A smart elderly care data system and data processing method based on big data

Abstract

This invention relates to the field of data processing technology, specifically to a smart elderly care data system and data processing method based on big data. The method includes: determining the user's stride variation coefficient for each complete gait cycle based on the differences in the left and right foot drive index and braking index in each complete gait cycle sequence, combined with the duration differences of different complete gait cycles; determining the time delay compensation coefficient between multi-source data by combining it with the heart rate data time series sequence; and determining the heart rate data time series correlation sequence segment from the heart rate data time series sequence based on the correlation between the heart rate data time series correlation sequence segment and the user's stride variation coefficient corresponding to the complete gait cycle, thereby determining the user's current individual abnormality detection index to determine whether intervention is necessary. This invention can ensure timely early warning and accurate intervention when a user experiences a sudden health condition, providing more reliable safety protection for the elderly.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a smart elderly care data system and data processing method based on big data. Background Technology

[0002] As the aging population becomes increasingly prevalent, traditional elderly care methods, which primarily rely on manual care, cannot meet the growing personalized and diverse needs of the elderly. Meanwhile, with advancements in information technology, intelligent technologies are gradually permeating the elderly care industry, giving rise to smart elderly care. By utilizing big data, the Internet of Things, and artificial intelligence, smart elderly care provides integrated services to the elderly, thereby improving their quality of life.

[0003] Existing problems: Current smart elderly care data systems achieve personalized elderly care services through the initial integration of multi-source heterogeneous data. However, due to the heterogeneity of cross-domain data sources and the complex correlations between multimodal data, user health risks undergo decay or nonlinear distortion during data fusion, resulting in the loss of feature-level information and the omission of key potential anomalies or dangerous signs. This data silo effect causes the system to issue erroneous warnings or judgments about users' real-time health status, failing to accurately reflect users' true health needs. Summary of the Invention

[0004] This invention provides a smart elderly care data system and data processing method based on big data to solve existing problems.

[0005] The present invention provides a smart elderly care data system and data processing method based on big data, which adopts the following technical solution:

[0006] One embodiment of the present invention provides a smart elderly care data processing method based on big data, the method comprising the following steps:

[0007] Acquire the user's heart rate data time series and complete gait cycle series, as well as the duration, start and end times, left and right foot drive index and braking index of each complete gait cycle;

[0008] Based on the differences in the left and right foot drive index and braking index in each complete gait cycle in the complete gait cycle sequence, and combined with the differences in the duration of different complete gait cycles, the user stride variation coefficient corresponding to each complete gait cycle is determined.

[0009] Based on the synchronicity of the user stride variation coefficient and the termination time in the heart rate data time series for all complete gait cycles in the complete gait cycle sequence, and combined with the duration of the complete gait cycle, the time delay compensation coefficient between multi-source data is determined.

[0010] Based on the start time and duration of the complete gait cycle in the complete gait cycle sequence, and combined with the time delay compensation coefficient between multi-source data, heart rate data time-series correlation sequence segments are divided from the heart rate data time-series sequence. Based on the correlation between the heart rate data time-series correlation sequence segments and the corresponding user stride variation coefficients of the complete gait cycles in the complete gait cycle sequence, and combined with the magnitude and differences of heart rate data in the heart rate data time-series correlation sequence segments, the user's current individual abnormality detection indicators are determined.

[0011] Based on the magnitude of the user's current individual anomaly detection indicators, determine whether intervention is necessary.

[0012] Furthermore, the specific steps for determining the user stride variation coefficient corresponding to each complete gait cycle are as follows:

[0013] For any complete gait cycle, the user motion coordination deviation factor within that complete gait cycle is determined based on the difference between the driving index and braking index of the left and right feet.

[0014] With a preset quantity threshold S, in the complete gait cycle sequence, the S preceding complete gait cycles adjacent to the i-th complete gait cycle are obtained to form the complete gait cycle sequence segment corresponding to the i-th complete gait cycle;

[0015] In the complete gait cycle sequence segment corresponding to the i-th complete gait cycle, the information entropy of the duration of all complete gait cycles is obtained, and then the mean of the absolute values ​​of the differences in the duration of all adjacent complete gait cycles is obtained. The normalized value of the sum of the information entropy and the mean is used as the gait cycle fluctuation of the user's activity trajectory.

[0016] Obtain the sum of user motion coordination deviation factors within all complete gait cycles in the complete gait cycle sequence segment corresponding to the i-th complete gait cycle. Then, normalize the product of the sum of user motion coordination deviation factors and the gait cycle fluctuation of the user's activity trajectory and denote it as the user stride variation coefficient corresponding to the i-th complete gait cycle.

[0017] Furthermore, the specific steps for determining the user's motion coordination deviation factor within any complete gait cycle, based on the difference between the left and right foot drive indices and the difference between the braking indices, are as follows:

[0018] For any complete gait cycle, obtain the absolute value of the difference between the driving index of the left and right feet, and record it as the first difference value. Then obtain the absolute value of the difference between the braking index of the left and right feet, and record it as the second difference value. The normalized value of the sum of the first difference value and the second difference value is recorded as the user motion coordination deviation factor within the given complete gait cycle.

[0019] Furthermore, the specific steps for determining the delay compensation coefficient between multi-source data are as follows:

[0020] In the complete gait cycle sequence, the end time of each complete gait cycle is assigned to the user stride variation coefficient corresponding to each complete gait cycle. In chronological order, the user stride variation coefficient time series is constructed using the user stride variation coefficients corresponding to all complete gait cycles.

[0021] Using the Fast Fourier Transform algorithm, the phase information of the user's stride coefficient variation time series and heart rate data time series under different frequency components is obtained. The phase information of the user's stride coefficient variation time series and heart rate data time series is arranged in order of increasing frequency, and the phase information sequences corresponding to the user's stride coefficient variation time series and heart rate data time series are obtained respectively.

[0022] Based on the similarity between the phase information sequences corresponding to the user's stride variation coefficient time series and the heart rate data time series, and combined with the duration of the complete gait cycle, the time delay compensation coefficient between the multi-source data is determined.

[0023] Furthermore, the specific steps for determining the time delay compensation coefficient between multi-source data based on the similarity between the phase information sequence corresponding to the user's stride variation coefficient time series and the heart rate data time series, combined with the duration of the complete gait cycle, are as follows:

[0024] Using the DTW algorithm, the DTW distance between the phase information sequence corresponding to the user's stride variation coefficient time series and the heart rate data time series is obtained. Then, the mean of the duration of all complete gait cycles in the complete gait cycle sequence is obtained. The product of the normalized value of the DTW distance and the mean of the duration is recorded as the time delay compensation coefficient between the multi-source data.

[0025] Furthermore, the specific steps for dividing the heart rate data time-series correlation sequence segment from the heart rate data time-series include the following:

[0026] In the complete gait cycle sequence segment corresponding to the last complete gait cycle in the complete gait cycle sequence, the start time of the middle complete gait cycle is recorded as the target time, and the sum of the durations of all complete gait cycles is recorded as the target duration.

[0027] The sum of the delay compensation coefficients between the target time and the multi-source data is denoted as the associated time.

[0028] Construct a related time range with the associated moment as the center and the duration as the target duration;

[0029] In the heart rate data time series, obtain the heart rate data time series segment within the associated time range, and denote it as the heart rate data time series associated sequence segment.

[0030] Furthermore, the specific steps for determining the user's current individual anomaly detection indicators are as follows:

[0031] In the complete gait cycle sequence segment corresponding to the last complete gait cycle in the complete gait cycle sequence, the termination time of each complete gait cycle is assigned to the user stride variation coefficient corresponding to each complete gait cycle. According to the time order, the user stride variation coefficient time sequence segment is formed by the user stride variation coefficients corresponding to all complete gait cycles.

[0032] Using the DTW algorithm, the inversely proportional normalized value of the DTW distance between the time series segment of the user's stride variation coefficient and the time series correlation segment of heart rate data is obtained, and is denoted as the correlation between the current user's stride variation coefficient and heart rate.

[0033] Within the time-series correlation sequence of heart rate data, the mean and variance of all heart rate data are obtained, and the normalized value of the sum of the mean and variance is recorded as the abnormal physiological response coefficient of the user in the current stage.

[0034] Based on the magnitude of the user's stride variation coefficient corresponding to the last complete gait cycle in the complete gait cycle sequence and the user's current stage of physiological response abnormality coefficient, combined with the correlation between the current user's stride variation coefficient and heart rate, the user's current individual abnormality detection index is determined.

[0035] Furthermore, the specific steps for determining the user's current individual abnormality detection index based on the magnitude of the user's stride variation coefficient corresponding to the last complete gait cycle in the complete gait cycle sequence and the user's current stage physiological response abnormality coefficient, combined with the correlation between the current user's stride variation coefficient and heart rate, are as follows:

[0036] The sum of the user's stride variation coefficient and the user's current stage physiological response abnormality coefficient corresponding to the last complete gait cycle in the complete gait cycle sequence is calculated. The product of the sum of the user's stride variation coefficient and the user's current stage physiological response abnormality coefficient and the correlation between the current user's stride variation coefficient and heart rate is recorded as the user's current individual abnormality detection index.

[0037] Furthermore, the specific steps involved in determining whether to intervene based on the magnitude of the user's current individual anomaly detection indicators are as follows:

[0038] If the normalized value of the user's current individual anomaly detection index is greater than the preset intervention threshold, an intervention command will be issued.

[0039] The present invention also proposes a smart elderly care data system based on big data, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program stored in the memory to implement the steps of the aforementioned smart elderly care data processing method based on big data.

[0040] The beneficial effects of the technical solution of the present invention are:

[0041] In this embodiment of the invention, based on the differences in the left and right foot drive index and braking index of each complete gait cycle in the complete gait cycle sequence, combined with the duration differences of different complete gait cycles, the user stride variation coefficient corresponding to each complete gait cycle is determined. Then, combined with the heart rate data time series sequence, the time delay compensation coefficient between multi-source data is determined to segment the heart rate data time series sequence. Thus, through gait coordination deviation factors and physiological data time delay compensation algorithms, the dynamic relationship between user behavior and physiological parameters is monitored in real time. Furthermore, the stride variation coefficient is analyzed using the gait cycle segmentation method to identify progressive health risks such as fall risk and chronic disease flare-ups. Based on the correlation between the heart rate data time series sequence segment and the corresponding user stride variation coefficient of the complete gait cycle, the user's current individual abnormality detection index is determined to determine whether intervention is necessary. Therefore, an attention mechanism is used to dynamically adjust the weights of multi-source data, optimize the correlation analysis between heart rate data and movement trajectory, and reduce the occurrence of misjudgments. Furthermore, by integrating anomaly detection indicators and threshold warning mechanisms, the invention can achieve intelligent identification from single data anomalies to multi-dimensional collaborative anomalies. Thus, when a user experiences a sudden health condition (such as a sharp increase in heart rate accompanied by gait imbalance), the invention can quickly activate a graded response plan, ensuring the timeliness of warnings and the accuracy of interventions, providing more reliable safety protection for the elderly. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a flowchart illustrating the steps of a smart elderly care data processing method based on big data according to the present invention.

[0044] Figure 2 This is a schematic diagram of the smart elderly care data system architecture. Detailed Implementation

[0045] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a smart elderly care data system and data processing method based on big data proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0046] Unless otherwise defined, 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 pertains.

[0047] The following description, in conjunction with the accompanying drawings, details a specific solution for a smart elderly care data system and data processing method based on big data provided by this invention.

[0048] Please see Figure 1 The diagram illustrates a flowchart of a smart elderly care data processing method based on big data, according to an embodiment of the present invention. The method includes the following steps:

[0049] Step S001: Obtain the user's heart rate data time series and complete gait cycle series, as well as the duration, start and end times, left and right foot drive index and braking index of each complete gait cycle.

[0050] In this embodiment, the data processing procedure for smart elderly care based on big data is as follows: multimodal high-synchronous data acquisition is performed, followed by data preprocessing. Then, in the data processing layer, real-time motion trajectory behavior and compensation data fusion loss information are analyzed to achieve data-driven anomaly detection. Finally, intelligent decision-making is performed in the data analysis layer. A schematic diagram of the smart elderly care data system architecture is shown below. Figure 2 As shown.

[0051] It should be noted that in this embodiment, the user's heart rate data is monitored in real time using non-contact millimeter-wave radar to obtain a time-series sequence of heart rate data. The acquisition frequency is 10 Hz, and the heart rate data is normalized using the minimum-maximum normalization method to unify the dimensions. The minimum-maximum normalization method is a well-known technique, and its specific method will not be described here. Simultaneously, UWB indoor positioning tags are used to track the user's activity trajectory in real time, and gait data is acquired using an IMU inertial sensor to accurately monitor the user's dynamic behavior. Gait data includes: gait cycle, gait cycle duration, start and end times, and the driving and braking indices of the gait cycle. Furthermore, a multi-source data spatiotemporal alignment algorithm (dynamic time warping algorithm, a well-known technique, and its specific method will not be described here) is used to synchronize the data sampling frequency of heterogeneous devices. Finally, an "information collection + cloud processing" working mode is adopted, where sensors, alarm buttons, and other devices installed in various parts of the room transmit the collected information in real time via NB-IoT (Narrowband Internet of Things) to the cloud system's data processing layer for data analysis and judgment.

[0052] It should be further explained that a gait cycle refers to the entire process from the heel of one foot striking the ground until the heel of the same foot strikes the ground again. This process corresponds to a duration and start and end times. In other words, a gait cycle describes the movement of one foot. In normal walking, the gait cycles of the left and right feet alternate. Therefore, a complete walking process involves continuous alternating gait cycles of the left and right feet. For example, after one gait cycle of the left foot ends, it is immediately followed by one gait cycle of the right foot, and so on. Therefore, in this embodiment, one gait cycle of the left (right) foot and the immediately following one gait cycle of the right (left) foot constitute a complete gait cycle. The user's complete gait cycle sequence, as well as the duration and start and end times of the complete gait cycle, are obtained in chronological order. The drive index of each gait cycle represents the maximum pitch angle when the toes leave the ground during the gait cycle, and the braking index of each gait cycle represents the maximum pitch angle when the heel strikes the ground during the gait cycle. Therefore, each complete gait cycle corresponds to a drive index and a braking index for the left and right feet.

[0053] Step S002: Based on the differences in the left and right foot drive index and braking index of each complete gait cycle in the complete gait cycle sequence, and combined with the differences in the duration of different complete gait cycles, determine the user stride variation coefficient corresponding to each complete gait cycle.

[0054] It should be noted that the core objective of the data processing layer is to establish a multimodal nonlinear correlation model by highly accurately fusing data from different fields, dynamically repairing the lack of feature-level information, and providing scientific data support for subsequent user health assessments. Therefore, real-time movement trajectories can be analyzed based on historical user behavior perception data. Abnormal activity trajectories in daily activities, such as falls or early signs of disease, cause abnormal changes in the user's gait, typically manifested as a reduced range of motion on one side of the body and asymmetrical gait cycles.

[0055] Preferably, in one embodiment of the present invention, the method for obtaining the user stride variation coefficient corresponding to each complete gait cycle includes:

[0056] In a complete gait cycle sequence, for any complete gait cycle, the absolute value of the difference between the driving index of the left and right feet is obtained and recorded as the first difference value. Then, the absolute value of the difference between the braking index of the left and right feet is obtained and recorded as the second difference value. The normalized value of the sum of the first difference value and the second difference value is recorded as the user motion coordination deviation factor in that any complete gait cycle.

[0057] It should be noted that: this embodiment uses A linear normalization function normalizes the sum of the first and second difference values ​​to between 0 and 1. The first difference value reflects the dynamic changes in the pitch angles of the user's left and right feet when the toes leave the ground during the complete gait cycle, while the second difference value reflects the dynamic changes in the pitch angles of the user's left and right feet when the heels land during the complete gait cycle. Therefore, combining the dynamic changes in the pitch angles of the user's left and right feet when they leave and land during the complete gait cycle measures the user's motion coordination deviation factor. The user's motion coordination deviation factor reflects motion asymmetry and coordination problems in the user's gait. A larger user motion coordination deviation factor indicates abnormal gait, a higher risk of falls, or potential health problems. However, the motion coordination deviation factor primarily focuses on the coordination of the left and right feet within a single complete gait cycle during daily activities and cannot explain the gait stability of the user's activity trajectory.

[0058] The preset quantity threshold S is 8, and this will be used as an example for explanation.

[0059] In the complete gait cycle sequence, the first S complete gait cycles adjacent to the i-th complete gait cycle are obtained to form the complete gait cycle sequence segment corresponding to the i-th complete gait cycle (the sequence segment formed by the i-4th to i+4th complete gait cycles).

[0060] It should be noted that: for the complete gait cycle at the beginning and end of a complete gait cycle sequence, for example: let i be 3, then the sequence segment is composed of the 1st to the 9th complete gait cycles.

[0061] In the complete gait cycle sequence segment corresponding to the i-th complete gait cycle, obtain the information entropy of the duration of all complete gait cycles, and then obtain the mean of the absolute values ​​of the differences in duration of all adjacent complete gait cycles. The normalized value of the sum of the information entropy and the mean is used as the gait cycle fluctuation of the user's activity trajectory.

[0062] It should be noted that information entropy is a well-known technique, and the specific method will not be described here. The higher the information entropy, the greater the difference in duration among all complete gait cycles in the sequence segment. This embodiment uses... The linear normalization function normalizes the sum of the information entropy and the mean to a value between 0 and 1. The gait cycle fluctuations of the user's activity trajectory reflect the gait stability and coordination within the phased activity trajectory; a larger value indicates poorer stability within the phased activity trajectory. Therefore, based on user behavior perception data, the system focuses on the stability of individual complete gait cycles during user activities through a real-time motion coordination deviation factor. It interprets the gait stability within the phased activity trajectory through phased gait cycle fluctuations, thereby quantifying the stride variation coefficient of the user's overall phases.

[0063] In the complete gait cycle sequence segment corresponding to the i-th complete gait cycle, obtain the sum of the user motion coordination deviation factors within all complete gait cycles. The normalized value of the product of this sum and the gait cycle fluctuation of the user's activity trajectory is denoted as the user stride variation coefficient corresponding to the i-th complete gait cycle.

[0064] It should be noted that the sum of the user's motor coordination deviation factors across all complete gait cycles in the sequence segment is obtained because, in the case of gradual accumulation of gait abnormalities (such as pre-fall symptoms or early symptoms of chronic diseases), the cumulative value can more sensitively reflect the overall intensity of persistent abnormalities. For example, when the cumulative user motor coordination deviation factor reaches 20 over 10 consecutive steps, it reveals the progressive changes in the user's health problems more effectively than a higher single-step user motor coordination deviation factor (10). This cumulative value helps to identify potential health risks earlier. The gait cycle fluctuations of the user's activity trajectory reflect a comprehensive indicator of the user's gait stability and coordination during a certain activity phase. The larger the value, the more severe the user's gait stability and coordination problems are during that phase, and the higher the health risk.

[0065] Step S003: Based on the synchronicity of the user stride variation coefficients corresponding to all complete gait cycles in the complete gait cycle sequence and the termination time in the heart rate data time series, and combined with the duration of the complete gait cycle, determine the time delay compensation coefficient between multi-source data.

[0066] It should be noted that, after analyzing real-time motion trajectory behavior, this embodiment uses multimodal data fusion based on an attention mechanism to compensate for lost feature information in the multi-source data fusion information. The core of this operation lies in dynamically allocating the weights of multi-source data through an attention mechanism to solve the problems of information loss and complex correlations in the cross-domain multi-source data fusion process. That is, firstly, based on the user's real-time heart rate data, the latency compensation between multi-source data during the user's daily activities is analyzed, and then the lost feature information in the multi-source data fusion information is compensated through correlation analysis between multi-source data.

[0067] It should be further noted that when the smart elderly care system performs multimodal data fusion analysis, the heterogeneity between multiple data sources leads to asynchronous data correlation, resulting in the loss of fused information. For example, a user may experience a sudden change in gait at the 10th second (a precursor to a fall), but the heart rate data at the sampling points at the 10th and 15th seconds may show normal values, which could easily lead the system to misjudge the situation as "no risk".

[0068] Preferably, in one embodiment of the present invention, the method for obtaining the delay compensation coefficient between multi-source data includes:

[0069] In the complete gait cycle sequence, the termination time of each complete gait cycle is assigned to the user stride variation coefficient corresponding to each complete gait cycle. In chronological order, the user stride variation coefficient time series is constructed using the user stride variation coefficients corresponding to all complete gait cycles.

[0070] Using the Fast Fourier Transform algorithm, the phase information of the user's stride coefficient variation time series and heart rate data time series under different frequency components is obtained. The phase information of all frequency components of the user's stride coefficient variation time series and heart rate data time series is arranged in ascending order of frequency, and the corresponding phase information sequences of the user's stride coefficient variation time series and heart rate data time series are obtained respectively.

[0071] It should be noted that the Fast Fourier Transform (FFT) algorithm is a well-known technique. The FFT algorithm can convert a time-domain signal into a frequency-domain signal, revealing different frequency components in the signal. Each frequency component corresponds to a specific frequency, and each frequency component is represented by a complex number. The real and imaginary parts of this complex number can be used to calculate phase information, which is represented by an angle value. Phase information describes the phase shift of the signal waveform at different frequencies. Analyzing the phase information at different frequency components can help understand the periodicity and synchronization of time-series data.

[0072] Using the DTW algorithm, the DTW distance between the phase information sequence corresponding to the user's stride variation coefficient time series and the heart rate data time series is obtained. Then, the mean of the duration of all complete gait cycles in the complete gait cycle sequence is obtained. The product of the normalized value of the DTW distance and the mean of the duration is recorded as the time delay compensation coefficient between the multi-source data.

[0073] It should be noted that the DTW (Dynamic Time Warping) algorithm is a well-known technique, and its specific method will not be described here. The smaller the DTW distance, the more similar the two sequences are, meaning the time series sequence of the user's stride variation coefficient is more synchronized with the heart rate data time series. This embodiment uses... A linear normalization function normalizes the DTW distance to between 0 and 1. Therefore, the normalized DTW distance quantifies the phase difference between gait activity and heart rate data in the frequency domain during daily activities, directly reflecting the degree of asynchrony between multi-source data. Thus, the normalized DTW distance is used as the weight of the average duration of all complete gait cycles to obtain the time delay compensation coefficient between multi-source data (user activity trajectory data and heart rate data). This value reflects the degree of asynchrony between gait data and heart rate data during daily activities, that is, the relationship between the phase difference and time delay of the two types of data. The larger the value, the more significant the time difference between the user's gait abrupt changes (e.g., pre-fall symptoms) and physiological data (heart rate) changes, indicating that in some emergency situations, heart rate data may not reflect potential risks in a timely manner. Therefore, in subsequent multi-source data association analysis to identify information loss, the time delay association analysis between multi-source data is optimized based on the time delay compensation coefficient.

[0074] Step S004: Based on the start time and duration of the complete gait cycle in the complete gait cycle sequence, and combined with the time delay compensation coefficient between multi-source data, divide the heart rate data time-series sequence into heart rate data time-series segments; based on the correlation between the heart rate data time-series segments and the corresponding user stride variation coefficients of the complete gait cycles in the complete gait cycle sequence, and combined with the magnitude and difference of heart rate data in the heart rate data time-series segments, determine the user's current individual abnormality detection indicators.

[0075] It should be noted that in a smart elderly care system, the correlation between multimodal data will dynamically change when a user exhibits abnormal behavior (such as pre-fall symptoms or a medical episode). For example, during a pre-fall symptom, gait data may show significant fluctuations or abrupt changes, while heart rate may increase or become irregular, indicating that the body is undergoing a physiological stress response.

[0076] Preferably, in one embodiment of the present invention, the method for obtaining the user's current individual anomaly detection indicators includes:

[0077] In the complete gait cycle sequence segment corresponding to the last complete gait cycle in the complete gait cycle sequence, the start time of the middle complete gait cycle is recorded as the target time, and the sum of the durations of all complete gait cycles is recorded as the target duration.

[0078] The sum of the delay compensation coefficients between the target time and the multi-source data is denoted as the associated time.

[0079] Centered on the associated moment, construct an associated time range with the target duration.

[0080] In the heart rate data time series, obtain the heart rate data time series segment within the associated time range, and denote it as the heart rate data time series associated sequence segment.

[0081] It should be noted that heart rate changes only gradually occur after a user's gait becomes unstable. Therefore, the sum of the time delay compensation coefficients between the target time and the multi-source data is used as the correlation time to align the motion trajectory data and heart rate data in time. This constructs a correlation time range of equal length to the complete gait cycle sequence segment corresponding to the last complete gait cycle, obtaining the temporal correlation sequence segment of heart rate data. Analyzing the last complete gait cycle is essentially real-time analysis of user anomalies.

[0082] In the complete gait cycle sequence segment corresponding to the last complete gait cycle in the complete gait cycle sequence, the termination time of each complete gait cycle is assigned to the user stride variation coefficient corresponding to each complete gait cycle. According to the time order, the user stride variation coefficient time sequence segment is constructed using the user stride variation coefficients corresponding to all complete gait cycles.

[0083] The DTW algorithm is used to obtain the DTW distance between the time series segment of the user's stride variation coefficient and the time series correlation segment of heart rate data. The inversely proportional normalized value is denoted as the correlation between the current user's stride variation coefficient and heart rate.

[0084] It should be noted that in this embodiment, the following is used: As DTW distance The inverse proportional normalized value, The function is linearly normalized, and both user stride variation coefficient and heart rate data have been normalized, eliminating dimensionless effects. DTW distance The larger the inverse proportional normalization value, the more likely the stride variation during the user's daily activities is accompanied by changes in heart rate data. This corresponds to abnormal user behavior, such as pre-fall symptoms or chronic disease flare-ups. The correlation between the current user's stride variation coefficient and heart rate can accurately quantify the lost abnormal feature information generated during the fusion of multi-source data. Further data-driven anomaly detection is needed.

[0085] Within the time-series correlation segment of heart rate data, obtain the mean and variance of all heart rate data, and record the normalized value of the sum of the mean and variance as the abnormal physiological response coefficient of the user in the current stage.

[0086] It should be noted that: this embodiment uses A linear normalization function normalizes the sum of the mean and variance to between 0 and 1. In daily activities, abnormal behaviors (such as falls or chronic conditions) often lead to abnormally high heart rates and significant data fluctuations. Therefore, by combining the mean and variance, a periodic physiological response anomaly coefficient is obtained, reflecting the degree of abnormality in the user's physiological state at a specific stage. Next, the system accurately quantifies the lost abnormal feature information generated during the fusion of multi-source data using a dynamic attention mechanism. This is combined with the abnormal behavioral patterns of users in daily activities reflected by individual data points to perform data-driven individual anomaly detection.

[0087] Calculate the sum of the user's stride variation coefficient corresponding to the last complete gait cycle in the complete gait cycle sequence and the user's current stage physiological response abnormality coefficient. Multiply this sum by the correlation between the current user's stride variation coefficient and heart rate, and record it as the user's current individual abnormality detection index.

[0088] It should be noted that individual user anomaly detection metrics are used to reflect risk warnings during daily activities, and anomaly detection is further optimized by analyzing the correlation between multimodal data. If gait abnormalities are highly correlated with abnormal physiological data in the corresponding associated window, it means that the user's abnormal behavior (such as falls or disease onset) has a higher severity and risk. The higher the user's current individual anomaly detection metrics, the more unstable or dangerous the user's health status, and the more timely the system needs to issue warnings to help monitor and take timely intervention measures.

[0089] Step S005: Determine whether to intervene based on the magnitude of the user's current individual abnormality detection indicators.

[0090] The preset intervention threshold is 0.5, and this will be used as an example for explanation.

[0091] If the normalized value of the user's current individual anomaly detection index is greater than the preset intervention threshold, an intervention command will be issued.

[0092] It should be noted that: this embodiment uses A linear normalization function normalizes the user's current individual anomaly detection indicators to a range of 0 to 1. The smart elderly care system dynamically adjusts its risk monitoring strategy based on the real-time fluctuations of these indicators and executes precise interventions tailored to the user's individual characteristics. In high-risk situations, the system automatically triggers an emergency response mechanism, contacting caregivers and intervening promptly to ensure the user's health and safety are addressed quickly. The specific intervention command triggering mechanism is based on threshold judgment; when the normalized value of the user's current individual anomaly detection indicator exceeds a preset intervention threshold, the system immediately generates and transmits an intervention command to ensure effective prevention of potential health risks.

[0093] The present invention also provides a smart elderly care data system based on big data, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program stored in the memory to implement the steps of the aforementioned smart elderly care data processing method based on big data.

[0094] This invention is now complete.

[0095] In summary, in this embodiment of the invention, based on the differences in the left and right foot drive index and braking index of each complete gait cycle in the complete gait cycle sequence, combined with the duration differences of different complete gait cycles, the user's stride variation coefficient corresponding to each complete gait cycle is determined. Then, combined with the heart rate data time series sequence, the time delay compensation coefficient between multi-source data is determined, which is used to segment the heart rate data time series correlation sequence segment from the heart rate data time series sequence. Based on the correlation between the heart rate data time series correlation sequence segment and the corresponding user stride variation coefficient of the complete gait cycle, the user's current individual abnormality detection index is determined, which is used to determine whether intervention is necessary. This invention can ensure the timeliness of early warning and the accuracy of intervention when a user experiences a sudden health condition, providing more reliable safety protection for the elderly.

[0096] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart elderly care data processing method based on big data, characterized in that, The method includes the following steps: Acquire the user's heart rate data time series and complete gait cycle series, as well as the duration, start and end times, left and right foot drive index and braking index of each complete gait cycle; Based on the differences in the left and right foot drive index and braking index in each complete gait cycle in the complete gait cycle sequence, and combined with the differences in the duration of different complete gait cycles, the user stride variation coefficient corresponding to each complete gait cycle is determined. Based on the synchronicity of the user stride variation coefficient and the termination time in the heart rate data time series for all complete gait cycles in the complete gait cycle sequence, and combined with the duration of the complete gait cycle, the time delay compensation coefficient between multi-source data is determined. Based on the start time and duration of the complete gait cycle in the complete gait cycle sequence, and combined with the time delay compensation coefficient between multi-source data, heart rate data time-series correlation sequence segments are divided from the heart rate data time-series sequence. Based on the correlation between the heart rate data time-series correlation sequence segments and the corresponding user stride variation coefficients of the complete gait cycles in the complete gait cycle sequence, and combined with the magnitude and differences of heart rate data in the heart rate data time-series correlation sequence segments, the user's current individual abnormality detection indicators are determined. Determine whether to intervene based on the magnitude of the user's current individual anomaly detection indicators; The specific steps for determining the time delay compensation coefficient between multi-source data are as follows: In the complete gait cycle sequence, the termination time of each complete gait cycle is assigned to the user's stride variation coefficient corresponding to each complete gait cycle. Following the time order, a user stride variation coefficient time series is constructed using the user stride variation coefficients corresponding to all complete gait cycles. Using the Fast Fourier Transform (FFT) algorithm, the phase information of the user stride variation coefficient time series and the heart rate data time series at different frequency components is obtained. The phase information of the user stride variation coefficient time series and the heart rate data time series are arranged in ascending order of frequency, resulting in phase information sequences corresponding to the user stride variation coefficient time series and the heart rate data time series. Using the Time-Distance Wheatstone (DTW) algorithm, the DTW distance between the phase information sequences corresponding to the user stride variation coefficient time series and the heart rate data time series is obtained. Then, the mean duration of all complete gait cycles in the complete gait cycle sequence is obtained. The product of the normalized DTW distance and the mean duration is recorded as the time delay compensation coefficient between the multi-source data. The specific steps for determining the user's current individual anomaly detection index are as follows: Within the complete gait cycle sequence segment corresponding to the last complete gait cycle in the complete gait cycle sequence, the termination time of each complete gait cycle is assigned to the user's stride variation coefficient (SVC) corresponding to that complete gait cycle. Following chronological order, the SVCs of all complete gait cycles are used to construct a time-series sequence segment of the user's stride variation coefficient. Using the DTW algorithm, the inversely proportional normalized value of the DTW distance between the user's stride variation coefficient time-series sequence segment and the heart rate data time-series correlation sequence segment is obtained, and this value is recorded as the correlation between the current user's stride variation coefficient and heart rate. Within the heart rate data time-series correlation sequence segment, the mean and variance of all heart rate data are obtained. The normalized value of the sum of the mean and variance is recorded as the user's current stage-specific physiological response anomaly coefficient. Based on the magnitude of the user's stride variation coefficient corresponding to the last complete gait cycle in the complete gait cycle sequence and the user's current stage-specific physiological response anomaly coefficient, combined with the correlation between the current user's stride variation coefficient and heart rate, the user's current individual anomaly detection index is determined.

2. The smart elderly care data processing method based on big data according to claim 1, characterized in that, The specific steps involved in determining the coefficient of variation of user stride length for each complete gait cycle are as follows: For any complete gait cycle, the user's motion coordination deviation factor within any complete gait cycle is determined based on the difference between the driving index and braking index of the left and right feet. With a preset quantity threshold S, in the complete gait cycle sequence, the S preceding complete gait cycles adjacent to the i-th complete gait cycle are obtained to form the complete gait cycle sequence segment corresponding to the i-th complete gait cycle; In the complete gait cycle sequence segment corresponding to the i-th complete gait cycle, obtain the information entropy of the duration of all complete gait cycles, and then obtain the mean of the absolute values ​​of the differences in the duration of all adjacent complete gait cycles. The normalized value of the sum of the information entropy and the mean is used as the gait cycle fluctuation of the user's activity trajectory. Obtain the sum of the user motion coordination deviation factors within all complete gait cycles in the complete gait cycle sequence segment corresponding to the i-th complete gait cycle. Then, normalize the product of the sum of the user motion coordination deviation factors and the gait cycle fluctuation of the user's activity trajectory, and denote it as the user stride variation coefficient corresponding to the i-th complete gait cycle.

3. The smart elderly care data processing method based on big data according to claim 2, characterized in that, For any complete gait cycle, the user's motion coordination deviation factor within any complete gait cycle is determined based on the differences in the driving index and braking index between the left and right feet. The specific steps include the following: For any complete gait cycle, obtain the absolute value of the difference between the driving index of the left and right feet, and record it as the first difference value. Then obtain the absolute value of the difference between the braking index of the left and right feet, and record it as the second difference value. The normalized value of the sum of the first difference value and the second difference value is recorded as the user motion coordination deviation factor within any complete gait cycle.

4. The smart elderly care data processing method based on big data according to claim 2, characterized in that, The specific steps involved in extracting time-series correlated segments of heart rate data from a time-series heart rate data sequence are as follows: In the complete gait cycle sequence segment corresponding to the last complete gait cycle in the complete gait cycle sequence, the start time of the middle complete gait cycle is recorded as the target time, and the sum of the durations of all complete gait cycles is recorded as the target duration. The sum of the delay compensation coefficients between the target time and the multi-source data is denoted as the associated time. Construct a related time range with the associated moment as the center and the duration as the target duration; In the heart rate data time series, obtain the heart rate data time series segment within the associated time range, and denote it as the heart rate data time series associated sequence segment.

5. The smart elderly care data processing method based on big data according to claim 1, characterized in that, Based on the magnitude of the user's stride variation coefficient corresponding to the last complete gait cycle in the complete gait cycle sequence and the user's current stage of physiological response abnormality coefficient, combined with the correlation between the current user's stride variation coefficient and heart rate, the user's current individual abnormality detection indicators are determined. The specific steps include the following: The sum of the user's stride variation coefficient and the user's current stage physiological response abnormality coefficient corresponding to the last complete gait cycle in the complete gait cycle sequence is calculated. The product of the sum of the user's stride variation coefficient and the user's current stage physiological response abnormality coefficient and the correlation between the current user's stride variation coefficient and heart rate is recorded as the user's current individual abnormality detection index.

6. The smart elderly care data processing method based on big data according to claim 1, characterized in that, Based on the magnitude of the user's current individual anomaly detection indicators, the decision to intervene is made, including the following specific steps: If the normalized value of the user's current individual anomaly detection index is greater than the preset intervention threshold, an intervention command will be issued.

7. A smart elderly care data system based on big data, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is executed by the processor, it implements the steps of the smart elderly care data processing method based on big data as described in any one of claims 1-6.

Images