of each sub-time period. This index reflects the degree of load balance across different time periods throughout the night. Finally, this time period difference index is fused with the dispersion of time intervals between adjacent events to form a composite event time distribution feature. This feature includes information on local density variations between adjacent events as well as information on the global balance of workload across different time periods, enabling the final event time distribution feature to comprehensively characterize the distribution of respiratory events throughout the night's sleep from two complementary dimensions.

[0118] As another implementation of this application, S105 may also include the following steps S1051-S1052.

[0119] S1051, normalize the event-related hypoxic load, event duration, and event temporal distribution characteristics respectively.

[0120] As an example, normalization can refer to the operation of mapping original indicators with different dimensions and numerical ranges to the same numerical interval through mathematical transformation. After normalization, the values ​​of each indicator are within a uniform scale range.

[0121] Specifically, as an example, the maximum and minimum values ​​of three indicators—event-related hypoxic load, event duration, and event temporal distribution characteristics—can be obtained from the entire dataset of monitored objects or historical datasets. For each indicator, a maximum-minimum normalization method is used: the original value is subtracted from the minimum value of the indicator, and then divided by the difference between the maximum and minimum values ​​of the indicator, resulting in a normalized value for the indicator in the range of 0 to 1. After this normalization process, the values ​​of the three indicators are all mapped to the range of 0 to 1. This step solves the incomparability problem caused by the different units and numerical ranges of the three indicators, ensuring that the three indicators have the same scale during subsequent fusion, and reducing the influence of any one indicator due to its excessively large original numerical range on the fusion result.

[0122] S1052, by integrating the normalized event-related hypoxia load, event duration, and event time distribution characteristics, a comprehensive event load index is obtained.

[0123] Specifically, as an example, three normalized values ​​are used as input, and integrated using either a multiplicative fusion method or a weighted summation fusion method. Multiplicative fusion multiplies the three normalized values ​​to obtain a comprehensive index; in this method, a lower value in any dimension will lower the final comprehensive index value. Weighted summation fusion assigns a weight coefficient to each dimension, multiplies it by the corresponding normalized value, and then sums the results. The weight coefficients for each dimension can be adjusted according to clinical needs. Furthermore, sleep stage correction factors, postural correction factors, respiratory effort correction factors, or individual clinical characteristic correction factors can be incorporated into the fusion process. These factors are normalized and then fused together with the three basic dimensions. The resulting comprehensive event load index is a continuous value within the range of 0 to 1. This step integrates information from three disparate dimensions into a single, comprehensive value that is easy to interpret and compare, allowing clinical users to obtain an overall assessment of the cardiovascular physiological burden of the monitored subject without having to analyze multiple independent indicators separately.

[0124] The sleep monitoring data processing method in this application normalizes three original indicators—event-related hypoxia load, event duration, and event temporal distribution characteristics—to a unified numerical range. This resolves the incomparability caused by differences in units and numerical ranges, ensuring that each dimension has the same scale in subsequent fusion and preventing any single dimension from dominating the fusion result due to an excessively large original numerical range. Fusion is then performed based on normalization, integrating the information from the three dimensions at comparable scales into a single value through multiplicative fusion or weighted summation. This value simultaneously reflects comprehensive information on the degree of hypoxia accumulation, event duration, and event distribution pattern. These steps aggregate physiological load information, originally scattered across multiple independent indicators, into a single, easily interpretable comprehensive event load indicator. Clinical users can use this single value to make an overall assessment of the cardiovascular physiological burden related to sleep-disordered breathing in the monitored subject.

[0125] It should be noted that after the sleep monitoring data of the subject to be monitored within the preset time period, the blood oxygen saturation signal is preprocessed in this embodiment. Various signal repair methods such as linear interpolation, spline interpolation, local regression, or model imputation can be used to restore missing segments. The threshold for removing outliers, the rules for processing missing segments, and the method for identifying artifacts can all be adaptively adjusted according to the type of acquisition equipment and the actual signal quality to ensure that the blood oxygen saturation time series entering the subsequent calculation process has reliable data quality.

[0126] This application also validated the technical solution of the present invention using data from a previous sleep monitoring cohort. This cohort included 4904 subjects with complete sleep monitoring signals and long-term follow-up data, with a median follow-up time of approximately 11.0 years. During the follow-up period, 341 cardiovascular death events were recorded. The validation results showed that the comprehensive event load index of the present invention, after adjusting for age, sex, body mass index, traditional cardiovascular risk factors, sleep structure indicators, mean nocturnal oxygen saturation, and traditional respiratory event frequency, remained independently associated with the risk of cardiovascular death. In ternary grouping analysis, the highest comprehensive event load group had a significantly higher risk of cardiovascular death than the lowest comprehensive event load group, with a hazard ratio of 1.72, a 95% confidence interval of 1.28 to 2.33, and a trend test p-value less than 0.001. In quintile grouping analysis, the highest comprehensive event load group also had a higher risk of cardiovascular death than the lowest comprehensive event load group, with a hazard ratio of 1.97, a 95% confidence interval of 1.34 to 2.91, and a trend test p-value less than 0.001. When analyzed as a continuous variable, the risk of cardiovascular death increases by one standard deviation for each increase in the comprehensive event load index, with a hazard ratio of 1.13 and a 95 percent confidence interval of 1.05 to 1.22.

[0127] In incremental predictive analysis, the consistency index of the clinical risk model improved to 0.807 after incorporating the comprehensive event load index, while the consistency index of the model after incorporating the traditional respiratory event frequency index or the single hypoxia load index was approximately 0.803. After incorporating the comprehensive event load index, the continuous net weight classification improvement index was 0.507, with a 95% confidence interval of 0.389 to 0.596; the comprehensive discrimination improvement index was 0.035, with a 95% confidence interval of 0.027 to 0.045. These results demonstrate that the comprehensive event load quantification method of this invention can provide additional quantitative information for cardiovascular risk assessment beyond traditional respiratory event frequency indicators and single hypoxia indicators.

[0128] Validation across different clinical outcomes showed that in coronary heart disease mortality, the highest quintile of the comprehensive event load had a higher risk than the lowest quintile, with a hazard ratio of 2.36, a 95% confidence interval of 1.45 to 3.86, and a trend test p-value less than 0.001. In heart failure, the highest quintile also had a higher risk than the lowest quintile, with a hazard ratio of 1.37, a 95% confidence interval of 1.03 to 1.83, and a trend test p-value of 0.035. Sensitivity analysis with different event weights showed that the association between this comprehensive index and cardiovascular mortality risk remained stable, indicating that the proposed solution has good robustness. In summary, the comprehensive event load index of this invention demonstrates good risk stratification ability in cardiovascular death, coronary heart disease mortality, and heart failure, and can improve the model's discrimination and risk reclassification ability based on existing clinical models.

[0129] Based on the sleep monitoring data processing method provided in the above embodiments, this application also provides specific implementation methods of the sleep monitoring data processing device. Please refer to the following embodiments.

[0130] First see Figure 5 The sleep monitoring data processing device 50 provided in this application embodiment includes the following modules: Data acquisition module 501 is used to acquire sleep monitoring data of the subject to be monitored within a preset time period; the sleep monitoring data includes blood oxygen saturation time series; Event recognition module 502 is used to identify respiratory events in blood oxygen saturation time series and obtain at least one respiratory event and its corresponding time window; The hypoxia calculation module 503 is used to calculate the event-related hypoxia load within a preset time period based on the blood oxygen saturation within the time window corresponding to each respiratory event. The event time module 504 is used to calculate the event duration and event time distribution characteristics within a preset time period based on the time window corresponding to each respiratory event; The data fusion module 505 is used to fuse event-related hypoxia load, event duration, and event time distribution characteristics to obtain cardiovascular event load indicators for the subjects to be monitored.

[0131] In some embodiments, the hypoxia calculation module 503 may further include the following modules: The reference value determination module is used to determine the blood oxygen saturation reference value for each first respiratory event based on the blood oxygen saturation before the first respiratory event; the first respiratory event is any respiratory event corresponding to the blood oxygen saturation time series. The area of ​​decline calculation module is used to calculate the area of ​​the decline curve of blood oxygen saturation relative to the blood oxygen saturation reference value within the time window corresponding to the first respiratory event. The hypoxia load calculation module is used to determine the area of ​​the decline curve of blood oxygen saturation per unit sleep time based on the total area of ​​the decline curve corresponding to all respiratory events and the duration of a preset time period, thereby obtaining the event-related hypoxia load.

[0132] In some embodiments, the reference value determination module may further include the following modules: The interval calculation module is used to calculate the time interval between the end time of the second respiratory event and the start time of the first respiratory event; the second respiratory event is the preceding respiratory event of the first respiratory event. The recovery degree calculation module is used to calculate the recovery degree of blood oxygen saturation within a time interval based on the blood oxygen saturation reference value of the second respiratory event. The reference value calculation module is used to determine the blood oxygen saturation reference value corresponding to the first respiratory event based on the time interval and the degree of recovery.

[0133] In some embodiments, the reference value calculation module is specifically used to: when the time interval is greater than a preset time interval and the recovery degree is greater than a preset recovery threshold, take the blood oxygen saturation characteristic value within a preset duration before the start time of the first respiratory event as the blood oxygen protection reference value of the first respiratory event; the blood oxygen saturation characteristic value includes the mean or median of blood oxygen saturation within a preset duration before the start time of the second respiratory event; when the time interval is less than or equal to the preset time interval or the recovery degree is less than or equal to the preset recovery threshold, take the maximum blood oxygen saturation value within the preset duration before the start time of the first respiratory event as the blood oxygen protection reference value of the first respiratory event.

[0134] In some embodiments, the event timing module 504 may further include the following modules: The weight determination module is used to determine the weight of each respiratory event based on its type. The time-weighted module is used to sum the duration of each respiratory event based on the weight of each respiratory event, so as to obtain the weighted duration of the respiratory event. The duration module is used to determine the weighted duration within a unit of sleep time based on the weighted duration of the breathing event and the duration of a preset time period, thus obtaining the event duration.

[0135] In some embodiments, the sleep monitoring data further includes sleep state parameters, and the weight determination module is specifically used for: Different basic weights are assigned to different types of respiratory events based on their type; The feature correction factor is determined based on the preset risk coefficients corresponding to the sleep state parameters and the physiological characteristic parameters of the subject to be monitored. The weights of each respiratory event are determined by adjusting the base weights using a feature correction factor.

[0136] In some embodiments, the event timing module 504 may further include the following modules: The event sequence module is used to sort all respiratory events according to the start time of each respiratory event, forming a respiratory event sequence; The time interval module is used to calculate the time interval between every two adjacent respiratory events in a respiratory event sequence; The event discretization module is used to calculate the event time distribution characteristics based on the degree of discretization of all time intervals.

[0137] In some embodiments, the event discretization module is further configured to: Divide the preset time period into multiple sub-time periods; For each sub-time period, the duration of each respiratory event within the sub-time period is weighted and summed with the preset weight of each respiratory event to obtain the weighted duration of the sub-time period. Based on the weighted duration of each sub-time period, determine the difference index of duration; The event time distribution characteristics are calculated based on the differences in the dispersion and duration of all time intervals.

[0138] In some embodiments, the data fusion module 505 is specifically used for: Normalization was performed on the event-related hypoxic load, event duration, and event temporal distribution characteristics. By integrating the normalized event-related hypoxia load, event duration, and event time distribution characteristics, a comprehensive event load index is obtained.

[0139] Figure 6 A schematic diagram of the hardware structure of the electronic device provided in an embodiment of this application is shown.

[0140] The electronic device may include a processor 601 and a memory 602 storing computer program instructions.

[0141] Specifically, the processor 601 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0142] Memory 602 may include mass storage for data or instructions. For example, and not limitingly, memory 602 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 602 may include removable or non-removable (or fixed) media. Where appropriate, memory 602 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 602 is non-volatile solid-state memory.

[0143] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the methods according to one aspect of this disclosure.

[0144] The processor 601 reads and executes computer program instructions stored in the memory 602 to implement any of the sleep monitoring data processing methods in the above embodiments.

[0145] In one example, the electronic device may also include a communication interface 603 and a bus 610. For example, Figure 6 As shown, the processor 601, memory 602, and communication interface 603 are connected through bus 610 and complete communication with each other.

[0146] The communication interface 603 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0147] Bus 610 includes hardware, software, or both, that couples components of an electronic device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 610 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, this application contemplates any suitable bus or interconnect.

[0148] Furthermore, in conjunction with the sleep monitoring data processing methods in the above embodiments, this application embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the sleep monitoring data processing methods in the above embodiments.

[0149] This application also provides a computer program product, including a computer program that, when executed by a processor, implements any of the sleep monitoring data processing methods described in the above embodiments.

[0150] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0151] The functional blocks shown in the above block diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0152] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0153] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0154] The above are merely specific embodiments of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A method for processing sleep monitoring data, characterized in that, include: Acquire sleep monitoring data of the subject to be monitored within a preset time period; The sleep monitoring data includes a time series of blood oxygen saturation. Identify respiratory events in the blood oxygen saturation time series to obtain at least one respiratory event and its corresponding time window; Based on the blood oxygen saturation within the time window corresponding to each respiratory event, the event-related hypoxia load within the preset time period is calculated respectively. Based on the time window corresponding to each respiratory event, calculate the event duration and event time distribution characteristics within the preset time period; By integrating the event-related hypoxia load, the event duration, and the event temporal distribution characteristics, the cardiovascular event load index of the subject to be monitored is obtained.

2. The method according to claim 1, characterized in that, The step of calculating the event-related hypoxic load within the preset time period based on the blood oxygen saturation within the time window corresponding to each respiratory event includes: For each first respiratory event, a reference value for blood oxygen saturation for the first respiratory event is determined based on the blood oxygen saturation before the first respiratory event; the first respiratory event is any respiratory event corresponding to the blood oxygen saturation time series. Within the time window corresponding to the first respiratory event, calculate the area under the curve of the decrease in blood oxygen saturation relative to the blood oxygen saturation reference value; Based on the total area of ​​the decline curves corresponding to all respiratory events and the duration of the preset time period, the area of ​​the decline curve of blood oxygen saturation per unit sleep time is determined, and the event-related hypoxia load is obtained.

3. The method according to claim 2, characterized in that, For each first respiratory event, determining a reference value for blood oxygen saturation in the first respiratory event based on the blood oxygen saturation before the first respiratory event includes: Calculate the time interval between the end time of the second respiratory event and the start time of the first respiratory event; the second respiratory event is the preceding respiratory event of the first respiratory event; Based on the blood oxygen saturation reference value of the second respiratory event, the degree of recovery of blood oxygen saturation within the time interval is calculated; Based on the time interval and the degree of recovery, a reference value for blood oxygen saturation corresponding to the first respiratory event is determined.

4. The method according to claim 3, characterized in that, The step of determining the blood oxygen saturation reference value corresponding to the first respiratory event based on the time interval and the degree of recovery includes: If the time interval is greater than a preset time interval and the recovery degree is greater than a preset recovery threshold, the blood oxygen saturation characteristic value within a preset duration before the start time of the first respiratory event is used as the blood oxygen protection reference value of the first respiratory event; the blood oxygen saturation characteristic value includes the mean or median of blood oxygen saturation within the preset duration before the start time of the second respiratory event; If the time interval is less than or equal to the preset time interval or the recovery degree is less than or equal to the preset recovery threshold, the maximum blood oxygen saturation value within the preset duration before the start time of the first respiratory event is taken as the blood oxygen protection reference value of the first respiratory event.

5. The method according to any one of claims 1-4, characterized in that, The step of calculating the duration of events within the preset time period based on the time window corresponding to each respiratory event includes: The weight of each respiratory event is determined according to its type. Based on the weights of each respiratory event, the duration of each respiratory event is weighted and summed to obtain the weighted duration of the respiratory event. The weighted duration of the respiratory event and the duration of the preset time period are used to determine the weighted duration within a unit of sleep time, thereby obtaining the event duration.

6. The method according to claim 5, characterized in that, The sleep monitoring data also includes sleep state parameters; The weight of each respiratory event is determined according to its type, including: Different basic weights are assigned to different types of respiratory events based on their type; The feature correction factor is determined based on the preset risk coefficients corresponding to the sleep state parameters and the physiological characteristic parameters of the subject to be monitored. The base weights are adjusted using the feature correction factor to determine the weights of each respiratory event.

7. The method according to any one of claims 1-4, characterized in that, The step of calculating the event time distribution characteristics within the preset time period based on the time window corresponding to each respiratory event includes: All respiratory events are sorted according to their start time to form a respiratory event sequence; Calculate the time interval between every two adjacent respiratory events in the respiratory event sequence; The event time distribution characteristics are calculated based on the degree of dispersion of all the time intervals.

8. The method according to claim 7, characterized in that, The step of calculating the event time distribution characteristics based on the dispersion of all the time intervals includes: The preset time period is divided into multiple sub-time periods; For each sub-time period, the duration of each respiratory event within the sub-time period is weighted and summed with the preset weight of each respiratory event to obtain the weighted duration of the sub-time period. Based on the weighted duration of each sub-time period, determine the difference index of duration; The event time distribution characteristics are calculated based on the dispersion of all the time intervals and the difference index of the duration.

9. The method according to claim 1, characterized in that, The comprehensive event load index is obtained by integrating the event-related hypoxic load, the event duration, and the event temporal distribution characteristics, including: The hypoxic load associated with the event, the duration of the event, and the temporal distribution characteristics of the event were normalized respectively. The comprehensive event load index is obtained by integrating the normalized event-related hypoxia load, event duration, and event time distribution characteristics.

10. A sleep monitoring data processing device, characterized in that, The device includes: The data acquisition module is used to acquire sleep monitoring data of the subject to be monitored within a preset time period; the sleep monitoring data includes a blood oxygen saturation time series. The event recognition module is used to identify respiratory events in the blood oxygen saturation time series and obtain at least one respiratory event and its corresponding time window; The hypoxia calculation module is used to calculate the event-related hypoxia load within the preset time period based on the blood oxygen saturation within the time window corresponding to each respiratory event. The event time module is used to calculate the event duration and event time distribution characteristics within the preset time period based on the time window corresponding to each respiratory event; The data fusion module is used to fuse the event-related hypoxia load, the event duration, and the event time distribution characteristics to obtain the cardiovascular event load index of the subject to be monitored.

Images