Sleep monitoring system and method

Abstract

This invention provides a sleep monitoring system and method. The system includes: an EEG acquisition module for acquiring raw EEG data and head angular acceleration data during a user's sleep process, as first raw data; a PPG acquisition module for acquiring raw PPG data and hand acceleration data during the user's sleep process, as second raw data; a local computing module for performing primary sleep analysis based on the first and second raw data to obtain primary analysis results; a cloud computing module for performing secondary sleep analysis based on the primary analysis results, the first and second raw data, to obtain secondary analysis results; and a user interaction module for visualizing the primary and / or secondary analysis results for user viewing. This invention reduces the computational burden on each module by distributing computational tasks through multi-level operations, resulting in low cost and good scalability.

Description

Technical Field

[0001] This invention relates to the field of sleep monitoring technology, and more particularly to a sleep monitoring system and method. Background Technology

[0002] Sleep quality is closely related to physical health. Statistics show that about one-third of the world's population experiences sleep problems to varying degrees, with difficulty falling asleep, light sleep, and frequent awakenings being particularly common. Long-term poor sleep quality not only affects daytime work efficiency and daily life, but is also closely linked to various chronic diseases such as hypertension, heart disease, diabetes, and depression.

[0003] Traditional sleep monitoring primarily relies on polysomnography (PSG) in hospitals. While offering high accuracy, it is costly, complex, restricts movement while wearing the device, and is bulky, making it unsuitable for everyday use by the general population. Some studies have used electroencephalogram (EEG) data for sleep analysis, but sleep analysis systems relying solely on EEG data have a limited data source, poor sleep staging performance, are susceptible to interference, and have limited physiological signal acquisition capabilities.

[0004] In addition, sleep analysis tasks are usually performed by a local server or a remote computing server. When multiple users are using the device, the server's computing pressure is high, making it unable to efficiently process complex data analysis, and the device's computing power is limited.

[0005] In summary, existing technologies suffer from high usage costs and heavy computational burdens. Summary of the Invention

[0006] This invention provides a sleep monitoring system and method to address the shortcomings of existing technologies, such as high cost and heavy computational burden, and to achieve sleep monitoring with low cost and low computational burden.

[0007] This invention provides a sleep monitoring system, comprising the following modules: The EEG acquisition module is used to collect raw EEG data and head angular acceleration data during the user's sleep process as the first raw data; The PPG acquisition module is used to collect raw PPG data and hand acceleration data during the user's sleep process as the second raw data. The local computing module is used to perform a first-level sleep analysis based on the first raw data and the second raw data to obtain the first-level analysis results. The first-level sleep analysis includes sleep stage calculation and physiological parameter calculation. The cloud computing module is used to perform secondary sleep analysis based on the primary analysis results, the first raw data, and the second raw data to obtain secondary analysis results. The secondary sleep analysis includes sleep efficiency calculation, sleep stage percentage calculation, physiological signal dynamic trend analysis, and abnormal data point detection. The user interaction module is used to visualize the primary analysis results and / or the secondary analysis results for users to view.

[0008] According to a sleep monitoring system provided by the present invention, the electroencephalogram (EEG) acquisition module is worn on a preset area of ​​the user's head during the acquisition process. The EEG acquisition module includes an EEG amplifier, electrode patches, and an angular acceleration measuring device. The EEG amplifier and the electrode patches are connected by magnetic coupling and metal contact positioning; and / or The PPG acquisition module is installed in a wearable portable device and is worn on a preset part of the user's body during the acquisition process. The PPG acquisition module includes a PPG sensor and a hand acceleration measurement device.

[0009] According to a sleep monitoring system provided by the present invention, the electrode patch is made of a skin-friendly adhesive material.

[0010] According to a sleep monitoring system provided by the present invention, a primary sleep analysis is performed based on the first raw data and the second raw data to obtain the primary analysis results, including: Using a preset analysis window, extract the first raw data and the second raw data that have been added between the last processing completion time and the current processing time, and use them as the current first raw data and the current second raw data. Based on the hand acceleration data in the current second data, determine the current body movement state; Preprocessing is performed on the EEG raw data in the current first raw data and the PPG raw data in the current second raw data to obtain EEG preprocessed data and PPG preprocessed data. When the current body movement state is the first body movement state, sleep stage calculation is performed based on the EEG preprocessing data to obtain the sleep stage result; When the current body movement state is the second body movement state, sleep staging is calculated based on the EEG preprocessing data and the PPG preprocessing data to obtain the sleep staging result; Based on the PPG raw data in the current second raw data, preset physiological parameters are calculated, wherein the preset physiological parameters include heart rate, heart rate variability, respiratory rate and / or blood oxygen saturation; Based on the sleep stage results and the preset physiological parameters, the first-level analysis results are obtained.

[0011] According to a sleep monitoring system provided by the present invention, determining the current body movement state based on hand acceleration data in the current second data includes: The hand acceleration data in the current second data is smoothed using a moving average to obtain an average hand acceleration result; Calculate the amplitude of hand acceleration based on the average hand acceleration result; If the maximum value of the hand acceleration amplitude is greater than a preset hand acceleration threshold, the current body movement state is determined to be the first body movement state. If the maximum value of the hand acceleration amplitude is not greater than the preset hand acceleration threshold, the current body movement state is determined to be the second body movement state. According to a sleep monitoring system provided by the present invention, the system preprocesses the EEG raw data in the current first raw data and the PPG raw data in the current second raw data to obtain EEG preprocessed data and PPG preprocessed data, including: The EEG raw data in the current first raw data is filtered, and the filtered EEG raw data in the current first raw data is standardized to obtain standardized EEG data. The standardized EEG data is segmented according to a preset segmentation window length to obtain preprocessed EEG data. The PPG raw data in the current second raw data is divided into DC component and AC component; The AC component is bandpass filtered to preserve the heart rate-related frequency band, thus obtaining the AC filtering result; The AC filtering results are then subjected to data standardization processing to obtain standardized PPG data; The standardized PPG data is segmented according to the preset segmentation window length to obtain PPG preprocessed data.

[0012] According to a sleep monitoring system provided by the present invention, a secondary sleep analysis is performed based on the primary analysis results, the first raw data, and the second raw data to obtain secondary analysis results, including: The head angular acceleration data in the current first raw data is smoothed using a moving average to obtain the average head angular acceleration result. Calculate the amplitude of head angular acceleration based on the average head angular acceleration result; Based on the sleep stage results, the head angular acceleration amplitude, and the hand acceleration amplitude, determine the sleep preparation start time and wake-up time, as well as the start and end times of each stage of the sleep stage results; The time spent in bed is determined based on the sleep start time and the wake-up time; The actual sleep time is determined based on the start and end times of each sleep stage as described in the sleep stage results. Sleep efficiency is determined based on the time spent in bed and the actual sleep time. Based on the start and end times of each stage of the sleep staging results and the actual sleep time, the percentage of each sleep stage in the sleep staging results is determined. Using a pre-set sliding window, dynamic trend analysis is performed on the preset physiological parameters, and abnormal data points are detected based on a set threshold. The secondary analysis results are obtained based on the sleep efficiency, the proportion of sleep stages, the results of the dynamic trend analysis, and the abnormal data points.

[0013] According to a sleep monitoring system provided by the present invention, when the current body movement state is a first body movement state, sleep staging is calculated based on the electroencephalogram (EEG) preprocessing data to obtain a sleep staging result, including: When the current body movement state is the first body movement state, the EEG preprocessing data is input into the pre-trained EEG sleep staging deep learning model to obtain the sleep staging result.

[0014] According to a sleep monitoring system provided by the present invention, when the current body movement state is a second body movement state, sleep staging is calculated based on the EEG preprocessing data and the PPG preprocessing data to obtain a sleep staging result, including: When the current body movement state is the second body movement state, the EEG preprocessing data is input into the pre-trained EEG sleep staging deep learning model to obtain the first recognition result; The PPG preprocessed data is input into a pre-trained PPG sleep staging deep learning model to obtain a second identification result; Based on the first identification result and the second identification result, the sleep staging result is obtained.

[0015] This invention provides a sleep monitoring method, comprising the following steps: Based on the EEG acquisition module, raw EEG data and head angular acceleration data during the user's sleep process are collected as the first raw data; Based on the PPG acquisition module, PPG raw data and hand acceleration data during the user's sleep process are collected as the second raw data; Based on the local computing module, a first-level sleep analysis is performed according to the first raw data and the second raw data to obtain the first-level analysis results. The first-level sleep analysis includes sleep stage calculation and physiological parameter calculation. Based on the cloud computing module, a second-level sleep analysis is performed according to the first-level analysis results, the first raw data, and the second raw data to obtain the second-level analysis results. The second-level sleep analysis includes sleep efficiency calculation, sleep stage percentage calculation, dynamic trend analysis of physiological signals, and abnormal data point detection. Based on the user interaction module, the primary analysis results and / or the secondary analysis results are visualized for users to view.

[0016] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the sleep monitoring method as described above.

[0017] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the sleep monitoring method as described above.

[0018] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the sleep monitoring method as described above.

[0019] The sleep monitoring system and method provided by this invention collect multimodal physiological signals from users through an EEG acquisition module and a PPG acquisition module, including raw EEG data, head angular acceleration data, raw PPG data, and hand acceleration data. The devices used for data collection are easy to wear and have low cost. Based on the collected data, a first-level sleep analysis is performed in a local computing module to obtain the first-level analysis results, quickly completing preliminary signal analysis and reducing communication pressure. Based on the first-level analysis results and the collected data, a second-level sleep analysis is performed in a cloud computing module to obtain the second-level analysis results, enabling refined modeling and large-scale data comparison analysis. A user interaction module is also included. This invention reduces the computational burden on each module by distributing computational tasks through multi-level operations, resulting in low cost and good scalability. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of the sleep monitoring system provided by the present invention; Figure 2This is a schematic diagram of the EEG amplifier and electrode patch of the sleep monitoring system provided by the present invention; Figure 3 This is a schematic diagram of a user wearing an EEG acquisition module in the sleep monitoring system provided by the present invention; Figure 4 This is a flowchart illustrating the sleep monitoring method provided by the present invention; Figure 5 This is a schematic diagram of the structure of the electronic device provided by the present invention.

[0022] Figure label: 201: EEG amplifier; 202: Electrode patch; 201-1: One of the metal contacts of the EEG amplifier; 201-2: Another metal contact of the EEG amplifier; 201-3: Another metal contact of the EEG amplifier; 201-4: Another metal contact of the EEG amplifier; 202-1: One of the metal contacts of the electrode patch; 202-2: Another metal contact of the electrode patch; 202-3: Another metal contact of the electrode patch; 202-4: Another metal contact of the electrode patch. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0024] The following is combined Figures 1-3 The sleep monitoring system and method of the present invention are described below. Figure 1 This is a schematic diagram of the sleep monitoring system provided by the present invention, as shown below. Figure 1 As shown, the system includes: The EEG acquisition module 110 is used to collect raw EEG data and head angular acceleration data during the user's sleep process as the first raw data.

[0025] The EEG acquisition module 110 has real-time acquisition capabilities, continuously acquiring EEG signals throughout the user's sleep process as raw EEG data. Simultaneously, it synchronously acquires head angular acceleration data under a unified clock reference.

[0026] Head angular acceleration data is used to record changes in the user's head movement and lying posture information, and may include head acceleration data, head angle data, etc., which are not limited in this invention.

[0027] Understandably, head angular acceleration data is used to help determine whether a user is in an effective sleep state or whether there are abnormal postural changes (such as frequent turning over, abnormal head elevation, etc.).

[0028] The raw EEG data and head angular acceleration data are referred to as the first raw data.

[0029] The PPG acquisition module 120 is used to acquire raw PPG data and hand acceleration data during the user's sleep process as second raw data.

[0030] It should be noted that PPG (Photo Plethysmo Graphy) is a photoplethysmography technique, a non-invasive detection method that uses photoelectric means to detect changes in blood volume in living tissue. The PPG acquisition module 120 can acquire raw PPG data through photoplethysmography, thereby calculating physiological parameters such as heart rate, blood oxygen, heart rate variability, and respiratory rate, and can also be used for sleep quality analysis.

[0031] Simultaneously, hand acceleration data is collected under a unified clock reference for acquiring raw PPG data.

[0032] Hand acceleration data is used to define the motion state of the user's hand, thereby determining whether the hand is in a state of micro-movement.

[0033] Understandably, hand acceleration data is used for dynamic decision-making to determine the specific data used for sleep stage assessment.

[0034] The sleep monitoring of this invention is based on multimodal physiological signals, which integrates various physiological signals including EEG, PPG, and acceleration. The data sources are diverse and not easily affected by interference. The local computing module 130 is used to perform a first-level sleep analysis based on the first raw data and the second raw data to obtain the first-level analysis results. The first-level sleep analysis includes sleep stage calculation and physiological parameter calculation.

[0035] The local computing module 130, as the core hub of the system, is located on the local edge device. It has a built-in lightweight analysis engine, which is mainly used to perform sleep-level analysis on the first and second raw data collected, so as to quickly complete the preliminary signal analysis at the local edge node and reduce the communication pressure.

[0036] The primary sleep analysis includes sleep stage calculation and physiological parameter calculation. Physiological parameter calculation includes the calculation of preset physiological parameters such as heart rate, blood oxygen saturation, heart rate variability, and respiratory rate.

[0037] To achieve data reception, the EEG acquisition module 110 and the PPG acquisition module 120 are communicatively connected to the local computing module 130. This invention does not limit the specific manner in which the EEG acquisition module 110 and the PPG acquisition module 120 are communicated with the local computing module 130.

[0038] In some embodiments, the EEG acquisition module 110 and PPG acquisition module 120 transmit data to the local computing module 130 via Bluetooth connection. Further, the local computing module 130 can be understood as having a Bluetooth communication module that supports Bluetooth pairing, device connection, etc., for further user operations.

[0039] Based on the above embodiments, in some embodiments, further operations performed by the user include: setting up a Wi-Fi network via a mobile phone on top of a Bluetooth connection to enable the system's network connectivity. It should be further noted that the system's network connectivity can be manually enabled based on a Bluetooth connection, or the Wi-Fi module can be configured separately to enable the system's network connectivity.

[0040] The local computing module 130 is also used to store the received first raw data and second raw data, and to add a time stamp.

[0041] It should be noted that, in some embodiments, a data receiving module is provided in the local computing module 130 to receive the first raw data transmitted by the EEG acquisition module 110 and the second raw data transmitted by the PPG acquisition module 120. The data receiving module can also be used to timestamp and store the data.

[0042] Furthermore, in some embodiments, the local computing module 130 provides an interface for users to query the first raw data and the second raw data according to time.

[0043] Furthermore, in some embodiments, the local computing module 130 is disposed in the multi-functional charging compartment.

[0044] Based on the above embodiments, the multi-functional charging case also includes a charging module and a power module. The charging module supplies power to the EEG acquisition module 110 and the PPG acquisition module 120, while the power module has a built-in rechargeable battery for supplying power to the internal systems of the multi-functional charging case.

[0045] After obtaining the primary analysis results, the primary analysis results, along with the first and second raw data, are sent to the cloud computing module 140 for further processing and result display. In the following embodiment, this sending task is implemented by a separately configured data upload module.

[0046] Furthermore, in one embodiment, while sending data, the connection status of the EEG acquisition module 110 and PPG acquisition module 120 with the local computing module 130 is reported to the cloud computing module 140 for user query.

[0047] The cloud computing module 140 is used to perform secondary sleep analysis based on the primary analysis results, the first raw data, and the second raw data to obtain secondary analysis results. The secondary sleep analysis includes sleep efficiency calculation, sleep stage percentage calculation, physiological signal dynamic trend analysis, and abnormal data point detection.

[0048] The cloud computing module 140 can be set up in a cloud data center to realize remote computing and data storage. The cloud computing module 140 is the data analysis core and information management hub of the system, with high-performance data processing and large-capacity storage capabilities.

[0049] It should be further explained that the cloud computing module 140 is mainly responsible for receiving the first-level analysis results, the first raw data and the second raw data uploaded from the local computing module 130, and performing higher-level statistical modeling and in-depth analysis (i.e., sleep secondary analysis) on this basis.

[0050] The secondary sleep analysis includes calculating the user's sleep efficiency (i.e., the ratio of actual sleep duration to time spent in bed), analyzing the proportion of each sleep stage (e.g., the proportion of REM sleep, light sleep, and deep sleep), tracking the dynamic changes of physiological signals throughout the night (e.g., heart rate, respiration, blood oxygenation, etc.), and detecting abnormal data points (e.g., identifying health risks such as nocturnal hypoxia, frequent awakenings, and abnormal heart rate fluctuations).

[0051] Furthermore, the cloud computing module 140 is also used for time-archiving and storage of user sleep data, supporting long-term retention and trend analysis, and providing standardized interface services for user terminal devices to retrieve individual sleep analysis reports and historical data.

[0052] Users can use the networking function described in the above embodiments to communicate with a remote computing and data storage server and upload the first-level analysis results, the first raw data, and the second raw data to the cloud.

[0053] The user interaction module 150 is used to visualize the primary analysis results and / or the secondary analysis results for users to view.

[0054] User interaction module 150 is used to interact with users so that they can view a visual version of the primary analysis results and / or secondary analysis results.

[0055] In some embodiments, an analysis results interface is set up to display the visualized results of primary and / or secondary analysis. The visualized results include: sleep scores and staging graphs, heart rate / respiration / blood oxygenation change curves, abnormal alerts and health advice, and historical record viewing.

[0056] To enhance user experience, the analysis results interface can be further customized, such as allowing users to view historical data for any date, compare sleep patterns on different days, support cross-device synchronization, support multi-device login, and enable automatic data synchronization.

[0057] Furthermore, in some embodiments, the user interaction module 150 supports users interacting with the system using smart terminals (such as mobile phones, tablets, and computers).

[0058] Furthermore, based on the above embodiments, the user connects to the system via a smart terminal through the user interaction module 150. Based on the Bluetooth connection, the user sets up a Wi-Fi network through the smart terminal, guides the user to access the user's home wireless network, and connects to the Internet to access the cloud computing module 140, thereby enabling data upload from the local computing module 130 to the cloud computing module 140.

[0059] The sleep monitoring system and method provided by this invention collect multimodal physiological signals from users through an EEG acquisition module and a PPG acquisition module, including raw EEG data, head angular acceleration data, raw PPG data, and hand acceleration data. The devices used for data collection are easy to wear and have low cost. Based on the collected data, a first-level sleep analysis is performed in a local computing module to obtain the first-level analysis results, quickly completing preliminary signal analysis and reducing communication pressure. Based on the first-level analysis results and the collected data, a second-level sleep analysis is performed in a cloud computing module to obtain the second-level analysis results, enabling refined modeling and large-scale data comparison analysis. A user interaction module is also included. This invention reduces the computational burden on each module by distributing computational tasks through multi-level operations, resulting in low cost and good scalability.

[0060] Furthermore, to increase the diversity of data sources, the system may also include other physiological signal acquisition modules to perform multi-dimensional, multimodal physiological signal acquisition. Other physiological signal acquisition modules include, for example, a mandibular electromyography (EMG) acquisition module and an electrooculography (EOG) acquisition module; this invention does not limit the scope of these modules.

[0061] The following provides a further description of the EEG acquisition module 110 and the PPG acquisition module 120. In some embodiments, the EEG acquisition module 110 is worn on a preset area of ​​the user's head during the acquisition process. The EEG acquisition module 110 includes an EEG amplifier, electrode patches, and an angular acceleration measurement device. The EEG amplifier and the electrode patches are connected by magnetic coupling and metal contact positioning; and / or The PPG acquisition module 120 is installed in a wearable portable device and is worn on a preset part of the user's body during the acquisition process. The PPG acquisition module 120 includes a PPG sensor and a hand acceleration measurement device.

[0062] Specifically, the EEG acquisition module 110 adopts a miniaturized design, which is highly portable and wearable. It includes an EEG amplifier and electrode patches, which are worn on a preset area of ​​the user's head during the acquisition process, such as the forehead.

[0063] like Figure 2 As shown, the EEG amplifier and electrode patch are positioned and connected using magnetic coupling and metal contacts. Figure 2 In the diagram, 201 is the EEG amplifier, 202 is the electrode patch, 201-1, 201-2, 201-3, and 201-4 are the metal contacts of the EEG amplifier, and 202-1, 202-2, 202-3, and 202-4 are the metal contacts of the electrode patch. The metal contacts of the EEG amplifier and the metal contacts of the electrode patch are magnetically coupled.

[0064] The EEG acquisition module 110 also integrates an angular acceleration measurement device. The angular acceleration measurement device and the EEG acquisition circuit of the EEG amplifier are placed on the same circuit board to capture head motion information in real time. The head angular acceleration data and EEG signals are acquired synchronously under a unified clock reference.

[0065] In some embodiments, the angular acceleration measuring device includes an angular accelerometer and an angle sensor.

[0066] When the EEG acquisition module 110 is worn, such as Figure 3 As shown, the device is worn on the forehead of the user's head, with the reference electrode and ground electrode positioned at the midline, and the EEG acquisition electrodes symmetrically distributed on both sides.

[0067] The PPG acquisition module 120 is installed in a wearable portable device, such as a ring or bracelet. During the acquisition process, the PPG acquisition module 120 is worn on a user-preset body part corresponding to the wearable portable device. For example, when the PPG acquisition module 120 is installed in a wearable ring, it is worn on the user's finger during the acquisition process.

[0068] The PPG acquisition module 120 includes a PPG sensor and a hand acceleration measurement device.

[0069] In some embodiments, the hand acceleration measuring device includes a triaxial accelerometer.

[0070] The embodiments of the present invention use portable micro-devices to collect multimodal physiological signals, which are low in preparation cost, easy to wear, and highly comfortable.

[0071] To further improve user comfort, in some embodiments, the electrode pads are made of a skin-friendly adhesive material.

[0072] The following further describes the steps of the local computing module 130 performing primary sleep analysis. In some embodiments, primary sleep analysis is performed based on the first raw data and the second raw data to obtain primary analysis results, including: Step 131: Using the preset analysis window, extract the first raw data and the second raw data added between the last processing completion time and the current processing time, and use them as the current first raw data and the current second raw data.

[0073] Using a preset analysis window, extract the newly added EEG raw data, head angular acceleration data, PPG raw data, and hand acceleration data from the time the last processing was completed to the current processing time.

[0074] Specifically, using a preset analysis window, the first raw data added between the last processing completion time and the current processing time is extracted as the current first raw data; using a preset analysis window, the second raw data added between the last processing completion time and the current processing time is extracted as the current second raw data.

[0075] The length of the extracted data is the preset analysis window length. The maximum integer multiple of the original data, with any insufficient portion carried over to the next calculation cycle. The total number of sampling points extracted from the current first and second original data. As shown in equation (1):

[0076] in, The time when the last processing was completed. For the current processing time, To preset the window length of the analysis window, Sampling rate, To find the function that yields the largest integer.

[0077] Step 132: Determine the current body movement state based on the hand acceleration data in the current second data.

[0078] Specifically, based on the hand acceleration data extracted from the current second data, the acceleration data amplitude is calculated, and the current body motion state is determined by the acceleration data amplitude.

[0079] Step 133: Preprocess the EEG raw data in the current first raw data and the PPG raw data in the current second raw data to obtain EEG preprocessed data and PPG preprocessed data.

[0080] Preprocessing is performed on the raw EEG data from the first set of raw data and the raw PPG data from the second set of raw data. Specifically, preprocessing is used to remove noise, extract valid data, and improve the efficiency of subsequent operations.

[0081] Step 134: When the current body movement state is the first body movement state, sleep stage calculation is performed based on the EEG preprocessing data to obtain the sleep stage result.

[0082] Step 135: When the current body movement state is the second body movement state, sleep staging is calculated based on the EEG preprocessing data and the PPG preprocessing data to obtain the sleep staging result.

[0083] This invention uses dynamic fusion decision-making based on the current body movement state to determine the data source for calculating sleep staging results. Specifically, the current body movement state is determined based on hand acceleration data, thereby dynamically deciding whether to use raw PPG data. When the current body movement state is the first state, the PPG signal is invalid, and raw PPG data is not used; sleep stage determination is based solely on raw EEG data to obtain the sleep staging result. When the current body movement state is the second state, the PPG signal is valid, and raw PPG data is used for sleep stage determination; that is, raw EEG data and raw PPG data are fused to jointly determine the sleep stage and obtain the sleep staging result. This effectively filters out PPG signal noise caused by hand tremors, friction, turning over, etc.

[0084] Step 136: Calculate preset physiological parameters based on the PPG raw data in the current second raw data, wherein the preset physiological parameters include heart rate, heart rate variability, respiratory rate and / or blood oxygen saturation.

[0085] The AC component is extracted from the raw PPG data and further processed to calculate preset physiological parameters. These preset physiological parameters include heart rate, heart rate variability, respiratory rate, and / or blood oxygen saturation.

[0086] It should be noted that the extraction of the AC component from the PPG signal can be performed during preprocessing. If this step is not performed during preprocessing, it can be done in step 136.

[0087] Step 137: Based on the sleep stage results and the preset physiological parameters, obtain the first-level analysis results.

[0088] The sleep staging results and preset physiological parameters constitute the primary analysis results.

[0089] Furthermore, in some embodiments, the portion of the local computing module 130 performing step 136 is incorporated into a wearable portable device, and after the calculation is completed, the calculation result is uploaded to the local computing module 130.

[0090] Similarly, the portion of the local computing module 130 that executes steps 131-137 can also be set up in other devices or modules to redistribute the multi-level computing tasks and distribute the computing pressure on the devices.

[0091] The following provides a further explanation of step 132. In some embodiments, determining the current body movement state based on the hand acceleration data in the current second data includes: Step 1321: Smooth the hand acceleration data in the current second data using a moving average to obtain the average hand acceleration result.

[0092] The hand acceleration data in the current second data is smoothed using the moving average method, as shown in equation (2): (2); in, This is the average result of hand acceleration. The window size during the moving average process. For acceleration data, For the first A window.

[0093] Step 1322: Calculate the amplitude of hand acceleration based on the average hand acceleration result.

[0094] The amplitude of hand acceleration is calculated based on the mapping of the average hand acceleration result to each axis.

[0095] Taking hand acceleration as a three-axis acceleration as an example, calculate and determine the amplitude of hand acceleration. : (4); in This represents the amplitude of hand acceleration. , , These are mappings of the average hand acceleration results on the X, Y, and Z axes, respectively.

[0096] Step 1323: If the maximum value of the hand acceleration amplitude is greater than the preset hand acceleration threshold, the current body movement state is determined to be the first body movement state.

[0097] If the hand acceleration amplitude If the hand acceleration exceeds a preset threshold, it is determined that the hand movement is large and the PPG signal quality may be significantly affected. Therefore, the current body movement state is defined as the first body movement state, and EEG signals are used for sleep staging calculation. Step 1324: If the maximum value of the hand acceleration amplitude is not greater than the preset hand acceleration threshold, determine the current body movement state as the second body movement state. If the hand acceleration amplitude If the hand acceleration is less than or equal to the preset hand acceleration threshold, it is determined to be a weak hand movement, and sleep staging is determined by combining PPG and EEG.

[0098] The following provides a further explanation of step 133. In some embodiments, the EEG raw data in the current first raw data and the PPG raw data in the current second raw data are preprocessed to obtain EEG preprocessed data and PPG preprocessed data, including: Step 1331: Filter the EEG raw data in the current first raw data, and perform data standardization processing on the filtered EEG raw data in the current first raw data to obtain standardized EEG data.

[0099] The filter can be selected according to the actual situation to filter the EEG raw data in the first raw data, as shown in equation (5): (5); in, The impulse response of the bandpass filter. The data is filtered by the Chebyshev I filter. The EEG raw data is the current first raw data input to the Chebyshev I filter. t This is the current sampling point.

[0100] In some embodiments, based on equation (5), the raw EEG data in the current first raw data is bandpass filtered by 0.5-30Hz using a Chebyshev I filter.

[0101] The amplitude-frequency response of the Chebyshev Type I filter is: (6); Where ε is the passband ripple factor, used to control the magnitude of the passband ripple. It is an nth-order Chebyshev polynomial. This is the filter cutoff angular frequency.

[0102] Furthermore, to eliminate interference from the power frequency signal of the indoor power supply system, this embodiment applies a notch filter after the bandpass filter: (7); in, It is a narrowband suppression filter centered on the power frequency. The data is filtered by a notch filter, which is the original EEG data in the current first original data after filtering in this embodiment.

[0103] Then, the EEG raw data in the current first raw data after filtering is divided into channels according to the mean of the training data. and standard deviation Data standardization is performed, as shown in equation (8): (8); in, To standardize EEG data, the mean value was... and standard deviation They are selected through training.

[0104] Step 1332: Segment the standardized EEG data according to the preset segmentation window length to obtain EEG preprocessed data.

[0105] The standardized EEG data was segmented into K windows of a preset window length to obtain preprocessed EEG data. It should be noted that the number of windows segmented from EEG and PPG data is the same and they correspond strictly in time.

[0106] In some embodiments, the preset segmentation window length is set to be the same as the preset analysis window length, both being... .

[0107] Step 1333: Divide the PPG raw data in the current second raw data into DC component and AC component.

[0108] PPG raw data from the current second raw data It is divided into a direct current component (DC) and an alternating current component (AC), as shown in equation (9): (9); in: (10); (11); In the above formula, τ The moving average window length, t For the current sampling point, For a slowly changing baseline, This refers to the dynamic signal of heartbeat and pulse.

[0109] Step 1334: Perform bandpass filtering on the AC component to retain the heart rate-related frequency band, and obtain the AC filtering result.

[0110] Bandpass filtering is applied to the AC component to retain the heart rate-related frequency band, resulting in the filtered data. Let be the result of AC filtering, as shown in equation (12): (12); in, For the bandpass filter response, For the input AC components, AC filtering results.

[0111] This embodiment removes high-frequency noise and low-frequency drift of the AC component through bandpass filtering, while preserving the pulse waveform characteristics.

[0112] Step 1335: Perform data standardization processing on the AC filtering results to obtain standardized PPG data.

[0113] The AC filtering results were then standardized to obtain standardized PPG data. As shown in equation (13): (13); Among them, the mean and standard deviation They are selected through training.

[0114] Step 1336: Segment the standardized PPG data according to the preset segmentation window length to obtain PPG preprocessed data.

[0115] The standardized PPG data is divided into K windows of a preset window length to obtain preprocessed PPG data. .

[0116] Based on the above embodiments, in order to provide a more detailed explanation of the calculation of preset physiological parameters, a specific embodiment for calculating preset physiological parameters based on the AC component in the raw PPG data is given, including: 1) Heart rate calculation: Perform a fast Fourier transform on the PPG preprocessed data for each window, as shown in equation (17): (17); in, It is a complex value in the frequency domain, and its amplitude spectrum is |X(f)|. For the first PPG data for each window.

[0117] Based on the results of the Fast Fourier Transform, a frequency range corresponding to the heart rate is selected, denoted as . Find the frequency with the largest amplitude. : (18); No. Estimated heart rate from PPG preprocessed data of one window The calculation is as follows: (19) 2) Heart rate variability calculation: Heart rate variability The calculation is shown in equation (20): (20); in, K represents the total number of windows for preprocessing data. Specifically, K is the total number of windows that the first and second raw data are divided into according to the Tw time window for the current analysis. Each window corresponds to a sleep stage result.

[0118] 3) Respiratory rate calculation: Preprocessing PPG data for a single window Perform Hilbert transform and calculate the instantaneous amplitude envelope. : (twenty one); in, This indicates the amplitude modulation of the PPG pulse with respiration. () represents the Hilbert transform, used to extract the instantaneous amplitude of a signal.

[0119] right Power spectrum analysis was performed to identify the energy peak corresponding to the respiratory rate. (twenty two); in, f Represents frequency, ranging from 0.08 to 0.4 Hz (corresponding to a typical respiratory rate of 5–24 breaths / min). Fast Fourier Transform converts a time-domain signal into a frequency-domain signal. The signal power spectrum reflects the energy level of each frequency component.

[0120] Find the frequency corresponding to the peak respiratory rate in the power spectrum. This leads to the calculation of respiratory rate. : (twenty three); (twenty four); in The peak frequency (Hz) corresponding to the respiratory rate. Respiratory rate, measured in breaths / min. Indicates to Find the parameters.

[0121] 4) Blood oxygen calculation: Red light long-wavelength (… ) and infrared long-wave ( The PPG signal of the red light is extracted using equations (10) and (11) to obtain the DC and AC components of two wavelengths, thus obtaining the long-wavelength DC component of the red light. Red light long-wavelength AC component Infrared long-wavelength DC component Infrared long-wave AC component .

[0122] Calculate the ratio R of the AC to DC components of red and infrared light: (25); Based on the empirical formula, the ratio R is converted to SpO2: (26); SpO2 is blood oxygen saturation, expressed as a percentage (%). A and B are empirical calibration coefficients, which can be obtained through equipment calibration.

[0123] The following further explains the steps of the cloud computing module 140 performing secondary sleep analysis. In some embodiments, secondary sleep analysis is performed based on the primary analysis results, the first raw data, and the second raw data to obtain secondary analysis results, including: Step 141: Use a moving average to smooth the head angular acceleration data in the current first raw data to obtain the average head angular acceleration result.

[0124] According to the moving average method of equation (2), the head angular acceleration data in the first original data is smoothed and filtered to obtain the average head angular acceleration result.

[0125] The specific steps will not be elaborated here. The smoothing and filtering results of the hand acceleration data can be directly obtained from step 1321.

[0126] Alternatively, instead of using the result obtained in step 1321, the hand acceleration data in the current second data can be smoothed and filtered according to the moving average method of equation (2) to obtain the average result of hand acceleration.

[0127] Step 142: Calculate the amplitude of head angular acceleration based on the average head angular acceleration result.

[0128] Referring to the calculation method of hand acceleration amplitude in step 1322, the head angular acceleration amplitude is calculated. That is, the calculation methods of the two are the same. The hand acceleration is the acceleration data of the three axes of x, y, and z, and the head angular acceleration data is the angular acceleration data of the x-axis, y-axis, and z-axis respectively. They are directly measured by the corresponding sensors. The amplitude calculation method is to calculate the sum of squares and then take the square root.

[0129] The amplitude of hand acceleration can be obtained directly from step 1321, or it can be recalculated.

[0130] Step 143: Based on the sleep stage results, the head angular acceleration amplitude, and the hand acceleration amplitude, determine the sleep preparation start time and wake-up time, as well as the start and end times of each stage of the sleep stage results.

[0131] Specifically, from the sleep staging results, the starting time of the first continuous low-motion (e.g., below the first angular velocity threshold) of the head angular acceleration amplitude and hand acceleration amplitude for a preset duration is extracted from the most recent complete sleep data before the first sleep stage (light sleep, deep sleep, REM sleep). This time is used as the sleep preparation start time. .

[0132] In one specific embodiment, the preset duration is at least 5 minutes. From the sleep staging results, the wake-up time is defined as the time when the amplitudes of head angular acceleration and hand acceleration first exceed a preset second angular velocity threshold after the last sleep stage, following the period of peak wakefulness. .

[0133] Based on the sleep stage results, the start and end times of each stage are extracted.

[0134] Step 144: Determine the time in bed based on the sleep start time and the wake-up time.

[0135] Time in bed for: (27).

[0136] Step 145: Determine the actual sleep time based on the start and end times of each sleep stage according to the sleep stage results.

[0137] Actual sleep time for: (28); in, , Indicates the first k The start and end times of each sleep stage (light sleep, deep sleep, REM sleep). This represents the number of sleep segments that are classified as light sleep, deep sleep, or REM sleep during a single use of the device in bed (analysis results for one night / multiple calculation cycles).

[0138] Step 146: Determine sleep efficiency based on the time spent in bed and the actual sleep time.

[0139] Based on TIB and TST, the sleep efficiency SE is calculated as follows: .

[0140] Step 147: Determine the percentage of sleep stages in each stage of the sleep staging results based on the start and end times of each stage and the actual sleep time.

[0141] Complete sleep data for each session is generated by K all (Different from) K sleep The specific differences are as follows K all This represents the number of results obtained during a single complete sleep phase using the device. K sleep The sequence consists of (the number of sleep stages in a single device use that are classified as light sleep, deep sleep, or REM sleep) and is composed of consecutive sleep state segments. k A segment corresponds to a stage marker. And has a corresponding start time. End time If a certain target stage is denoted as The cumulative duration of this stage is... The calculation is as follows: (30); in, This indicates the sleep state identifier for the k-th sleep segment (specifically: light sleep / deep sleep / REM). I{·} This represents an indicator function; if the condition within the parentheses is true, the value is 1; otherwise, it is 0.

[0142] but TST If the actual total sleep time is given, then the corresponding percentage of each stage is calculated. for: (31).

[0143] Step 148: Using a pre-set sliding window, perform dynamic trend analysis on the preset physiological parameters and detect abnormal data points based on the set threshold.

[0144] The sliding window method is used to perform dynamic trend analysis on preset physiological parameters (such as heart rate, respiratory rate, blood oxygen, etc.) and detect abnormal data points based on the set threshold.

[0145] Abnormal events include health risks such as nocturnal hypoxia, frequent awakenings, and abnormal heart rate fluctuations. Specifically, this includes: analyzing the trends of physiological signals such as heart rate, respiratory rate, and blood oxygenation; detecting abnormal fluctuations by calculating the rate of change of signals and comparing them with thresholds; and tracking the number of awakenings, hypoxia events, and heart rate fluctuations each night.

[0146] Furthermore, in some embodiments, dynamic trend analysis of preset physiological parameters and detection of abnormal data points based on set thresholds can also be performed in the local computing module 130.

[0147] The following is an example of detecting outlier data points: 1) Calculation of signal change rate: To detect mutations or fluctuations, the rate of change of physiological signals is calculated. : (32); in, Δt represents the signal value at a certain point in time (such as heart rate, blood oxygen, etc.), and Δt is the time sampling interval.

[0148] If the rate of change exceeds the set threshold, it is marked as an abnormal event.

[0149] 2) Anomaly detection: Preset physiological parameters (such as heart rate, respiratory rate, blood oxygen, etc.) are divided into fixed time windows, totaling K windows, and the mean and standard deviation within each window are calculated.

[0150] Based on threshold conditions, if a data point (such as heart rate, blood oxygen, etc.) exceeds a preset safety range, it is marked as abnormal. Specifically, this includes: Abnormal heart rate: If the heart rate is too high (>120 bpm) or too low (<40 bpm); Abnormal blood oxygenation: If blood oxygen levels are below 90%; Abnormal breathing: If the respiratory rate exceeds the normal range (e.g., <12 bpm or >20 bpm).

[0151] For each time window, mark the data points that exceed the threshold: (33); in, , These are the upper and lower limits of the safe threshold range for this physiological signal.

[0152] 3) Frequent arousal detection: Based on the results of sleep staging, if the number of awakenings throughout the night exceeds a set threshold, it is marked as a frequent awakening event.

[0153] Step 149: Based on the sleep efficiency, the proportion of sleep stages, the results of the dynamic trend analysis, and the abnormal data points, obtain the secondary analysis results.

[0154] The results of secondary analysis consist of sleep efficiency, sleep stage percentage, dynamic trend analysis, and outlier data points.

[0155] Step 134 is further explained below. In some embodiments, when the current body movement state is the first body movement state, sleep staging is calculated based on the EEG preprocessing data to obtain sleep staging results, including: Step 1341: When the current body movement state is the first body movement state, the EEG preprocessing data is input into the pre-trained EEG sleep staging deep learning model to obtain the sleep staging result.

[0156] When the current body movement state is determined to be the first body movement state, the amount of hand movement is relatively large, and the PPG signal quality may be significantly affected.

[0157] Sleep staging is calculated using only EEG data. The preprocessed EEG data obtained by windowing is input into the trained EEG sleep staging deep learning model for temporal pattern recognition, as shown in Equation (14): (14); in, This represents the preprocessed EEG data for the k-th window; A deep learning model for EEG sleep staging; This represents the probability distribution of sleep stages (wakefulness, REM sleep, light sleep, deep sleep) output by the model.

[0158] The maximum probability is the sleep stage result for the current time window.

[0159] It is understood that the deep learning model for EEG sleep staging is based on a deep neural network model and is trained using window EEG data samples labeled with sleep stages. This invention does not restrict the specific base model or training method, and can be selected according to the actual situation.

[0160] The following provides a further explanation of step 135. In some embodiments, when the current body movement state is a second body movement state, sleep staging is calculated based on the EEG preprocessing data and the PPG preprocessing data to obtain a sleep staging result, including: Step 1351: When the current body movement state is the second body movement state, the EEG preprocessing data is input into the pre-trained EEG sleep staging deep learning model to obtain the first recognition result.

[0161] When the current body movement state is determined to be the second body movement state, the hand movement is weak and the PPG signal quality is relatively stable. The EEG + PPG joint fusion model is used to realize multimodal sleep stage recognition.

[0162] Specifically, the preprocessed EEG data obtained through windowing is first input into the trained EEG sleep staging deep learning model for temporal pattern recognition, as shown in Equation (14), which will not be elaborated upon here. The result output by the model is recorded as the first recognition result.

[0163] Step 1352: Input the PPG preprocessed data into the pre-trained PPG sleep staging deep learning model to obtain the second recognition result.

[0164] In addition to using the above-mentioned EEG sleep staging deep learning model, a PPG sleep staging deep learning model is added, as shown in equation (15): (15); in, This represents the PPG preprocessed data for the k-th window; A deep learning model for PPG sleep staging; This represents the probability distribution of sleep stages (wakefulness, REM sleep, light sleep, deep sleep) output by the model.

[0165] The highest probability value is the second recognition result for the current time window.

[0166] It is understood that the PPG sleep staging deep learning model is based on a deep neural network model and is trained using PPG data samples labeled with sleep stages. This invention does not limit the specific base model or training method, and can be selected according to the actual situation.

[0167] Step 1353: Obtain the sleep staging result based on the first identification result and the second identification result.

[0168] The output of the two models , Weighted fusion was performed to obtain the sleep staging results. As shown in equation (16): (16); in, Weights for the deep learning model of EEG sleep staging. Weights for the PPG sleep staging deep learning model. The results of the merged sleep staging are denoted as sleep staging results.

[0169] The following describes a specific process for sleep monitoring based on the sleep monitoring system provided by this invention. In this embodiment, the local computing module 130 is located in the multi-functional charging compartment, the PPG acquisition module 120 is located in the smart ring, and the EEG acquisition module 110 is located in the EEG collector.

[0170] The first step involves the user connecting the multi-functional charging case to their home Wi-Fi network via their mobile phone or tablet before going to sleep. This completes the device's network initialization (only the first use requires a setup process), and ensures that both the EEG acquisition module 110 and the PPG acquisition module 120 are operational. Once connected, all acquisition devices establish stable communication with the multi-functional charging case via Bluetooth, preparing for the data acquisition process.

[0171] The second step involves the user wearing the EEG acquisition device and smart ring while asleep, at which point the system initiates real-time data acquisition. The EEG acquisition module 110 continuously collects raw EEG data, while simultaneously recording the user's head posture and micro-movements via an angular accelerometer and angle sensor, obtaining head angular acceleration data, which constitutes the first set of raw data. The smart ring dynamically determines whether to initiate the acquisition of physiological data such as heart rate, blood oxygen, and respiratory rate based on acceleration thresholds, ensuring that PPG signals are stable before recording, thus obtaining raw PPG data and hand acceleration data, which serve as the second set of raw data.

[0172] The third step involves the multi-functional charging case acting as a local edge computing node, receiving and caching data from the EEG collector and the smart ring, and performing first-level analysis. EEG signals are used for real-time sleep stage determination, while PPG signals are used to assist in sleep staging and generate physiological indicators such as heart rate, respiration, and blood oxygenation. The system employs a dynamic fusion decision mechanism: sleep staging is determined at fixed intervals, and a single sleep staging result may represent one or more minimum sleep cycles. When the PPG signal quality is good and hand movement is minimal, the current body movement state is determined to be the second body movement state, and the EEG + PPG joint model is used for sleep staging; otherwise, the current body movement state is determined to be the first body movement state, and only EEG signals are used for sleep staging.

[0173] In the fourth step, after the first-level analysis is completed, the multi-functional charging compartment, while connected to the network, uploads the first-level processing results and the first and second raw data to the cloud computing module 140 in the remote computing and data storage server via the Wi-Fi module.

[0174] The fifth step involves a remote computing and data storage server serving as the second-level computing platform. This server performs in-depth processing and modeling of the uploaded data, completing the second-level computing tasks. Specifically, this includes calculating sleep efficiency, the proportion of each sleep stage, the overall physiological curve trend throughout the night, detecting abnormal physiological events (such as hypoxia and frequent awakenings), and archiving the data.

[0175] Step 6: After the server analysis is complete, the results are returned to the user interaction module 150 via API for user viewing. Users can view the text and image report on their mobile phones or PCs, including a complete sleep stage chart, physiological indicator curves, scoring suggestions, and abnormal alerts. It also supports selecting any date to compare and analyze historical records, enabling personalized health management.

[0176] The sleep monitoring system provided by this invention achieves high-precision sleep analysis in non-medical settings through the organic cooperation of its various modules. This effectively solves the problems of traditional devices being bulky, complex to operate, and requiring professional intervention, while significantly improving user convenience and data accuracy. The modular structure of the system also facilitates future expansion and maintenance. By rationally allocating computing tasks among multiple levels of devices, the computational burden on individual devices can be effectively reduced, extending device battery life and improving the overall system efficiency and stability.

[0177] The sleep monitoring method provided by the present invention is described below. The sleep monitoring method described below can be referred to in correspondence with the sleep monitoring system described above. Figure 4 This is a schematic diagram of the sleep monitoring method provided by the present invention, as shown below. Figure 4 As shown, the method includes the following steps: Step 410: Based on the EEG acquisition module, collect raw EEG data and head angular acceleration data during the user's sleep process as the first raw data; Step 420: Based on the PPG acquisition module, collect the raw PPG data and hand acceleration data during the user's sleep process as the second raw data; Step 430: Based on the local computing module, perform a first-level sleep analysis according to the first raw data and the second raw data to obtain the first-level analysis results. The first-level sleep analysis includes sleep stage calculation and physiological parameter calculation. Step 440: Based on the cloud computing module, perform secondary sleep analysis according to the primary analysis results, the first raw data and the second raw data to obtain secondary analysis results. The secondary sleep analysis includes sleep efficiency calculation, sleep stage ratio calculation, physiological signal dynamic trend analysis and abnormal data point detection. Step 450: Based on the user interaction module, visualize the primary analysis results and / or the secondary analysis results for the user to view.

[0178] According to a sleep monitoring method provided by the present invention, the electroencephalogram (EEG) acquisition module is worn on a preset area of ​​the user's head during the acquisition process. The EEG acquisition module includes an EEG amplifier, electrode patches, and an angular acceleration measuring device. The EEG amplifier and the electrode patches are connected by magnetic coupling and metal contact positioning; and / or The PPG acquisition module is installed in a wearable portable device and is worn on a preset part of the user's body during the acquisition process. The PPG acquisition module includes a PPG sensor and a hand acceleration measurement device.

[0179] According to a sleep monitoring method provided by the present invention, the electrode patch is made of a skin-friendly adhesive material.

[0180] According to a sleep monitoring method provided by the present invention, a primary sleep analysis is performed based on the first raw data and the second raw data to obtain the primary analysis results, including: Using a preset analysis window, extract the first raw data and the second raw data that have been added between the last processing completion time and the current processing time, and use them as the current first raw data and the current second raw data. Based on the hand acceleration data in the current second data, determine the current body movement state; Preprocessing is performed on the EEG raw data in the current first raw data and the PPG raw data in the current second raw data to obtain EEG preprocessed data and PPG preprocessed data. When the current body movement state is the first body movement state, sleep stage calculation is performed based on the EEG preprocessing data to obtain the sleep stage result; When the current body movement state is the second body movement state, sleep staging is calculated based on the EEG preprocessing data and the PPG preprocessing data to obtain the sleep staging result; Based on the PPG raw data in the current second raw data, preset physiological parameters are calculated, wherein the preset physiological parameters include heart rate, heart rate variability, respiratory rate and / or blood oxygen saturation; Based on the sleep stage results and the preset physiological parameters, the first-level analysis results are obtained.

[0181] According to a sleep monitoring method provided by the present invention, determining the current body movement state based on hand acceleration data in the current second data includes: The hand acceleration data in the current second data is smoothed using a moving average to obtain an average hand acceleration result; Calculate the amplitude of hand acceleration based on the average hand acceleration result; If the maximum value of the hand acceleration amplitude is greater than a preset hand acceleration threshold, the current body movement state is determined to be the first body movement state. If the maximum value of the hand acceleration amplitude is not greater than the preset hand acceleration threshold, the current body movement state is determined to be the second body movement state. According to a sleep monitoring method provided by the present invention, the EEG raw data in the current first raw data and the PPG raw data in the current second raw data are preprocessed to obtain EEG preprocessed data and PPG preprocessed data, including: The EEG raw data in the current first raw data is filtered, and the filtered EEG raw data in the current first raw data is standardized to obtain standardized EEG data. The standardized EEG data is segmented according to a preset segmentation window length to obtain preprocessed EEG data. The PPG raw data in the current second raw data is divided into DC component and AC component; The AC component is bandpass filtered to preserve the heart rate-related frequency band, thus obtaining the AC filtering result; The AC filtering results are then subjected to data standardization processing to obtain standardized PPG data; The standardized PPG data is segmented according to the preset segmentation window length to obtain PPG preprocessed data.

[0182] According to a sleep monitoring method provided by the present invention, a secondary sleep analysis is performed based on the primary analysis results, the first raw data, and the second raw data to obtain secondary analysis results, including: The head angular acceleration data in the current first raw data is smoothed using a moving average to obtain the average head angular acceleration result. Calculate the amplitude of head angular acceleration based on the average head angular acceleration result; Based on the sleep stage results, the head angular acceleration amplitude, and the hand acceleration amplitude, determine the sleep preparation start time and wake-up time, as well as the start and end times of each stage of the sleep stage results; The time spent in bed is determined based on the sleep start time and the wake-up time; The actual sleep time is determined based on the start and end times of each sleep stage as described in the sleep stage results. Sleep efficiency is determined based on the time spent in bed and the actual sleep time. Based on the start and end times of each stage of the sleep staging results and the actual sleep time, the percentage of each sleep stage in the sleep staging results is determined. Using a pre-set sliding window, dynamic trend analysis is performed on the preset physiological parameters, and abnormal data points are detected based on a set threshold. The secondary analysis results are obtained based on the sleep efficiency, the proportion of sleep stages, the results of the dynamic trend analysis, and the abnormal data points.

[0183] According to a sleep monitoring method provided by the present invention, when the current body movement state is a first body movement state, sleep staging is calculated based on the electroencephalogram (EEG) preprocessing data to obtain sleep staging results, including: When the current body movement state is the first body movement state, the EEG preprocessing data is input into the pre-trained EEG sleep staging deep learning model to obtain the sleep staging result.

[0184] According to a sleep monitoring method provided by the present invention, when the current body movement state is a second body movement state, sleep staging is calculated based on the EEG preprocessing data and the PPG preprocessing data to obtain a sleep staging result, including: When the current body movement state is the second body movement state, the EEG preprocessing data is input into the pre-trained EEG sleep staging deep learning model to obtain the first recognition result; The PPG preprocessed data is input into a pre-trained PPG sleep staging deep learning model to obtain a second identification result; Based on the first identification result and the second identification result, the sleep staging result is obtained.

[0185] Figure 5 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 5As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other through the communication bus 540. The processor 510 can call logic instructions in the memory 530 to execute a sleep monitoring method, which includes: acquiring raw EEG data and head angular acceleration data during the user's sleep process based on an EEG acquisition module, as first raw data; acquiring raw PPG data and hand acceleration data during the user's sleep process based on a PPG acquisition module, as second raw data; performing a first-level sleep analysis based on the first and second raw data using a local computing module, to obtain a first-level analysis result, wherein the first-level sleep analysis includes sleep stage calculation and physiological parameter calculation; performing a second-level sleep analysis based on the first-level analysis result, the first raw data, and the second raw data using a cloud computing module, to obtain a second-level analysis result, wherein the second-level sleep analysis includes sleep efficiency calculation, sleep stage percentage calculation, physiological signal dynamic trend analysis, and abnormal data point detection; and visualizing the first-level analysis result and / or the second-level analysis result for the user to view using a user interaction module.

[0186] Furthermore, the logical instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0187] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the sleep monitoring method provided by the above methods. The method includes: acquiring raw EEG data and head angular acceleration data during the user's sleep process based on an EEG acquisition module as first raw data; acquiring raw PPG data and hand acceleration data during the user's sleep process based on a PPG acquisition module as second raw data; performing a first-level sleep analysis based on a local computing module according to the first raw data and the second raw data to obtain a first-level analysis result, wherein the first-level sleep analysis includes sleep stage calculation and physiological parameter calculation; performing a second-level sleep analysis based on a cloud computing module according to the first-level analysis result, the first raw data, and the second raw data to obtain a second-level analysis result, wherein the second-level sleep analysis includes sleep efficiency calculation, sleep stage percentage calculation, physiological signal dynamic trend analysis, and abnormal data point detection; and visualizing the first-level analysis result and / or the second-level analysis result for the user to view based on a user interaction module.

[0188] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the sleep monitoring method provided by the above-described methods. This method includes: acquiring raw EEG data and head angular acceleration data during a user's sleep process using an EEG acquisition module as first raw data; acquiring raw PPG data and hand acceleration data during the user's sleep process using a PPG acquisition module as second raw data; performing a first-level sleep analysis based on the first and second raw data using a local computing module to obtain a first-level analysis result, wherein the first-level sleep analysis includes sleep stage calculation and physiological parameter calculation; performing a second-level sleep analysis based on the first-level analysis result, the first raw data, and the second raw data using a cloud computing module to obtain a second-level sleep analysis result, wherein the second-level sleep analysis includes sleep efficiency calculation, sleep stage percentage calculation, dynamic trend analysis of physiological signals, and abnormal data point detection; and visualizing the first-level analysis result and / or the second-level analysis result using a user interaction module for user viewing.

[0189] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0190] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0191] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A sleep monitoring system, characterized in that, include: The EEG acquisition module is used to collect raw EEG data and head angular acceleration data during the user's sleep process as the first raw data; The PPG acquisition module is used to collect raw PPG data and hand acceleration data during the user's sleep process as the second raw data. The local computing module is used to perform a first-level sleep analysis based on the first raw data and the second raw data to obtain the first-level analysis results. The first-level sleep analysis includes sleep stage calculation and physiological parameter calculation. The cloud computing module is used to perform secondary sleep analysis based on the primary analysis results, the first raw data, and the second raw data to obtain secondary analysis results. The secondary sleep analysis includes sleep efficiency calculation, sleep stage percentage calculation, physiological signal dynamic trend analysis, and abnormal data point detection. The user interaction module is used to visualize the primary analysis results and / or the secondary analysis results for users to view.

2. The sleep monitoring system according to claim 1, characterized in that, During the data acquisition process, the EEG acquisition module is worn on a preset area of ​​the user's head. The EEG acquisition module includes an EEG amplifier, electrode patches, and an angular acceleration measurement device. The EEG amplifier and the electrode patches are connected via magnetic coupling and metal contact positioning; and / or The PPG acquisition module is installed in a wearable portable device and is worn on a preset part of the user's body during the acquisition process. The PPG acquisition module includes a PPG sensor and a hand acceleration measurement device.

3. The sleep monitoring system according to claim 2, characterized in that, The electrode pads are made of a skin-friendly adhesive material.

4. The sleep monitoring system according to claim 2, characterized in that, Based on the first raw data and the second raw data, a primary sleep analysis is performed to obtain the primary analysis results, including: Using a preset analysis window, extract the first raw data and the second raw data that have been added between the last processing completion time and the current processing time, and use them as the current first raw data and the current second raw data. Based on the hand acceleration data in the current second data, determine the current body movement state; Preprocessing is performed on the EEG raw data in the current first raw data and the PPG raw data in the current second raw data to obtain EEG preprocessed data and PPG preprocessed data. When the current body movement state is the first body movement state, sleep stage calculation is performed based on the EEG preprocessing data to obtain the sleep stage result; When the current body movement state is the second body movement state, sleep staging is calculated based on the EEG preprocessing data and the PPG preprocessing data to obtain the sleep staging result; Based on the PPG raw data in the current second raw data, preset physiological parameters are calculated, wherein the preset physiological parameters include heart rate, heart rate variability, respiratory rate and / or blood oxygen saturation; Based on the sleep stage results and the preset physiological parameters, the first-level analysis results are obtained.

5. The sleep monitoring system according to claim 4, characterized in that, Based on the hand acceleration data in the current second data, the current body movement state is determined, including: The hand acceleration data in the current second data is smoothed using a moving average to obtain an average hand acceleration result; Calculate the amplitude of hand acceleration based on the average hand acceleration result; If the maximum value of the hand acceleration amplitude is greater than a preset hand acceleration threshold, the current body movement state is determined to be the first body movement state. If the maximum value of the hand acceleration amplitude is not greater than the preset hand acceleration threshold, the current body movement state is determined to be the second body movement state.

6. The sleep monitoring system according to claim 4, characterized in that, Preprocessing is performed on the EEG raw data in the current first raw data and the PPG raw data in the current second raw data to obtain EEG preprocessed data and PPG preprocessed data, including: The EEG raw data in the current first raw data is filtered, and the filtered EEG raw data in the current first raw data is standardized to obtain standardized EEG data. The standardized EEG data is segmented according to a preset segmentation window length to obtain preprocessed EEG data. The PPG raw data in the current second raw data is divided into DC component and AC component; The AC component is bandpass filtered to preserve the heart rate-related frequency band, thus obtaining the AC filtering result; The AC filtering results are then subjected to data standardization processing to obtain standardized PPG data; The standardized PPG data is segmented according to the preset segmentation window length to obtain PPG preprocessed data.

7. The sleep monitoring system according to claim 5, characterized in that, Based on the primary analysis results, the first raw data, and the second raw data, a secondary sleep analysis is performed to obtain the secondary analysis results, including: The head angular acceleration data in the current first raw data is smoothed using a moving average to obtain the average head angular acceleration result. Calculate the amplitude of head angular acceleration based on the average head angular acceleration result; Based on the sleep stage results, the head angular acceleration amplitude, and the hand acceleration amplitude, determine the sleep preparation start time and wake-up time, as well as the start and end times of each stage of the sleep stage results; The time spent in bed is determined based on the sleep start time and the wake-up time; The actual sleep time is determined based on the start and end times of each sleep stage as described in the sleep stage results. Sleep efficiency is determined based on the time spent in bed and the actual sleep time. Based on the start and end times of each stage of the sleep staging results and the actual sleep time, the percentage of each sleep stage in the sleep staging results is determined. Using a pre-set sliding window, dynamic trend analysis is performed on the preset physiological parameters, and abnormal data points are detected based on a set threshold. The secondary analysis results are obtained based on the sleep efficiency, the proportion of sleep stages, the results of the dynamic trend analysis, and the abnormal data points.

8. The sleep monitoring system according to claim 4 or 5, characterized in that, When the current body movement state is the first body movement state, sleep staging is calculated based on the preprocessed EEG data to obtain sleep staging results, including: When the current body movement state is the first body movement state, the EEG preprocessing data is input into the pre-trained EEG sleep staging deep learning model to obtain the sleep staging result.

9. The sleep monitoring system according to claim 4 or 5, characterized in that, When the current body movement state is the second body movement state, sleep staging is calculated based on the EEG preprocessing data and the PPG preprocessing data to obtain sleep staging results, including: When the current body movement state is the second body movement state, the EEG preprocessing data is input into the pre-trained EEG sleep staging deep learning model to obtain the first recognition result; The PPG preprocessed data is input into a pre-trained PPG sleep staging deep learning model to obtain a second identification result; Based on the first identification result and the second identification result, the sleep staging result is obtained.

10. A sleep monitoring method, characterized in that, include: Based on the EEG acquisition module, raw EEG data and head angular acceleration data during the user's sleep process are collected as the first raw data; Based on the PPG acquisition module, PPG raw data and hand acceleration data during the user's sleep process are collected as the second raw data; Based on the local computing module, a first-level sleep analysis is performed according to the first raw data and the second raw data to obtain the first-level analysis results. The first-level sleep analysis includes sleep stage calculation and physiological parameter calculation. Based on the cloud computing module, a second-level sleep analysis is performed according to the first-level analysis results, the first raw data, and the second raw data to obtain the second-level analysis results. The second-level sleep analysis includes sleep efficiency calculation, sleep stage percentage calculation, dynamic trend analysis of physiological signals, and abnormal data point detection. Based on the user interaction module, the primary analysis results and / or the secondary analysis results are visualized for users to view.

Images