Intelligent old-age care platform data structure distribution method and system

Abstract

The invention relates to the technical field of data processing, in particular to an intelligent old-age care platform data structure distribution method and system, and the method comprises the steps: obtaining old-age health monitoring data through an intelligent device, the old-age health monitoring data comprising streaming data; analyzing the streaming data to obtain an outlier detection result, obtaining the distribution condition of outliers, and calculating the abnormality of the streaming data through the outlier detection result and the distribution condition of the outliers; performing priority adjustment from high to low on the streaming data based on the abnormality, and performing real-time distribution on the computing nodes by adopting the smart pension platform according to the adjusted priority; analyzing the streaming data to obtain the abnormality of the health streaming data of the elder, and adjusting the priorities of different data; and the computing nodes are distributed in real time according to the priorities, and input streams of all operators are balanced, so that the purpose of load balancing is achieved, and computing processing delay is prevented.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a data structure allocation method and system for a smart elderly care platform. Background Art

[0002] With the aging of the population becoming more and more serious, the issue of elderly care has become a major challenge facing the country and families. At present, my country mainly has three types of elderly care: institutional elderly care, community elderly care, and home elderly care. However, all three types of elderly care have certain limitations. Among them, institutional elderly care can provide professional and high-quality services, but it has high costs and is difficult to popularize. In addition, there is still a lack of high recognition of elderly care institutions in society. Institutional elderly care is not popular at this stage. Community elderly care can solve the problems encountered by the elderly in a relatively timely manner, and they can also enjoy the care of community service personnel, but there is also the problem of community service personnel having no sense of responsibility and perfunctory work. Home elderly care is currently the most popular elderly care method in my country, accounting for more than 90%, but elderly people living alone will encounter more limitations and troubles. To solve these problems, the "Internet + elderly care" of information and communication technology is usually adopted. This type of elderly care can improve the efficiency of elderly care services to achieve smart elderly care.

[0003] The elderly are a special user group. Smart elderly care platforms have very high requirements for real-time data calculation and analysis. Currently, most of them rely on elderly care platforms based on distributed cloud servers to implement these functions. However, this elderly care platform has load imbalance, which may lead to delays in the server's calculation and processing of some more important data, and cannot meet the response speed of current needs, which may pose potential risks to the health and safety of the elderly. Summary of the invention

[0004] In order to solve the technical problem that the existing smart elderly care platform is unbalanced in load and cannot process and analyze data in real time and accurately, the purpose of the present invention is to provide a data structure allocation method for a smart elderly care platform. The technical solution adopted is as follows: Acquiring health monitoring data of the elderly through a smart device, wherein the health monitoring data of the elderly includes streaming data; Analyze the streaming data to obtain the outlier detection results and the distribution of the outliers. Calculate the abnormality of the streaming data based on the outlier detection results and the distribution of the outliers. Based on abnormalities, the priority of streaming data is adjusted from high to low, and the smart elderly care platform is used to allocate computing nodes in real time according to the adjusted priority.

[0005] Preferably, obtaining health monitoring data of the elderly through smart devices includes: Obtain a smart elderly care terminal, and obtain health monitoring data of the elderly based on the smart elderly care terminal, wherein the streaming data includes any physiological data such as real-time blood pressure, real-time heart rate, real-time blood sugar, real-time blood lipids and real-time step count.

[0006] Preferably, analyzing the streaming data to obtain outlier detection results includes: Perform segmentation processing based on streaming data, and use windows to limit all segmented streaming data to the left and right edge ranges of the window to form a data set; Calculate the bandwidth of the data points in the window based on the data set, and obtain the kernel density estimate of the data points by combining the bandwidth with the multivariate Gaussian function; Based on the kernel density estimation value, it is determined whether it exceeds the preset outlier threshold. If not, the data point currently being analyzed is determined to be an outlier.

[0007] Preferably, the bandwidth of the data points in the window is calculated based on the data set, and the corresponding calculation formula is:

[0008]

[0009] in, Represents data points bandwidth; Representation dataset On data points; Representation dataset The standard deviation of Representation dataset The mean of Represents data points of Neighborhood average distance; Indicated in the dataset In the data point The distance is no greater than A collection of points with distances; Represents data points of The number of data points in the neighborhood; Representation dataset On data points; Represents data points To data point The Euclidean distance of .

[0010] Preferably, the calculation formula of the multivariate Gaussian function is:

[0011] in, Representation dataset The kernel function of Represents the dimension of streaming data; Representation dataset Model; Indicates An exponential function with base .

[0012] Preferably, the kernel density estimate of the data point is obtained by combining the bandwidth with the multivariate Gaussian function, and the corresponding calculation formula is:

[0013] in, Represents data points The kernel density estimate of ; Representation dataset On data points; Representation dataset The number of windows in the Representation dataset On data points; Represents data points bandwidth; Represents data points To data point The Euclidean distance of Represents the kernel function.

[0014] Preferably, obtaining the distribution of outliers includes: The windows are labeled from left to right in ascending order based on the size of the numbers as time, and the corresponding labels are the time values ​​of the data points in the data set. The time series of the outliers is obtained based on the one-to-one correspondence between the outliers and the time values. The difference of the time series of the outliers is calculated to obtain the time difference series of the outliers, and the distribution of the outliers is obtained.

[0015] Preferably, calculating the abnormality of the streaming data based on the outlier detection results and the distribution of the outliers includes: Calculate the abnormality of the data set through the outlier detection results and the distribution of outliers; Based on the abnormality, it is determined whether the abnormal threshold is exceeded. If so, it is determined that the currently analyzed streaming data has an abnormality; if not, it is determined that the smart elderly care platform processes the elderly health monitoring data normally.

[0016] Preferably, the calculation formula for calculating the abnormality of the data set is:

[0017] in, Representation dataset abnormality; Represents data points The kernel density estimate of ; Representation dataset On data points; represents the outlier threshold; Outliers The corresponding value in the time difference series; Represents a sum function.

[0018] To solve the above technical problems, the present invention provides another technical solution as follows: a smart elderly care platform data structure allocation system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the smart elderly care platform data structure allocation method as described in any one of the above when executing the computer program.

[0019] The present invention has the following beneficial effects: 1. Obtain the health monitoring data of the elderly in real time through smart devices; analyze the streaming data to obtain the abnormality of the elderly health flow data and adjust the priority of different data; allocate the computing nodes in real time according to the priority of different data, balance the input streams of each operator to achieve the purpose of load balancing and prevent computing processing delays. That is, considering the abnormality of the health monitoring data of the elderly, dynamically allocate the computing nodes of the smart elderly care platform to achieve the optimal load, and optimize the data allocation of the smart elderly care platform from the perspective of the importance and urgency of health monitoring data.

[0020] 2. A smart elderly care platform data structure allocation system provided by the present invention has the same beneficial effects as a smart elderly care platform data structure allocation method provided by the present invention, and will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 A flowchart of a method for allocating data structure of a smart elderly care platform provided by an embodiment of the present invention; Figure 2 A structural diagram of a window of a data structure allocation method for a smart elderly care platform provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the data structure allocation method and system of a smart elderly care platform proposed by the present invention, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0024] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0025] The following is a detailed description of a method and system for allocating data structures of a smart elderly care platform provided by the present invention in conjunction with the accompanying drawings.

[0026] Due to the randomness of the health monitoring data of the elderly and the unbalanced load of the existing distributed cloud server elderly care platform, there is a problem of computing delay in the server's processing of data, which cannot meet the response speed of current needs and affects the health and safety of the elderly. In one embodiment of the present invention, a data structure allocation method for a smart elderly care platform is provided, which obtains the health monitoring data of the elderly in real time through smart devices; analyzes the streaming data therein to obtain the abnormality of the elderly health flow data, and adjusts the priority of different data; allocates the computing nodes in real time according to the priority of different data, and balances the input streams of each operator to achieve the purpose of load balancing and prevent computing delays. In order to realize a data structure allocation method for a smart elderly care platform, a data structure allocation system for a smart elderly care platform is provided. The system is essentially a software system, which is composed of various devices that realize corresponding functions. The specific steps in the method are now introduced in detail.

[0027] See also Figure 1 , which shows a flowchart of a method for allocating data structure of a smart elderly care platform provided by an embodiment of the present invention, the method comprising: Step S1: Obtaining health monitoring data of the elderly through a smart device, where the health monitoring data of the elderly includes streaming data; Step S2: Analyze the streaming data to obtain outlier detection results and the distribution of outliers, and calculate the abnormality of the streaming data based on the outlier detection results and the distribution of outliers; Step S3: Based on the abnormality, the priority of the streaming data is adjusted from high to low, and the smart elderly care platform is used to allocate the computing nodes in real time according to the adjusted priority.

[0028] In this embodiment, the smart elderly care platform refers to an elderly care platform based on distributed cloud servers, which uses advanced cloud computing technology to provide comprehensive and multi-level elderly care services for the elderly; the platform ensures the efficiency and reliability of data processing and storage through a distributed cloud server architecture, and provides a stable and safe service experience for the elderly; among them, the distributed cloud server can realize rapid data processing and transmission, which greatly improves the response speed of elderly care services; the distributed cloud server has a strong data storage capacity and can store a large amount of user data and elderly care service information; the distributed cloud server has good scalability and flexibility, and can support the continuous upgrading and expansion of elderly care services; the distributed cloud server also has a high security and privacy protection capability, and through advanced encryption technology and security mechanisms, it ensures that the personal information and data of the elderly are not leaked, thereby protecting privacy rights.

[0029] It can be explained that the load imbalance phenomenon in data structure allocation usually refers to the situation in which there are significant differences in the amount of tasks and processing loads undertaken by each node or resource in the system or network; specifically, this imbalance may be manifested as some nodes or resources being overused, while other nodes or resources are relatively idle and fail to fully realize their potential. This phenomenon may lead to a decline in system performance and even bottlenecks, affecting the overall operating efficiency and stability; for example, in a distributed computing system, if the task distribution is uneven, some computing nodes may be overloaded due to processing too many tasks, while other nodes may be idle due to fewer tasks. In this case, the overall processing capacity of the system cannot be fully utilized, and may even cause failures of overloaded nodes, thereby affecting the reliability of the entire system; therefore, in response to the current load imbalance in the elderly care platform based on distributed cloud servers, a data structure allocation method for a smart elderly care platform is proposed to ensure that each node or resource can reasonably share the task and avoid computing delays.

[0030] Furthermore, step S1 includes: Obtain smart elderly care terminals and obtain health monitoring data of the elderly based on the smart elderly care terminals. The streaming data includes any physiological data such as real-time blood pressure, real-time heart rate, real-time blood sugar, real-time blood lipids and real-time step count.

[0031] It is clarified that smart elderly care terminals refer to smart devices based on the Internet of Things, such as smart bracelets, smart mattresses, and all-in-one health check-up machines, which are used to obtain health monitoring data for the elderly. That is, the data collected in real time is transmitted to the cloud or related management platforms in real time through the Internet of Things technology to facilitate subsequent analysis and processing.

[0032] Specifically, the health status of the elderly is comprehensively monitored through smart elderly care terminals to obtain health monitoring data for the elderly; among them, streaming data is real-time physiological data; real-time blood pressure refers to the changes in the blood pressure of the elderly, and abnormal conditions such as high blood pressure or low blood pressure are discovered in time; real-time heart rate refers to the heart rate of the elderly, to understand the heart health status, and to discover problems such as arrhythmia in time; real-time blood sugar is used for diabetic patients or elderly people at risk of diabetes to prevent life-threatening situations; real-time blood lipids refer to the blood lipid levels of the elderly to prevent the occurrence of cardiovascular diseases; real-time step count refers to the daily activity level of the elderly to assess whether the amount of exercise of the elderly meets the standard.

[0033] Optionally, the health monitoring data of the elderly also includes static data, which includes any health data such as health status, lifestyle and medical history, which covers key information such as the health status, lifestyle and medical history of the elderly over a period of time, so as to understand the overall health status of the elderly. Health status refers to the basic health indicators of the elderly, such as weight, height, BMI (Body Mass Index), etc.; lifestyle refers to the elderly’s living habits, such as diet, sleep, exercise, etc.; medical history refers to the elderly’s past medical history, including previous diseases, surgical history, drug allergy history, etc.

[0034] Comprehensive analysis of streaming data and static data can provide more accurate and personalized health management services for the elderly, ensuring that the health status of the elderly is fully monitored and intervened in a timely manner; however, streaming data is highly random data and is the main reason for load imbalance. It cannot meet the response speed and processing capabilities of the elderly care platform. Therefore, in this embodiment, the streaming data is mainly processed.

[0035] It is understandable that although the elderly care platform based on distributed cloud servers can distribute data processing tasks to multiple working nodes for parallel processing, the generation time, amount of data and content of streaming data are all random, and the load of data processing tasks is usually difficult to predict, resulting in an imbalance in the amount of data flowing into different subtasks of parallel data processing. This causes load imbalance in the system. When the load of some working nodes exceeds their processing capacity, output delays will occur, while other working nodes may have more idle time. This imbalance will become more and more serious over time, eventually leading to a decline in the overall performance of the elderly care platform, making the response time of data processing tasks unable to meet application requirements.

[0036] To solve the problem of load imbalance, load balancing and load distribution in traditional parallel and distributed environments mostly tend to focus on the load of the computing nodes themselves, and pay attention to the balance between nodes from the perspective of computing load. Although this can optimize the speed of parallel computing from a global macro perspective, it ignores the data balance between computing tasks, and data imbalance is the root cause of load imbalance; when the data is within the normal range, the elderly care platform will not fluctuate. For data that exceeds or is lower than the normal range, the possible situations are: the elderly are in danger, and the data of the smart elderly care platform has problems, and then the platform fails, affecting the platform's processing, that is, by analyzing the streaming data with high randomness to obtain abnormalities, analyze the problems of the entire smart elderly care platform.

[0037] Furthermore, in step S2, the streaming data is analyzed to obtain outlier detection results, including: Step S21: performing segmentation processing based on the streaming data, using a window to limit all the segmented streaming data to the left and right edge ranges of the window to form a data set; Step S22: Calculate the bandwidth of the data points in the window according to the data set, and obtain the kernel density estimation value of the data points by combining the bandwidth with the multivariate Gaussian function; Step S23: judging whether the kernel density estimation value exceeds a preset outlier threshold value based on the kernel density estimation value; if not, determining that the data point currently being analyzed is an outlier point.

[0038] To explain, outliers refer to one or more data points in a set of data that are significantly different from other data points, showing abnormal or inconsistent characteristics. They usually deviate from the distribution trend of the overall data, making them particularly prominent in statistical analysis.

[0039] See also Figure 2 It can be explained that in step S21, the window technology is specifically used to process the streaming data, wherein the window technology refers to dividing the streaming data into a series of time periods or data volume intervals, so that data processing becomes more controllable and efficient, that is, the window technology is divided into multiple types such as time window, counting window and sliding window, which are specifically applied according to unique application scenarios and advantages; the time window refers to processing data within a certain time range, such as processing data once per second; the counting window divides the window according to the amount of data, such as processing 100 data as a window; the sliding window combines the concepts of time and data volume, and a fixed time interval and data volume can be set, such as processing once every 5 seconds, and processing 100 data each time; through the window technology, the streaming data can be effectively analyzed and processed in real time, and abnormal situations in the streaming data can be discovered in time or trend prediction can be performed; it can not only improve the efficiency of streaming data processing, but also enhance the flexibility and real-time performance of streaming data processing.

[0040] For the convenience of explanation and calculation, in this embodiment, the health monitoring data of the elderly selects three indicators, namely real-time blood pressure, real-time heart rate, and real-time blood sugar, that is, the three indicators of the health monitoring data of the elderly are formed into a three-dimensional data, and then data fusion is performed on it to achieve data dimensionality reduction, and the collected health monitoring data is stored in the cache of the smart device for three days, that is, the time length of the acquired data stream is 72 hours; the sliding window is mainly used to process the streaming data; among them, the collection cycle of the health monitoring data is 1 hour, and the selected window size is 24, that is, the streaming data of the last 24 hours is taken as the object, that is, the streaming data is segmented, and the segmented streaming data is all limited to the left and right edge range of the window using the window to form a data set; in addition, the size of the window can be selected by the experimenter according to the actual situation such as the data length that the cache can store and the specific parameters of the smart device.

[0041] It can be explained that "using a window to limit all segmented streaming data to the left and right edge range of the window" means using window technology to define a range of data volume to ensure that all collected data is between the start and end points of this window, so as to effectively control the inflow of data and avoid excessive data inflow affecting the processing results; "forming a data set" means that when streaming data flows in, data is collected according to the defined window size 24, and all streaming data between the start and end points of the window are taken into consideration. When the window ends, the collected data will be integrated into a data set, which includes all streaming data records in the window for subsequent analysis.

[0042] To explain, "HOP (Hand Off Point)" refers to the sliding step of the window, that is, in window technology, the window will periodically slide forward a certain step. The size of the step determines the frequency of window updates and is related to the real-time and accuracy of data processing. The sliding step includes a fixed step at a fixed time interval and a dynamic step adjusted according to the dynamic changes of the data. To better explain, in the design of sliding windows, there is usually overlap between different windows, and this overlap can capture the continuity and trend of the data. When the sliding step is equal to the window size, there is no overlap between different windows, and each window will be completely independent at this time; the determination of the sliding step can be dynamically adjusted according to the actual processed data, that is, if it is necessary to capture subtle changes or trends, select overlapping windows, otherwise, only a general trend is needed, and non-overlapping windows are sufficient for analysis; "old tuples" refer to tuples that have existed in the sliding window for a period of time, that is, historical streaming data. Although they have not expired, they have existed for a period of time relative to the tuples that have just entered the window, but in the window sliding The data tuples in the sliding window will be retained during the sliding process until they expire. "Stream data tuples" refer to tuples in the data stream processed in the sliding window technology, which are the main objects processed by the sliding window technology. The window will continuously slide to include new stream data tuples. "New tuples" refer to the latest data tuples entering the sliding window, that is, the tuples that have just entered the window from the data stream, which are the latest data. "Expired tuples of one hop" refer to tuples that have slid out of the window range in the sliding window, that is, due to the sliding of the window, some tuples will no longer belong to the scope of the current window due to the passage of time and are considered expired. It can be explained that expired tuples usually no longer participate in subsequent data processing and analysis. "Newly added tuples of one hop" refer to tuples that have just entered the window range in the sliding window. As the window slides, new data tuples will enter the window. These newly entered tuples are newly added tuples. It can be explained that the newly added tuples will participate in subsequent data processing and analysis. The sliding window technology can be used to realize real-time processing and analysis of streaming data.

[0043] Furthermore, in step S22, the bandwidth of the data points in the window is calculated according to the data set, and the corresponding calculation formula is:

[0044]

[0045] in, Represents data points bandwidth; Representation dataset On data points; Representation dataset The standard deviation of Representation dataset The mean of Represents data points of Neighborhood average distance; Indicated in the dataset In the data point The distance is no greater than A collection of points with distances; Represents data points of The number of data points in the neighborhood; Representation dataset On data points; Represents data points To data point The Euclidean distance of .

[0046] kNN (k-Nearest Neighbors) is a machine learning algorithm. The core idea of ​​the algorithm is to find the closest distance by calculating the distance between the sample to be predicted and the known sample. neighbors, and then predict the label or value of the sample to be predicted based on the labels or values ​​of these neighbors; specifically, in the classification task, the kNN algorithm selects the closest sample based on the distance between the sample to be predicted and each sample in the training set. samples; through the voting mechanism, according to this The category label of the nearest neighbor sample determines the category of the sample to be predicted, and the distance is usually calculated using the Euclidean distance. The Euclidean distance is a method of measuring the straight-line distance between two points in a multidimensional space. It is obtained by calculating the square root of the sum of the squares of the differences between the two points in each dimension. In two-dimensional space, the Euclidean distance can be simply understood as the straight-line distance between two points; in three-dimensional space, it is the length of the straight-line segment between two points.

[0047] It can be shown that using data points of Neighborhood average distance With the dataset The overall coefficient of dispersion The fitted value of the product is used as the bandwidth; The size of the data point In its The degree of drift within the neighborhood, and the dispersion coefficient Representation dataset The overall degree of dispersion.

[0048] In this embodiment, , that is, using data points The first two points are calculated with the last two points. Take real-time blood pressure as an example. The choice of the value indicates that, without drug intervention, an episode of essential hypertension will often last at least 6 hours, so The selection of the value needs to be determined by the experimenter based on the size of the data set and the cycle of collecting streaming data in actual situations.

[0049] Furthermore, in step S22, the calculation formula of the multivariate Gaussian function is:

[0050] in, Representation dataset The kernel function of Represents the dimension of streaming data; Representation dataset Model; Indicates An exponential function with base .

[0051] It is explained that the kernel function is a function widely used in mathematics and machine learning. Its main function is to map the original data to a high-dimensional feature space so that classification or regression analysis can be more easily performed in the new space. The core idea is to implicitly perform high-dimensional mapping by calculating the inner product between data points without explicitly calculating the mapped coordinates. It can be explained that kernel functions are now mostly used in vector machines. By selecting a suitable kernel function, the vector machine can find the optimal decision boundary in the high-dimensional space and improve the accuracy of classification or regression tasks. The introduction of the kernel function not only simplifies the calculation process, but also enables the vector machine to handle complex nonlinear problems.

[0052] Furthermore, in step S22, the kernel density estimation value of the data point is obtained by combining the bandwidth with the multivariate Gaussian function, and the corresponding calculation formula is:

[0053] in, Represents data points The kernel density estimate of ; Representation dataset On data points; Representation dataset The number of windows in the Representation dataset On data points; Represents data points bandwidth; Represents data points To data point The Euclidean distance of Represents the kernel function.

[0054] To explain, Representation dataset In this embodiment, the size of the sliding window is 24, that is, ; Represents data points The bandwidth, that is ,in, Representation dataset The original bandwidth; Indicates the dimension of streaming data. In this embodiment, the three indicators of real-time blood pressure, real-time heart rate, and real-time blood sugar are mainly analyzed, namely .

[0055] Understandably, in the calculation of the kernel density estimate, due to , so the bandwidth cannot be too large; if the bandwidth is too small, the data set will have too few data points and the error will be large; if the data points In its k The greater the degree of discreteness within the neighborhood, the greater the possibility that the data point is an outlier, so the bandwidth of the data point should be larger when performing kernel density estimation, so as to estimate the kernel density of the data point in a larger range; in addition, in this embodiment, the real-time blood pressure is mainly analyzed. According to common sense, blood pressure attacks often show similar characteristics in time, that is, outliers are most likely not to appear independently but to be concentrated in a time range, so it is necessary to set a different bandwidth for each data point for estimation.

[0056] It can be explained that in step S23, the outlier threshold is Based on the above formula, the kernel density estimate of each data point in the streaming data can be obtained, that is, , the larger the value, the higher the consistency of the data point with the data points within its bandwidth; outliers are obtained based on the outlier threshold, that is, The data points that meet the judgment conditions are outliers; the smaller the kernel density estimate, the more likely it is that the elderly person's health may be in danger.

[0057] Furthermore, in step S2, the distribution of outliers is obtained, including: The windows are labeled from left to right in ascending order based on the size of the numbers as time, and the corresponding labels are the time values ​​of the data points in the data set. The time series of the outliers is obtained based on the one-to-one correspondence between the outliers and the time values. The difference of the time series of the outliers is calculated to obtain the time difference series of the outliers, and the distribution of the outliers is obtained.

[0058] Specifically, the dataset The data points in the sliding window are marked as time from left to right according to the number 1 to 24, and the number is regarded as the time value of each data point. The outliers are detected based on the above steps, and the time values ​​corresponding to the outliers are obtained based on the time number 1 to 24, that is, the time series of the outliers are obtained. ; Then calculate the difference of the time series of the outlier, that is, calculate the difference between adjacent time values ​​in the time series to obtain the time difference series of the outlier ; Then we can get the temporal distribution of outliers, that is, the temporal distribution characteristics of outliers, such as the frequency of outliers, the regularity of intervals, etc., to understand the behavior pattern of outliers in the time dimension and provide a basis for subsequent data processing.

[0059] Furthermore, in step S2, the abnormality of the streaming data is calculated based on the outlier detection results and the distribution of the outliers, including: Calculate the abnormality of the data set through the outlier detection results and the distribution of outliers; Based on the abnormality, it is determined whether the abnormal threshold is exceeded. If so, it is determined that the currently analyzed streaming data has an abnormality; if not, it is determined that the smart elderly care platform processes the elderly health monitoring data normally.

[0060] It should be noted that the abnormality of streaming data refers to indicators that are abnormal or deviate from the normal range during real-time monitoring. For example, if the real-time heart rate of an elderly person is too fast or too slow, if the heart rate continues to be too high, it may indicate arrhythmia or other heart disease. If the heart rate is too low, it may indicate insufficient heart pumping function. If the real-time monitoring data of blood pressure shows abnormal fluctuations, such as sudden increases or decreases, it may indicate symptoms of high or low blood pressure, or even be a sign of certain acute diseases. By discovering abnormalities, users of the smart elderly care platform can make follow-up actions based on the abnormalities.

[0061] Furthermore, the formula for calculating the abnormality of the data set is:

[0062] in, Representation dataset abnormality; Represents data points The kernel density estimate of ; Representation dataset On data points; represents the outlier threshold; Outliers The corresponding value in the time difference series; Represents a sum function.

[0063] To explain, based on the above calculations, we know that the smaller the kernel density estimate, the more abnormal the elderly person's body is. Taking real-time blood pressure as an example, generally, hypertension in the elderly is caused by a rapid increase in blood pressure to an abnormal value in a short period of time, and then the blood pressure remains near the value. The denser the distribution of outliers, the more continuous it is in the time series. At this time, it is considered that the data points in the sliding window in the streaming data are abnormal, that is, the health monitoring data of the elderly is abnormal. The smaller it is, the greater the abnormality of the streaming data; The smaller it is, the greater the abnormality of the streaming data.

[0064] It can be explained that, in this embodiment, the abnormal threshold is , that is, the abnormality of the current data set exceeds the abnormal threshold, , it means that the current analyzed streaming data is abnormal, and the elderly health monitoring data is abnormal; otherwise, At this time, the smart elderly care platform processes the health monitoring data of the elderly normally.

[0065] For better explanation, the abnormal threshold is dynamically adjusted according to the specific analyzed streaming data combined with different physiological indicators and health conditions. For example, for heart rate, the normal range may be 60-100 beats / minute, and for blood pressure, the normal range may be 90-120 / 60-80 mmHg. In addition, there may be special circumstances in the setting of abnormal thresholds. For example, for the elderly with a history of heart disease, the abnormal threshold needs to be set lower to avoid potential risks.

[0066] To explain, in step S3, the priority of the streaming data is adjusted from high to low based on the abnormality, and the smart elderly care platform is used to allocate the computing nodes in real time according to the adjusted priority.

[0067] It can be understood that the abnormality is calculated according to the aforementioned steps S1 and S2. For abnormal data, it is considered that the elderly is in a dangerous situation or the smart device has a fault, and the above abnormal data needs to be processed with priority for subsequent rapid response.

[0068] Specifically, based on the step description of the aforementioned step S21, it is known that the streaming data includes different data sets, the abnormality of each data set is calculated and the priority is adjusted from high to low; as an optional implementation method, the computing speed and load capacity of the distributed computing nodes of the smart elderly care platform are sorted from high to low, and the data sets with larger abnormalities are preferentially allocated to high-performance computing nodes to achieve the purpose of load balancing, ensuring that abnormal data can be processed in a timely manner and responded to quickly, and from the perspective of the importance and urgency of health monitoring data, the data allocation of the smart elderly care platform is optimized, thereby improving the overall performance and user experience of the smart elderly care platform.

[0069] It can be understood that the health monitoring data of the elderly is obtained in real time through smart devices; the streaming data is analyzed to obtain the abnormality of the health flow data of the elderly, and the priority of different data is adjusted; the computing nodes are allocated in real time according to the priority of different data, and the input streams of each operator are balanced to achieve the purpose of load balancing and prevent computing processing delays. That is, the factors of the abnormality of the health monitoring data of the elderly are considered to dynamically allocate the computing nodes of the smart elderly care platform to achieve the optimal load, and the data allocation of the smart elderly care platform is optimized from the perspective of the importance and urgency of health monitoring data.

[0070] An embodiment of the present invention further proposes a smart elderly care platform data structure allocation system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the system implements the steps of a smart elderly care platform data structure allocation method provided in an embodiment of the present invention. The system has the same beneficial effects as the aforementioned smart elderly care platform data structure allocation method, which will not be elaborated here.

[0071] It can be understood that when the various devices of a smart elderly care platform data structure distribution system are in operation, it is necessary to utilize a smart elderly care platform data structure distribution method provided by the aforementioned embodiment. Therefore, whether the memory, processor and computer program are integrated or different hardware is configured to produce functions similar to the effects achieved by the present invention, they all fall within the scope of protection of the present invention.

[0072] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. 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.

[0073] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for allocating data structure of a smart elderly care platform, characterized in that: The method comprises: Acquiring health monitoring data of the elderly through a smart device, wherein the health monitoring data of the elderly includes streaming data; Analyze the streaming data to obtain the outlier detection results and the distribution of the outliers. Calculate the abnormality of the streaming data based on the outlier detection results and the distribution of the outliers. Based on abnormalities, the priority of streaming data is adjusted from high to low, and the smart elderly care platform is used to allocate computing nodes in real time according to the adjusted priority.

2. The data structure allocation method of the smart elderly care platform according to claim 1, characterized in that: Obtain elderly health monitoring data through smart devices, including: Obtain a smart elderly care terminal, and obtain health monitoring data of the elderly based on the smart elderly care terminal, wherein the streaming data includes any physiological data such as real-time blood pressure, real-time heart rate, real-time blood sugar, real-time blood lipids and real-time step count.

3. The data structure allocation method of the smart elderly care platform according to claim 1, characterized in that: Analyze the streaming data to obtain outlier detection results, including: Perform segmentation processing based on streaming data, and use windows to limit all segmented streaming data to the left and right edge ranges of the window to form a data set; Calculate the bandwidth of the data points in the window based on the data set, and obtain the kernel density estimate of the data points by combining the bandwidth with the multivariate Gaussian function; Based on the kernel density estimation value, it is determined whether it exceeds the preset outlier threshold. If not, the data point currently being analyzed is determined to be an outlier.

4. The data structure allocation method of the smart elderly care platform according to claim 3, characterized in that: Calculate the bandwidth of the data points in the window based on the data set. The corresponding calculation formula is: in, Represents data points bandwidth; Representation dataset On data points; Representation dataset The standard deviation of Representation dataset The mean of Represents data points of Neighborhood average distance; Indicated in the dataset In the data point The distance is no greater than A collection of points with distances; Represents data points of The number of data points in the neighborhood; Representation dataset On data points; Represents data points To data point The Euclidean distance of .

5. The data structure allocation method of the smart elderly care platform according to claim 4, characterized in that: The calculation formula of the multivariate Gaussian function is: in, Representation dataset The kernel function of Represents the dimension of streaming data; Representation dataset Model; Indicates An exponential function with base .

6. The data structure allocation method of the smart elderly care platform according to claim 5, characterized in that: The kernel density estimate of the data point is obtained by combining the bandwidth with the multivariate Gaussian function. The corresponding calculation formula is: in, Represents data points The kernel density estimate of ; Representation dataset On data points; Representation dataset The number of windows in the Representation dataset On data points; Represents data points bandwidth; Represents data points To data point The Euclidean distance of Represents the kernel function.

7. The data structure allocation method of the smart elderly care platform according to claim 6, characterized in that: Get the distribution of outliers, including: The windows are labeled from left to right in ascending order based on the size of the numbers as time, and the corresponding labels are the time values ​​of the data points in the data set. The time series of the outliers is obtained based on the one-to-one correspondence between the outliers and the time values. The difference of the time series of the outliers is calculated to obtain the time difference series of the outliers, and the distribution of the outliers is obtained.

8. The data structure allocation method of the smart elderly care platform according to claim 7, characterized in that: The abnormality of streaming data is calculated based on the outlier detection results and the distribution of outliers, including: Calculate the abnormality of the data set through the outlier detection results and the distribution of outliers; Based on the abnormality, it is determined whether the abnormal threshold is exceeded. If so, it is determined that the currently analyzed streaming data has an abnormality; if not, it is determined that the smart elderly care platform processes the elderly health monitoring data normally.

9. The method for allocating data structure of a smart elderly care platform according to claim 8, characterized in that: The formula for calculating the abnormality of a dataset is: in, Representation dataset abnormality; Represents data points The kernel density estimate of ; Representation dataset On data points; represents the outlier threshold; Outliers The corresponding value in the time difference series; Represents a sum function.

10. A data structure distribution system for a smart elderly care platform, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the smart elderly care platform data structure allocation method as described in any one of claims 1 to 9 are implemented.

Images