Sleep management method for analyzing sleep record data

Abstract

The invention discloses a sleep management method for analyzing sleep record data, and relates to the technical field of sleep management.The method comprises the steps that monitoring data of a sleep monitor for conducting sleep monitoring on a target user in a preset historical window is called, and L historical sleep monitoring data logs of L historical monitoring nodes are obtained; obtaining L historical sleep monitoring index groups; performing clustering analysis to obtain Q clustering historical sleep monitoring index group sets; performing centralized screening on the sleep monitoring index groups to obtain Q clustered historical sleep monitoring index centralized groups; and analyzing the Q clustered historical sleep monitoring index concentrated groups by using the sleep management scheme identification network layer to obtain a target sleep management scheme. The technical problems that in the prior art, the fitness between sleep management and the actual sleep condition of the user is low, and the reliability of sleep management is low are solved. The technical effects of improving the data analysis efficiency and improving the sleep management quality are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of sleep management, and in particular to a sleep management method for analyzing sleep record data. Background Art

[0002] With the development of intelligent health management, sleep monitors and other devices have been widely used to record and analyze sleep data. These devices can collect multi-dimensional data such as sleep duration, deep sleep ratio, number of awakenings, and sleep latency, helping users understand their sleep status. However, most current sleep monitoring methods are mainly based on single-point or short-term data analysis, ignoring the potential patterns of historical data within a continuous time range, resulting in low reliability of sleep management. Summary of the invention

[0003] The present application provides a sleep management method for analyzing sleep record data, which is used to solve the technical problems in the prior art that sleep management has a low fit with the user's actual sleep situation and has low reliability of sleep management.

[0004] In view of the above problems, the present application provides a sleep management method for analyzing sleep record data, the method comprising:

[0005] Retrieving monitoring data of a sleep monitor that monitors the sleep of a target user within a preset historical window, and obtaining L historical sleep monitoring data logs of L historical monitoring nodes, where L is a positive integer;

[0006] Based on the preset sleep monitoring indicator groups, L historical sleep monitoring data logs are extracted respectively to obtain L historical sleep monitoring indicator groups;

[0007] Performing cluster analysis on the L historical sleep monitoring indicator groups of the L historical monitoring nodes from two dimensions: the similarity of the historical monitoring nodes and the similarity of the historical sleep monitoring indicators, to obtain Q clustered historical sleep monitoring indicator group sets, where Q is a positive integer less than or equal to L;

[0008] Traversing the Q clustered historical sleep monitoring indicator group sets to perform centralized screening of the sleep monitoring indicator groups, and obtaining Q clustered historical sleep monitoring indicator centralized groups;

[0009] The sleep management solution identification network layer is used to analyze the Q clustered historical sleep monitoring indicator concentrated groups to obtain a target sleep management solution.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0011] This application obtains L historical sleep monitoring data logs of L historical monitoring nodes by calling the monitoring data of the sleep monitor that monitors the sleep of the target user within the preset historical window, where L is a positive integer, and then extracts the L historical sleep monitoring data logs based on the preset sleep monitoring indicator group to obtain L historical sleep monitoring indicator groups, and then clusters the L historical sleep monitoring indicator groups of L historical monitoring nodes from the two dimensions of the similarity of the historical monitoring nodes and the similarity of the historical sleep monitoring indicators to obtain Q clustered historical sleep monitoring indicator group sets, where Q is a positive integer less than or equal to L, and then traverses the Q clustered historical sleep monitoring indicator group sets to perform centralized screening of the sleep monitoring indicator groups to obtain Q clustered historical sleep monitoring indicator centralized groups, and uses the sleep management plan identification network layer to analyze the Q clustered historical sleep monitoring indicator centralized groups to obtain the target sleep management plan. The technical effect of in-depth analysis of sleep record data and improving the reliability of sleep management is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Attached Figure 1 The present invention provides a sleep management method for analyzing sleep record data.

[0013] Attached Figure 2 It is a schematic diagram of a process for obtaining Q clustered historical sleep monitoring indicator concentration groups in a sleep management method for analyzing sleep record data provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0014] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0015] It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or inherent to these processes, methods, products or devices.

[0016] Embodiment, as attached Figure 1 As shown, the present application provides a sleep management method for analyzing sleep record data, wherein the method comprises:

[0017] S1: Retrieve monitoring data of a sleep monitor that monitors the sleep of a target user within a preset historical window, and obtain L historical sleep monitoring data logs of L historical monitoring nodes, where L is a positive integer;

[0018] In a possible embodiment, the target user is any user who needs sleep management. The sleep monitor is a device that records the sleep data of the target user. The preset history window is a time period for sleep record data analysis pre-set by a person skilled in the art. The L historical monitoring nodes are discrete time points for monitoring the target user within the preset history window. The L historical sleep monitoring data logs are detailed sleep records stored on each monitoring node, including multiple data (such as heart rate, respiratory rate, etc.). Among them, the historical monitoring nodes and the historical sleep monitoring data logs correspond one to one.

[0019] Optionally, the monitoring data of the target user in the preset historical window is retrieved from the sleep monitor and sorted into L independent historical monitoring nodes in chronological order. The monitoring data of each node is stored in the form of a log, containing raw data of multiple dimensions. By clarifying the historical monitoring window and monitoring node, it is ensured that the extracted data has a sufficient time span and retains detailed time distribution information, achieving the technical effect of laying the data foundation for subsequent cluster analysis.

[0020] S2: extracting L historical sleep monitoring data logs based on the preset sleep monitoring indicator groups to obtain L historical sleep monitoring indicator groups;

[0021] Furthermore, the preset sleep monitoring indicator group includes deep sleep duration, number of awakenings, sleep onset latency and total sleep duration.

[0022] In one embodiment, the preset sleep monitoring index group is a set of indicators for sleep quality assessment predefined by those skilled in the art, including deep sleep duration, number of awakenings, sleep latency and total sleep duration. Among them, deep sleep duration is the cumulative duration of the deep sleep stage (usually related to physical recovery). The number of awakenings is the number of times the user wakes up at night, reflecting the continuity of sleep. The sleep latency is the time from going to bed to entering stable sleep, which measures the efficiency of falling asleep. The total sleep duration is the total sleep time for the whole night, which is a basic indicator for evaluating sleep health.

[0023] Optionally, key information is extracted from L historical sleep monitoring data logs in turn according to preset sleep monitoring indicator groups (such as deep sleep duration, number of awakenings, etc.), and each group of extracted indicators is organized into a standardized historical sleep monitoring indicator group. Exemplarily, historical sleep monitoring indicator group 1 corresponding to log 1 is: {deep sleep duration = 90 minutes, number of awakenings = 3, sleep latency = 20 minutes, total sleep duration = 7 hours}; historical sleep monitoring indicator group 2 corresponding to log 2 is: {deep sleep duration = 100 minutes, number of awakenings = 2, sleep latency = 15 minutes, total sleep duration = 7.5 hours}.

[0024] By extracting key information from complex raw data, reducing data dimensions and focusing on core indicators of sleep quality, we achieve a technical effect that facilitates subsequent clustering analysis and generation of management plans.

[0025] S3: performing cluster analysis on the L historical sleep monitoring indicator groups of the L historical monitoring nodes from two dimensions: the similarity of the historical monitoring nodes and the similarity of the historical sleep monitoring indicators, to obtain Q clustered historical sleep monitoring indicator group sets, where Q is a positive integer less than or equal to L;

[0026] Further, cluster analysis is performed on the L historical sleep monitoring indicator groups of the L historical monitoring nodes from two dimensions: the similarity of the historical monitoring nodes and the similarity of the historical sleep monitoring indicators, to obtain Q clustered historical sleep monitoring indicator group sets. Step S3 of the embodiment of the present application also includes:

[0027] Perform Q extractions from the L historical sleep monitoring indicator groups in a random manner without replacement to obtain Q initial cluster centers;

[0028] Clustering the L historical sleep monitoring indicator groups based on the Q initial cluster centers to obtain Q initial cluster historical sleep monitoring indicator group sets;

[0029] Using the clustering cost loss function, the overall clustering cost loss of the Q initial clustering historical sleep monitoring indicator group sets is calculated from two dimensions: the similarity of historical monitoring nodes and the similarity of historical sleep monitoring indicators, to obtain the initial clustering cost loss amount;

[0030] Calculating the means of the Q initial clustered historical sleep monitoring indicator group sets respectively to obtain the means of the Q initial clustered historical sleep monitoring indicator groups, and updating the Q initial cluster centers according to the means of the Q initial clustered historical sleep monitoring indicator groups to obtain Q updated cluster centers;

[0031] Combining the L historical sleep monitoring indicator groups, constructing Q updated cluster historical sleep monitoring indicator group sets of the Q updated cluster centers, and using the cluster cost loss function to perform overall cluster cost loss analysis to obtain an updated cluster cost loss amount;

[0032] Determine whether the updated cluster cost loss is less than or equal to the initial cluster cost loss. If so, calculate the means of the Q updated cluster historical sleep monitoring indicator group sets respectively, and update the Q updated cluster centers according to the calculation results until the difference between the two updated cluster cost losses of two adjacent updates is less than a preset difference. Cluster the L historical sleep monitoring indicator groups according to the Q updated cluster centers obtained from the last update to obtain Q cluster historical sleep monitoring indicator group sets.

[0033] Furthermore, step S3 of the embodiment of the present application also includes:

[0034] Construct a clustering cost loss function, where the clustering cost loss function is:

[0035]

[0036] Among them, LOSS is the initial clustering cost loss, n i is the initial cluster historical sleep monitoring indicator group set of the i-th initial cluster center among the Q initial cluster centers, is the initial cluster historical sleep monitoring indicator group of the i-th initial cluster center, μ ij is the jth historical sleep monitoring indicator group in the initial cluster historical sleep monitoring indicator group set of the i-th initial cluster center, is the historical monitoring node corresponding to the initial cluster historical sleep monitoring indicator group of the i-th initial cluster center, t ij is the historical monitoring node corresponding to the jth historical sleep monitoring indicator group in the set of historical sleep monitoring indicator groups of the initial cluster of the i-th initial cluster center, and λ is the weight for balancing the similarity between the historical monitoring nodes and the similarity between the historical sleep monitoring indicators.

[0037] In one possible embodiment, the degree of similarity of historical monitoring nodes refers to the time interval between the sleep data of different historical monitoring nodes. Nodes with high similarity indicate that they are adjacent in time and are suitable for joint analysis. The degree of similarity of historical sleep monitoring indicators is based on the feature similarity of sleep indicators (such as deep sleep duration, number of awake times, etc.), which indicates the degree of proximity of data content. Cluster analysis is performed on the L historical sleep monitoring indicator groups of L historical monitoring nodes from two dimensions to obtain a set of Q clustered historical sleep monitoring indicator groups. Among them, the set of Q clustered historical sleep monitoring indicator groups is the result of cluster analysis, and each group contains one or more historical sleep monitoring indicator groups with similar characteristics. By performing cluster analysis from two dimensions, the technical effect of improving the reliability of clustering L historical sleep monitoring indicator groups and the quality of cluster analysis results is achieved.

[0038] In one embodiment, random extraction without replacement is a random sampling method, which extracts Q different samples from L historical sleep monitoring indicator groups as initial cluster centers to ensure that there is no duplication between samples. Randomly select Q samples from L historical indicator groups as initial cluster centers without replacement to obtain Q initial cluster centers, each center corresponding to a cluster. For example, the distances between the L historical sleep monitoring indicator groups and the Q initial cluster centers are calculated (such as based on Euclidean distance or cosine similarity), and the L historical sleep monitoring indicator groups are assigned to the set to which the initial cluster center closest to them belongs, to obtain Q initial cluster historical sleep monitoring indicator group sets. Each initial cluster historical sleep monitoring indicator group set contains one or more historical sleep monitoring indicator groups.

[0039] The Q initial cluster historical sleep monitoring indicator group sets and the Q initial cluster centers are input into the cluster cost loss function to perform an overall cost loss analysis, and the initial cluster cost loss amount is obtained. By using the two dimensions of the similarity of historical monitoring nodes and the similarity of historical sleep monitoring indicators, λ is used in the cluster cost loss function to balance the weights of the two when performing loss analysis. The technical effect of improving the quality and reliability of the overall cost loss analysis is achieved.

[0040] Optionally, the means of the Q initial clustered historical sleep monitoring indicator group sets are calculated respectively to obtain the means of the Q initial clustered historical sleep monitoring indicator group, and the Q initial cluster centers are updated according to the means of the Q initial clustered historical sleep monitoring indicator group to obtain Q updated cluster centers. Then, it is analyzed whether the Q updated cluster centers can better represent the concentration of the Q initial clustered historical sleep monitoring indicator group sets.

[0041] Preferably, the L historical sleep monitoring indicator groups are clustered again based on the Q updated cluster centers to obtain Q updated cluster historical sleep monitoring indicator group sets. The accuracy of clustering is further improved. The cluster cost loss function is used to perform overall cluster cost loss analysis to obtain the updated cluster cost loss amount. The updated cluster cost loss amount reflects the overall cost loss of clustering based on the Q updated cluster centers.

[0042] Further, it is determined whether the updated cluster cost loss is less than or equal to the initial cluster cost loss. If so, the means of the Q updated cluster historical sleep monitoring indicator group sets are calculated respectively, and the Q updated cluster centers are updated according to the calculation results, until the difference between the two updated cluster cost losses of two adjacent updates is less than the preset difference, indicating that the clustering result has reached a relatively reliable level, and the L historical sleep monitoring indicator groups are clustered according to the Q updated cluster centers obtained from the last update, to obtain Q cluster historical sleep monitoring indicator group sets. The preliminary grouping of data is achieved, providing an important basis for subsequent optimization and analysis.

[0043] S4: traversing the Q clustered historical sleep monitoring indicator group sets to perform centralized screening of the sleep monitoring indicator groups, and obtaining Q clustered historical sleep monitoring indicator centralized groups;

[0044] Further, as attached Figure 2 As shown, the Q clustered historical sleep monitoring indicator group sets are traversed to perform centralized screening of the sleep monitoring indicator groups to obtain Q clustered historical sleep monitoring indicator centralized groups. Step S4 of the embodiment of the present application further includes:

[0045] Calculate the means of the Q clustered historical sleep monitoring indicator group sets respectively to obtain Q clustered historical sleep monitoring indicator mean groups;

[0046] Taking the Q clustered historical sleep monitoring indicator mean groups as Q starting centralized screening points, constructing Q starting centralized screening point neighborhoods in the Q clustered historical sleep monitoring indicator group sets according to a preset centralized screening similarity threshold;

[0047] The Q clustered historical sleep monitoring indicator group sets are subjected to sleep monitoring indicator group centralized screening based on the Q starting centralized screening points and the Q starting centralized screening point neighborhoods to obtain Q clustered historical sleep monitoring indicator centralized groups.

[0048] Further, based on the Q starting concentrated screening points and the Q starting concentrated screening point neighborhoods, the Q clustered historical sleep monitoring indicator group sets are subjected to sleep monitoring indicator group concentrated screening to obtain Q clustered historical sleep monitoring indicator concentrated groups. Step S4 of the embodiment of the present application further includes:

[0049] Randomly extracting Q clustered historical sleep monitoring indicator groups from the edges of the neighborhoods of the screening points in the Q starting sets as Q iterative clustered historical sleep monitoring groups;

[0050] Constructing Q iterative clustering historical sleep monitoring group neighborhoods of the Q iterative clustering historical sleep monitoring groups according to a preset centralized screening similarity threshold;

[0051] Determine whether the neighborhood density of the Q iterative clustered historical sleep monitoring groups' neighborhoods is greater than or equal to the neighborhood density of the Q starting concentrated screening point neighborhoods. If so, use the Q iterative clustered historical sleep monitoring groups as Q iterative concentrated screening points, and use the Q iterative clustered historical sleep monitoring groups' neighborhoods as Q iterative concentrated screening point neighborhoods. Based on the Q iterative concentrated screening points and the Q iterative concentrated screening point neighborhoods, continue to perform sleep monitoring indicator group concentrated screening on the Q clustered historical sleep monitoring indicator group sets until the preset maximum screening times are met, and use the Q iterative concentrated screening points obtained from the last screening as the Q clustered historical sleep monitoring indicator concentrated groups.

[0052] Further, if not, Q clustered historical sleep monitoring indicator groups are randomly extracted again from the edges of the neighborhoods of the screening points in the Q starting sets, and the Q iterative clustered historical sleep monitoring groups are updated.

[0053] In one embodiment, after obtaining the Q clustered historical sleep monitoring indicator group sets, in order to improve the reliability of the subsequent sleep management plan, it is necessary to analyze the general situation of each cluster to provide a reliable basis for the generation of the sleep management plan. Preferably, the Q clustered historical sleep monitoring indicator group sets are obtained by performing sleep monitoring indicator group centralized screening on the Q clustered historical sleep monitoring indicator group sets. Among them, the Q clustered historical sleep monitoring indicator group is the data that best represents the general situation of the Q clustered historical sleep monitoring indicator group sets.

[0054] The means of the Q clustered historical sleep monitoring indicator group sets are calculated respectively to obtain Q clustered historical sleep monitoring indicator mean value groups, wherein the Q clustered historical sleep monitoring indicator mean value groups reflect the average level of the Q clustered historical sleep monitoring indicator group sets.

[0055] Furthermore, the Q clustered historical sleep monitoring indicator mean groups are used as Q starting centralized screening points, and Q starting centralized screening point neighborhoods are constructed in the Q clustered historical sleep monitoring indicator group sets according to the preset centralized screening similarity threshold. In other words, the Q clustered historical sleep monitoring indicator group sets are respectively similarly calculated with the corresponding Q clustered historical sleep monitoring indicator mean groups using the cosine similarity formula, and the clustered historical sleep monitoring indicator groups whose calculation results are greater than the preset centralized screening similarity threshold (the minimum similarity that can be included in the neighborhood pre-set by those skilled in the art) are added to the Q starting centralized screening point neighborhoods.

[0056] Furthermore, Q clustered historical sleep monitoring indicator groups are randomly extracted from the edges of the neighborhoods of the Q starting concentrated screening points as Q iterative clustered historical sleep monitoring groups. Based on the same construction principle as the neighborhoods of the Q starting concentrated screening points, Q iterative clustered historical sleep monitoring group neighborhoods of the Q iterative clustered historical sleep monitoring groups are constructed according to the preset concentrated screening similarity threshold. The neighborhoods of the Q iterative clustered historical sleep monitoring groups contain data whose similarity to the Q iterative clustered historical sleep monitoring groups is within the preset concentrated screening similarity threshold.

[0057] Furthermore, the number of clustered historical sleep monitoring indicator groups in the neighborhood of the Q starting concentrated screening points is counted, and the calculated result is compared with the preset concentrated screening similarity threshold of 2 times to obtain the neighborhood density of the neighborhood of the Q starting concentrated screening points. Based on the same calculation principle, the neighborhood density of the neighborhood of the Q iterative clustered historical sleep monitoring groups is obtained.

[0058] Furthermore, it is determined whether the neighborhood density of the neighborhoods of the Q iterative clustered historical sleep monitoring groups is greater than or equal to the neighborhood density of the neighborhoods of the Q starting concentrated screening points. If so, it indicates that the Q iterative clustered historical sleep monitoring groups are more representative, and the Q iterative clustered historical sleep monitoring groups are used as Q iterative concentrated screening points, and the neighborhoods of the Q iterative clustered historical sleep monitoring groups are used as the neighborhoods of the Q iterative concentrated screening points. Based on the Q iterative concentrated screening points and the Q iterative concentrated screening point neighborhoods, the Q clustered historical sleep monitoring indicator group sets are continuously subjected to sleep monitoring indicator group concentrated screening until the preset maximum screening times (the maximum screening times pre-set by a person skilled in the art) is met, and the Q iterative concentrated screening points obtained from the last screening are used as the Q clustered historical sleep monitoring indicator concentrated groups.

[0059] If not, Q clustered historical sleep monitoring indicator groups are randomly extracted from the edges of the neighborhoods of the Q starting concentrated screening points again, and the Q iterative clustered historical sleep monitoring groups are updated. Then, based on the above steps, the analysis is performed again. Preferably, the number of re-analysis is counted, and when the number exceeds the preset allowed number, the Q starting concentrated screening points are used as the Q clustered historical sleep monitoring indicator concentrated groups.

[0060] S5: Analyze the Q clustered historical sleep monitoring indicator concentrated groups using the sleep management solution identification network layer to obtain a target sleep management solution.

[0061] Further, the sleep management solution identification network layer is used to analyze the Q clustered historical sleep monitoring indicator concentration groups to obtain a target sleep management solution. Step S5 of the embodiment of the present application further includes:

[0062] Acquire a plurality of sample clustered historical sleep monitoring indicator concentrated group sets and a plurality of corresponding sample target sleep management plans as a training data set;

[0063] The training data set is used to perform supervised training on a framework built based on a feedforward neural network to learn a one-to-one mapping relationship between a concentrated group set of clustered historical sleep monitoring indicators and a target sleep management plan until the training converges, thereby obtaining a trained sleep management plan identification network layer.

[0064] In one possible embodiment, the sleep management solution identification network layer is a model built based on a feedforward neural network (FNN), which is used to input a clustered historical sleep monitoring indicator set and output a target sleep management solution. Through the supervised training of the sleep management solution identification network layer, the system can efficiently and accurately convert the clustered historical sleep monitoring indicator set into a specific target sleep management solution. This step realizes the intelligence from data analysis to solution output, greatly improving the practicality and applicability of the method.

[0065] Preferably, a sample cluster historical sleep monitoring index concentration group set and a corresponding plurality of sample target sleep management schemes are obtained, wherein the sample target sleep management scheme is a target output corresponding to the training sample one by one, including personalized sleep adjustment suggestions, such as improving the deep sleep duration or reducing the number of awakenings.

[0066] The neural network model is trained with known inputs (sample clusters of historical sleep monitoring indicators) and outputs (target sleep management plans) to learn the clear correspondence between indicator sets and management plans, ensuring that each cluster feature group corresponds to a specific management plan until the model can accurately predict the output of unseen samples.

[0067] Optionally, a feedforward neural network is used to define the network layer structure (such as input layer, hidden layer, and output layer), supervised learning is performed using the training data set, and the model parameters are optimized through the back propagation algorithm to gradually reduce the prediction error. The model is continuously trained until the training error converges and the performance of the model on the validation data set is stable. The Q clustered historical sleep monitoring indicators are grouped and input into the trained network model, and the target sleep management plan corresponding to each set of data is output.

[0068] Through supervised training of the sleep management solution identification network layer, the clustered historical sleep monitoring indicators can be efficiently and accurately converted into a specific target sleep management solution. This realizes the intelligence from data analysis to solution output, greatly improving the reliability of sleep management.

[0069] In summary, the embodiments of the present application have at least the following technical effects:

[0070] 1. This application achieves a comprehensive assessment of the user's sleep quality by retrieving the target user's historical sleep monitoring data and extracting key indicators (such as deep sleep duration, number of awakenings, etc.), and performs cluster analysis from the two dimensions of time proximity and indicator similarity, so that the potential patterns of historical data can be mined and the accuracy of the analysis results can be improved.

[0071] 2. This application uses clustering methods to achieve effective data dimensionality reduction, reducing redundant information while retaining the core features of the data, and further optimizes the clustering results through centralized screening to ensure that each indicator group is more representative and consistent, providing high-quality input for subsequent analysis.

[0072] 3. This application uses the sleep management solution identification network layer and realizes a one-to-one mapping relationship between cluster indicator groups and sleep management solutions through supervised training, which can automatically generate personalized sleep management suggestions based on user data. By learning from a large amount of training data, the reliability and adaptability of solution generation are improved, making the solution more in line with the actual needs of users.

[0073] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0074] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0075] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A sleep management method for analyzing sleep record data, characterized in that: The method comprises: Retrieving monitoring data of a sleep monitor that monitors the sleep of a target user within a preset historical window, and obtaining L historical sleep monitoring data logs of L historical monitoring nodes, where L is a positive integer; Based on the preset sleep monitoring indicator groups, L historical sleep monitoring data logs are extracted respectively to obtain L historical sleep monitoring indicator groups; Performing cluster analysis on the L historical sleep monitoring indicator groups of the L historical monitoring nodes from two dimensions: the similarity of the historical monitoring nodes and the similarity of the historical sleep monitoring indicators, to obtain Q clustered historical sleep monitoring indicator group sets, where Q is a positive integer less than or equal to L; Traversing the Q clustered historical sleep monitoring indicator group sets to perform centralized screening of the sleep monitoring indicator groups, and obtaining Q clustered historical sleep monitoring indicator centralized groups; The sleep management solution identification network layer is used to analyze the Q clustered historical sleep monitoring indicator concentrated groups to obtain a target sleep management solution.

2. The sleep management method for analyzing sleep record data according to claim 1, characterized in that: The preset sleep monitoring indicator group includes deep sleep duration, awakening times, sleep onset latency and total sleep duration.

3. The sleep management method for analyzing sleep record data according to claim 1, characterized in that: Cluster analysis is performed on the L historical sleep monitoring indicator groups of the L historical monitoring nodes from two dimensions: the similarity of the historical monitoring nodes and the similarity of the historical sleep monitoring indicators, to obtain Q clustered historical sleep monitoring indicator group sets, including: Perform Q extractions from the L historical sleep monitoring indicator groups in a random manner without replacement to obtain Q initial cluster centers; Clustering the L historical sleep monitoring indicator groups based on the Q initial cluster centers to obtain Q initial cluster historical sleep monitoring indicator group sets; Using the clustering cost loss function, the overall clustering cost loss of the Q initial clustering historical sleep monitoring indicator group sets is calculated from two dimensions: the similarity of historical monitoring nodes and the similarity of historical sleep monitoring indicators, to obtain the initial clustering cost loss amount; Calculating the means of the Q initial clustered historical sleep monitoring indicator group sets respectively to obtain the means of the Q initial clustered historical sleep monitoring indicator groups, and updating the Q initial cluster centers according to the means of the Q initial clustered historical sleep monitoring indicator groups to obtain Q updated cluster centers; Combining the L historical sleep monitoring indicator groups, constructing Q updated cluster historical sleep monitoring indicator group sets of the Q updated cluster centers, and using the cluster cost loss function to perform overall cluster cost loss analysis to obtain an updated cluster cost loss amount; Determine whether the updated cluster cost loss is less than or equal to the initial cluster cost loss. If so, calculate the means of the Q updated cluster historical sleep monitoring indicator group sets respectively, and update the Q updated cluster centers according to the calculation results until the difference between the two updated cluster cost losses of two adjacent updates is less than a preset difference. Cluster the L historical sleep monitoring indicator groups according to the Q updated cluster centers obtained from the last update to obtain Q cluster historical sleep monitoring indicator group sets.

4. The sleep management method for analyzing sleep record data according to claim 3, characterized in that: include: Construct a clustering cost loss function, where the clustering cost loss function is: Among them, LOSS is the initial clustering cost loss, n i is the initial cluster historical sleep monitoring indicator group set of the i-th initial cluster center among the Q initial cluster centers, is the initial cluster historical sleep monitoring indicator group of the i-th initial cluster center, μ ij is the jth historical sleep monitoring indicator group in the initial cluster historical sleep monitoring indicator group set of the i-th initial cluster center, is the historical monitoring node corresponding to the initial cluster historical sleep monitoring indicator group of the i-th initial cluster center, t ij is the historical monitoring node corresponding to the jth historical sleep monitoring indicator group in the set of historical sleep monitoring indicator groups of the initial cluster of the i-th initial cluster center, and λ is the weight for balancing the similarity between the historical monitoring nodes and the similarity between the historical sleep monitoring indicators.

5. The sleep management method for analyzing sleep record data according to claim 1, characterized in that: Traversing the Q clustered historical sleep monitoring indicator group sets to perform centralized screening of the sleep monitoring indicator groups, and obtaining Q clustered historical sleep monitoring indicator centralized groups, including: Calculate the means of the Q clustered historical sleep monitoring indicator group sets respectively to obtain Q clustered historical sleep monitoring indicator mean groups; Taking the Q clustered historical sleep monitoring indicator mean groups as Q starting centralized screening points, constructing Q starting centralized screening point neighborhoods in the Q clustered historical sleep monitoring indicator group sets according to a preset centralized screening similarity threshold; The Q clustered historical sleep monitoring indicator group sets are subjected to sleep monitoring indicator group centralized screening based on the Q starting centralized screening points and the Q starting centralized screening point neighborhoods to obtain Q clustered historical sleep monitoring indicator centralized groups.

6. The sleep management method for analyzing sleep record data according to claim 5, characterized in that: The Q clustered historical sleep monitoring indicator group sets are subjected to sleep monitoring indicator group centralized screening based on the Q starting centralized screening points and the Q starting centralized screening point neighborhoods to obtain Q clustered historical sleep monitoring indicator centralized groups, including: Randomly extracting Q clustered historical sleep monitoring indicator groups from the edges of the neighborhoods of the screening points in the Q starting sets as Q iterative clustered historical sleep monitoring groups; Constructing Q iterative clustering historical sleep monitoring group neighborhoods of the Q iterative clustering historical sleep monitoring groups according to a preset centralized screening similarity threshold; Determine whether the neighborhood density of the Q iterative clustered historical sleep monitoring groups' neighborhoods is greater than or equal to the neighborhood density of the Q starting concentrated screening point neighborhoods. If so, use the Q iterative clustered historical sleep monitoring groups as Q iterative concentrated screening points, and use the Q iterative clustered historical sleep monitoring groups' neighborhoods as Q iterative concentrated screening point neighborhoods. Based on the Q iterative concentrated screening points and the Q iterative concentrated screening point neighborhoods, continue to perform sleep monitoring indicator group concentrated screening on the Q clustered historical sleep monitoring indicator group sets until the preset maximum screening times are met, and use the Q iterative concentrated screening points obtained from the last screening as the Q clustered historical sleep monitoring indicator concentrated groups.

7. The sleep management method for analyzing sleep record data according to claim 6, characterized in that: If not, Q clustered historical sleep monitoring indicator groups are randomly extracted again from the edges of the neighborhoods of the screening points in the Q starting sets, and the Q iterative clustered historical sleep monitoring groups are updated.

8. The sleep management method for analyzing sleep record data according to claim 1, characterized in that: The sleep management scheme identification network layer is used to analyze the Q clustered historical sleep monitoring indicator concentrated groups to obtain a target sleep management scheme, including: Acquire a plurality of sample clustered historical sleep monitoring indicator concentrated group sets and a plurality of corresponding sample target sleep management plans as a training data set; The training data set is used to perform supervised training on a framework built based on a feedforward neural network to learn a one-to-one mapping relationship between a concentrated group set of clustered historical sleep monitoring indicators and a target sleep management plan until the training converges, thereby obtaining a trained sleep management plan identification network layer.

Images