A vital signs monitoring method and system

Abstract

The application discloses a kind of vital sign monitoring method and system, belong to health monitoring field.It includes: based on preset embedding dimension and preset time delay, HR time series data and SpO2 time series data are reconstructed in phase space, and the maximum Lyapunov index MLE of reconstructed phase space is calculated again;If MLE indicates that the human oxygen regulation system of target monitoring object is unstable, then obtain the temperature fluctuation time series data and altitude fluctuation time series data of first preset length, and based on at least one of temperature fluctuation time series data, altitude fluctuation time series data and acceleration time series data, HR time series data, SpO2 time series data are filtered, and obtain causeless HR time series data and causeless SpO2 time series data;The above-mentioned time series data is clustered respectively, and if the greater of two dispersions is not less than preset dispersion threshold, brain edema early warning is sent out.The scheme can realize early warning to high altitude cerebral edema in high altitude scene.

Description

Technical Field

[0001] This application belongs to the field of health monitoring technology, specifically relating to a method and system for monitoring vital signs. Background Technology

[0002] High-altitude areas are characterized by low oxygen levels, rapid changes in air pressure, and large temperature fluctuations. When people enter such environments, their oxygen regulation system is easily stimulated, triggering altitude sickness. High-altitude cerebral edema (HACE), a severe acute mountain sickness unique to high altitudes, has a rapid onset and progression. Its early symptoms are highly similar to those of ordinary altitude sickness, making it easily confused with other conditions. If not identified and intervened in a timely manner, it can rapidly cause damage to the central nervous system and, in severe cases, endanger life. Therefore, achieving accurate early differentiation between HACE and ordinary altitude sickness, and providing timely warnings, is an urgent problem to be solved in high-altitude vital sign monitoring. Summary of the Invention

[0003] The purpose of this application is to provide a vital signs monitoring method and system that can solve the problem in related technologies that cannot accurately distinguish between early HACE and ordinary altitude sickness and provide timely early warning.

[0004] On the one hand, embodiments of this application provide a method for monitoring vital signs, including: When an alert is received that the target monitoring object has symptoms of altitude sickness, the target monitoring object's heart rate (HR) time series data, blood oxygen saturation (SpO2) time series data, and acceleration time series data are acquired for the first preset duration before the current monitoring time. The motion weights at each acquisition time are obtained based on the acceleration time series data, the preset embedding dimension is obtained based on the degrees of freedom of the human oxygen regulation system, and the preset time delay is obtained based on the fluctuation period of the human oxygen regulation system. Based on the preset embedding dimension and the preset time delay, the HR time series data and the SpO2 time series data are reconstructed in phase space. Then, the maximum Lyapunov exponent MLE of the reconstructed phase space is calculated using the minimum data volume method, combined with the motion weights of each spatial vector in the reconstructed phase space. The motion weight of each spatial vector in the reconstructed phase space corresponds to the motion weight at its core acquisition time. If the MLE indicates that the human oxygen regulation system of the target monitoring object is unstable, then the temperature fluctuation time series data and altitude fluctuation time series data of the first preset duration are acquired, and the HR time series data and SpO2 time series data are filtered based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data and the acceleration time series data to obtain the causeless HR time series data and the causeless SpO2 time series data. Clustering is performed on the uninduced HR time series data and the uninduced SpO2 time series data respectively to obtain the dispersion of the uninduced HR time series data and the uninduced SpO2 time series data. If the larger of the two dispersions is not less than a preset dispersion threshold, a cerebral edema warning is issued.

[0005] In one optional embodiment of this application, obtaining the motion weights at each acquisition time based on the acceleration time-series data includes: For each acquisition moment, if the acceleration timing value at the acquisition moment is not greater than a first preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 1; if the acceleration timing value at the acquisition moment is greater than the first preset acceleration threshold and not greater than a second preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 0.3; if the acceleration timing value at the acquisition moment is greater than the second preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 0; wherein, the first preset acceleration threshold is less than the second preset acceleration threshold; The motion weights of each spatial vector in the reconstructed phase space are combined to calculate the MLE of the reconstructed phase space using the minimum data amount method: Determine the nearest neighbor of each spatial vector in the reconstructed phase space, and determine multiple time steps; For each time step, obtain the Euclidean distance between each spatial vector and its nearest neighbor at that time step, take the logarithm of each Euclidean distance to obtain the logarithmic Euclidean distance, and then perform a weighted average of the logarithmic Euclidean distances corresponding to each spatial vector based on the corresponding motion weights to obtain the weighted average logarithmic Euclidean distance corresponding to that time step. The MLE is obtained by performing a linear fit with each time step as the independent variable and each weighted average log Euclidean distance as the dependent variable.

[0006] In one optional embodiment of this application, the method further includes: Acquire a first preset number of high-altitude clinical time-series sample data, and obtain the corresponding fluctuation period and degrees of freedom of the human oxygen regulation system based on the high-altitude clinical time-series sample data; the high-altitude clinical time-series sample data includes HR time-series data and SpO2 time-series sample data; The method of obtaining a preset embedding dimension based on the degrees of freedom of the human oxygen regulation system and obtaining a preset time delay based on the fluctuation period of the human oxygen regulation system includes: Based on the degrees of freedom of the human oxygen regulation system, the Takens embedding theorem is used to determine a first traversal interval containing multiple initial embedding dimensions. The median value of the fluctuation period of the human oxygen regulation system is used as a temporary time delay. The phase space of the high-altitude clinical time series sample data is reconstructed based on each initial embedding parameter of the first traversal interval and the temporary time delay. The root mean square error of the spatial vector in each reconstructed phase space and the high-altitude clinical time series sample data is obtained. The initial embedding dimension corresponding to the minimum root mean square error is determined as the preset embedding dimension. Based on the fluctuation cycle of the human oxygen regulation system, a second traversal interval containing multiple initial time delays is determined. The phase space of the high-altitude clinical time series sample data is reconstructed based on each initial time delay of the second traversal interval and the preset embedding dimension. The root mean square error of the spatial vector in each reconstructed phase space corresponding to the high-altitude clinical time series sample data is obtained, and the initial time delay corresponding to the minimum root mean square error is determined as the preset time delay.

[0007] In one optional embodiment of this application, the method further includes: Based on the HR time-series data, SpO2 time-series data, and acceleration time-series data of the target monitoring object when it enters different preset altitude ranges and does not show symptoms of altitude sickness, the MLE baseline of the target monitoring object in different preset altitude ranges is obtained. Based on the time-series sample data of clinical altitude sickness but no cerebral edema in different preset altitude ranges, a second preset number of time-series samples are obtained to obtain the MLE offset threshold corresponding to different preset altitude ranges. The MLE offset threshold indicates the presence of altitude sickness but no cerebral edema in the corresponding altitude range. If the MLE is not greater than the sum of the MLE baseline and the MLE offset threshold in the corresponding altitude range, it indicates that the target monitoring object's human oxygen regulation system is not unstable, and that the target monitoring object only has altitude sickness and no cerebral edema, and an altitude sickness warning is issued. If the MLE is greater than the sum of the MLE baseline and the MLE offset threshold for the corresponding altitude range, it indicates that the target monitoring object's human oxygen regulation system is unstable, and that the target monitoring object has altitude sickness and needs further monitoring for cerebral edema.

[0008] In one optional embodiment of this application, the HR time series data and the SpO2 time series data are filtered based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data, and the acceleration time series data to obtain causeless HR time series data and causeless SpO2 time series data: If the time series value at any acquisition time in the temperature fluctuation time series data is not less than a preset temperature threshold, the time series value at any acquisition time in the altitude fluctuation time series data is not less than a preset altitude threshold, or the time series value at any acquisition time in the acceleration time series data is not less than a third preset acceleration threshold, then the time series value of the HR time series data and the time series value of the SpO2 time series data corresponding to the acquisition time are removed.

[0009] In one optional embodiment of this application, the step of clustering the uninduced HR time-series data and the uninduced SpO2 time-series data respectively to obtain the dispersion of the uninduced HR time-series data and the uninduced SpO2 time-series data respectively includes: The midpoint of the mean range of SpO2 time series data without cause and the midpoint of the mean range of HR time series data without cause under normal clinical conditions at high altitude were taken as their respective stable cluster centers, and the midpoint of the mean range of SpO2 time series data without cause and the midpoint of the mean range of HR time series data without cause under clinical cerebral edema conditions at high altitude were taken as their respective discrete cluster centers. Based on the stable cluster centers and the discrete cluster centers, the discreteness of the uninduced HR time series data and the uninduced SpO2 time series data is obtained by using a preset K-Means clustering algorithm.

[0010] In one optional embodiment of this application, the method further includes: The target monitoring object is acquired at preset time intervals for a first preset duration prior to the current moment, including HR time series data, SpO2 time series data, acceleration time series data, temperature fluctuation time series data, and altitude fluctuation time series data. Based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data, and the acceleration time series data, the HR time series data and the SpO2 time series data are filtered to obtain uninduced HR time series data and uninduced SpO2 time series data. Clustering is performed on the uninduced HR time series data and the uninduced SpO2 time series data respectively to obtain the dispersion of the uninduced HR time series data and the uninduced SpO2 time series data. If the larger of the two dispersions is not less than a preset dispersion threshold, a cerebral edema warning is issued.

[0011] Secondly, embodiments of this application provide a vital signs monitoring system, comprising: The time-series data acquisition module is used to acquire the heart rate (HR), blood oxygen saturation (SpO2), and acceleration time-series data of the target monitoring object for a first preset duration before the current monitoring time when an early warning of altitude sickness symptoms is received. The MLE acquisition module is used to acquire motion weights at each acquisition time based on the acceleration time series data, acquire a preset embedding dimension based on the degrees of freedom of the human oxygen regulation system, and acquire a preset time delay based on the fluctuation period of the human oxygen regulation system; it is used to reconstruct the phase space of the HR time series data and SpO2 time series data based on the preset embedding dimension and the preset time delay, and then calculate the maximum Lyapunov exponent (MLE) of the reconstructed phase space using the minimum data volume method, combining the motion weights of each spatial vector in the reconstructed phase space; wherein, the motion weight of each spatial vector in the reconstructed phase space corresponds to the motion weight of its core acquisition time; The altitude sickness screening module is used to acquire temperature fluctuation time-series data and altitude fluctuation time-series data for the first preset duration if the MLE indicates that the human oxygen regulation system of the target monitoring object is unstable. Based on at least one of the temperature fluctuation time-series data, the altitude fluctuation time-series data and the acceleration time-series data, the module filters the HR time-series data and the SpO2 time-series data to obtain causeless HR time-series data and causeless SpO2 time-series data. The cerebral edema screening module is used to cluster the uninduced HR time series data and the uninduced SpO2 time series data respectively, and obtain the dispersion of the uninduced HR time series data and the uninduced SpO2 time series data respectively. If the larger of the two dispersions is not less than a preset dispersion threshold, a cerebral edema warning is issued.

[0012] In one optional embodiment of this application, obtaining the motion weights at each acquisition time based on the acceleration time-series data includes: For each acquisition moment, if the acceleration timing value at the acquisition moment is not greater than a first preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 1; if the acceleration timing value at the acquisition moment is greater than the first preset acceleration threshold and not greater than a second preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 0.3; if the acceleration timing value at the acquisition moment is greater than the second preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 0; wherein, the first preset acceleration threshold is less than the second preset acceleration threshold; The motion weights of each spatial vector in the reconstructed phase space are combined to calculate the MLE of the reconstructed phase space using the minimum data amount method: Determine the nearest neighbor of each spatial vector in the reconstructed phase space, and determine multiple time steps; For each time step, obtain the Euclidean distance between each spatial vector and its nearest neighbor at that time step, take the logarithm of each Euclidean distance to obtain the logarithmic Euclidean distance, and then perform a weighted average of the logarithmic Euclidean distances corresponding to each spatial vector based on the corresponding motion weights to obtain the weighted average logarithmic Euclidean distance corresponding to that time step. The MLE is obtained by linear fitting with each time step as the independent variable and each weighted average log Euclidean distance as the dependent variable.

[0013] In one optional embodiment of this application, the system further includes a human-like modulation system feature extraction module, used for: Acquire a first preset number of high-altitude clinical time-series sample data, and obtain the corresponding fluctuation period and degrees of freedom of the human oxygen regulation system based on the high-altitude clinical time-series sample data; the high-altitude clinical time-series sample data includes HR time-series data and SpO2 time-series sample data; The method of obtaining a preset embedding dimension based on the degrees of freedom of the human oxygen regulation system and obtaining a preset time delay based on the fluctuation period of the human oxygen regulation system includes: Based on the degrees of freedom of the human oxygen regulation system, the Takens embedding theorem is used to determine a first traversal interval containing multiple initial embedding dimensions. The median value of the fluctuation period of the human oxygen regulation system is used as a temporary time delay. The phase space of the high-altitude clinical time series sample data is reconstructed based on each initial embedding parameter of the first traversal interval and the temporary time delay. The root mean square error of the spatial vector in each reconstructed phase space and the high-altitude clinical time series sample data is obtained. The initial embedding dimension corresponding to the minimum root mean square error is determined as the preset embedding dimension. Based on the fluctuation cycle of the human oxygen regulation system, a second traversal interval containing multiple initial time delays is determined. The phase space of the high-altitude clinical time series sample data is reconstructed based on each initial time delay of the second traversal interval and the preset embedding dimension. The root mean square error of the spatial vector in each reconstructed phase space corresponding to the high-altitude clinical time series sample data is obtained, and the initial time delay corresponding to the minimum root mean square error is determined as the preset time delay.

[0014] In an optional embodiment of this application, the high-intensity rejection screening module is further used for: Based on the HR time-series data, SpO2 time-series data, and acceleration time-series data of the target monitoring object when it enters different preset altitude ranges and does not show symptoms of altitude sickness, the MLE baseline of the target monitoring object in different preset altitude ranges is obtained. Based on the time-series sample data of clinical altitude sickness but no cerebral edema in different preset altitude ranges, a second preset number of time-series samples are obtained to obtain the MLE offset threshold corresponding to different preset altitude ranges. The MLE offset threshold indicates the presence of altitude sickness but no cerebral edema in the corresponding altitude range. If the MLE is not greater than the sum of the MLE baseline and the MLE offset threshold in the corresponding altitude range, it indicates that the target monitoring object's human oxygen regulation system is not unstable, and that the target monitoring object only has altitude sickness and no cerebral edema, and an altitude sickness warning is issued. If the MLE is greater than the sum of the MLE baseline and the MLE offset threshold for the corresponding altitude range, it indicates that the target monitoring object's human oxygen regulation system is unstable, and that the target monitoring object has altitude sickness and needs further monitoring for cerebral edema.

[0015] In one optional embodiment of this application, the HR time series data and the SpO2 time series data are filtered based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data, and the acceleration time series data to obtain causeless HR time series data and causeless SpO2 time series data: If the time series value at any acquisition time in the temperature fluctuation time series data is not less than a preset temperature threshold, the time series value at any acquisition time in the altitude fluctuation time series data is not less than a preset altitude threshold, or the time series value at any acquisition time in the acceleration time series data is not less than a third preset acceleration threshold, then the time series value of the HR time series data and the time series value of the SpO2 time series data corresponding to the acquisition time are removed.

[0016] In one optional embodiment of this application, the step of clustering the uninduced HR time-series data and the uninduced SpO2 time-series data respectively to obtain the dispersion of the uninduced HR time-series data and the uninduced SpO2 time-series data respectively includes: The midpoint of the mean range of SpO2 time series data without cause and the midpoint of the mean range of HR time series data without cause under normal clinical conditions at high altitude were taken as their respective stable cluster centers, and the midpoint of the mean range of SpO2 time series data without cause and the midpoint of the mean range of HR time series data without cause under clinical cerebral edema conditions at high altitude were taken as their respective discrete cluster centers. Based on the stable cluster centers and the discrete cluster centers, the discreteness of the uninduced HR time series data and the uninduced SpO2 time series data is obtained by using a preset K-Means clustering algorithm.

[0017] In one optional embodiment of this application, the system further includes a cerebral edema prevention module, used for: The target monitoring object is acquired at preset time intervals for a first preset duration prior to the current moment, including HR time series data, SpO2 time series data, acceleration time series data, temperature fluctuation time series data, and altitude fluctuation time series data. Based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data, and the acceleration time series data, the HR time series data and the SpO2 time series data are filtered to obtain uninduced HR time series data and uninduced SpO2 time series data. Clustering is performed on the uninduced HR time series data and the uninduced SpO2 time series data respectively to obtain the dispersion of the uninduced HR time series data and the uninduced SpO2 time series data. If the larger of the two dispersions is not less than a preset dispersion threshold, a cerebral edema warning is issued.

[0018] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement any of the above-described vital sign monitoring methods.

[0019] Fourthly, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the vital signs monitoring methods described above.

[0020] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements any of the vital signs monitoring methods described above.

[0021] The proposed solution, upon receiving an alert indicating altitude sickness symptoms in a target monitored object, acquires heart rate (HR) time-series data, blood oxygen saturation (SpO2) time-series data, and acceleration time-series data of the target monitored object for a first preset duration prior to the current monitoring time. Based on the acceleration time-series data, it acquires motion weights for each acquisition moment; based on the degrees of freedom of the human oxygen regulation system, it acquires a preset embedding dimension; and based on the fluctuation period of the human oxygen regulation system, it acquires a preset time delay. Based on the preset embedding dimension and the preset time delay, it reconstructs the phase space of the HR and SpO2 time-series data. Then, combining the motion weights of each spatial vector in the reconstructed phase space, it calculates the maximum Lyapunov exponent M of the reconstructed phase space using the minimum data amount method. If the MLE indicates that the human oxygen regulation system of the target monitoring object is unstable, then the temperature fluctuation time series data and altitude fluctuation time series data of the first preset duration are acquired, and the HR time series data and SpO2 time series data are filtered based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data and the acceleration time series data to obtain causeless HR time series data and causeless SpO2 time series data; the causeless HR time series data and causeless SpO2 time series data are clustered respectively to obtain the dispersion of the causeless HR time series data and causeless SpO2 time series data respectively; if the larger of the two dispersions is not less than a preset dispersion threshold, then a cerebral edema warning is issued. This scheme effectively eliminates external interferences such as motion, temperature, and altitude by collecting heart rate, blood oxygen, and acceleration data, combined with phase space reconstruction and MLE calculation, and can accurately distinguish between ordinary altitude sickness and early high-altitude cerebral edema. At the same time, through causeless data screening and cluster dispersion determination, it achieves objective quantification of central instability state, significantly reducing false positives and false negatives. The overall method has low computational power, can run continuously in real time, is compatible with wearable devices, and can provide early, rapid, and reliable automatic early warning of high-altitude cerebral edema in high-altitude scenarios. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0023] Figure 1 A flowchart illustrating a vital signs monitoring method provided by the present invention; Figure 2 A structural block diagram of a vital signs monitoring system provided by the present invention; Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0025] The technical solutions disclosed in the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0026] Figure 1 This is a flowchart illustrating a vital signs monitoring method provided in an embodiment of this application, as shown below. Figure 1 As shown, the method may include: Step S101: When a warning is received that the target monitoring object has symptoms of altitude sickness, the heart rate (HR), blood oxygen saturation (SpO2), and acceleration time series data of the target monitoring object are obtained for a first preset time period before the current monitoring time.

[0027] The target monitoring subjects refer to individuals entering high-altitude environments who require early HACE screening. These include high-altitude rescue personnel, plateau explorers, plateau workers (such as border guards and road construction workers), and plateau experimental personnel, consistent with the population sampled in previous clinical validation. The current monitoring time refers to the specific point in time when this step is performed and time-series data is acquired; it serves as the benchmark for defining the "first preset duration before the current monitoring time." For example, if the current monitoring time is 10:00 and the first preset duration is 1 minute, then time-series data from 9:59 to 10:00 will be acquired. The first preset duration refers to the length of time for a single acquisition of time-series data, used for subsequent motion weight calculations and phase space reconstruction; it must ensure the data has statistical significance. For example, the first preset duration is set to 1 minute (60 data points, acquisition frequency 1Hz).

[0028] Heart rate (HR) time-series data refers to a series of heart rate data continuously acquired at a fixed acquisition frequency (e.g., 1 Hz), with the unit being "beats / min". SpO2 (spO2) time-series data refers to a series of pulse oximetry data continuously acquired at a fixed acquisition frequency (e.g., 1 Hz), with the unit being "%". Acceleration time-series data refers to a series of motion acceleration data of the target monitoring object continuously acquired at a fixed acquisition frequency (e.g., 1 Hz), with the unit being "g" (1g ≈ 9.8 m / s²), used to determine motion status and motion weights.

[0029] It is understandable that, for the sake of ease of subsequent processing, the above time series data are collected at the same frequency and can all be collected by non-invasive wearable devices of the target monitoring object, such as smartwatches, bracelets, etc.

[0030] Specifically, when an alert is received indicating that a target monitoring subject is exhibiting symptoms of altitude sickness, it means that such symptoms have been detected. These symptoms may only be identified using existing technologies, but they could be misjudged. The target monitoring subject may already be in an early stage of cerebral edema, thus requiring further differentiation using the method described in this application. This step is fundamental to the entire HACE early screening method. Its core principle is to simultaneously collect three types of time-series data—HR, SpO2, and acceleration—to provide basic data support for subsequent motion weight calculation, phase space reconstruction, and MLE calculation. It is understood that the time-series data used in this step is valid data obtained after preprocessing steps such as simultaneous acquisition, temporal preprocessing, and validity verification. Simultaneous acquisition ensures data consistency and timestamp uniformity; temporal preprocessing removes significant noise and ensures data usability; and finally, after validity verification, the aforementioned time-series data for this step is obtained.

[0031] Step S102: Obtain motion weights at each acquisition time based on the acceleration time series data, obtain a preset embedding dimension based on the degrees of freedom of the human oxygen regulation system, and obtain a preset time delay based on the fluctuation period of the human oxygen regulation system.

[0032] Here, "acquisition time" refers to the specific time point at which each time-series data (HR, SpO2, acceleration, or time-series value) is acquired. If the acquisition frequency is 1Hz, then there is one acquisition time per second. If there are 60 acquisition times within one minute (t=0 to t=59), each acquisition time corresponds to a set of HR, SpO2, and acceleration data. Motion weight refers to the weight value determined based on the acceleration time-series data and used to correct the validity of the HR and SpO2 time-series data. The value range is [0,1]. The larger the weight, the more reliable the HR and SpO2 data at the corresponding acquisition time (less affected by motion interference). For example, when the motion weight is set to 1, the corresponding acquisition time is a resting state (acceleration ≤ 0.1g), and the HR and SpO2 data are completely valid; when set to 0, the corresponding acquisition time is a period of vigorous motion (acceleration > 0.3g), and the HR and SpO2 data are invalid.

[0033] The human oxygen regulation system refers to the physiological system responsible for regulating blood oxygen supply, heart rate fluctuations, and maintaining oxygen balance in the body. Its core components include the central nervous system, circulatory system, and respiratory system. Instability of this system can occur in the early stages of HACE. Degrees of freedom refer to the dimensions of independent movement of the human oxygen regulation system, determining the embedding dimension of phase space reconstruction and serving as a core parameter for phase space reconstruction.

[0034] The preset embedding dimension refers to the spatial dimension used for phase space reconstruction of HR and SpO2 time-series data. Determined based on the degrees of freedom of the human oxygen regulation system, it is a core parameter for phase space reconstruction. The fluctuation period refers to the time interval between repeated states of the human oxygen regulation system, reflecting the dynamic evolution of the system, and is used to determine the preset time delay for phase space reconstruction. For example, the fluctuation period of the human oxygen regulation system has been clinically validated to be 8-12 seconds, with a median of 10 seconds. The preset time delay refers to the time lag parameter used to construct the high-dimensional vector in phase space reconstruction. Determined based on the fluctuation period of the human oxygen regulation system, it ensures that the reconstructed phase space accurately reflects the system's dynamics.

[0035] Specifically, this step is the core preparatory step for subsequent phase space reconstruction and MLE calculation. The core principle is to determine motion weights through acceleration time series data, correct the validity of HR and SpO2 data, and determine the preset embedding dimension and preset time delay required for phase space reconstruction based on the characteristics of the human oxygen regulation system (degrees of freedom, fluctuation period), so as to provide parameter support for subsequent accurate calculation of MLE and investigation of central nervous system mutations without cause.

[0036] Step S103: Based on the preset embedding dimension and the preset time delay, the HR time series data and the SpO2 time series data are reconstructed in phase space. Then, the maximum Lyapunov exponent MLE of the reconstructed phase space is calculated using the minimum data volume method, combining the motion weights of each spatial vector in the reconstructed phase space. The motion weight of each spatial vector in the reconstructed phase space corresponds to the motion weight of its core acquisition time.

[0037] Phase space reconstruction refers to the method of converting one-dimensional HR and SpO2 time-series data into a set of vectors in a high-dimensional space by setting a preset embedding dimension and a preset time delay, thus reconstructing the dynamic evolution process of the human oxygen regulation system. It is the foundation of chaotic time series analysis. For example, 1 minute (60 data points) of HR time-series data, with an embedding dimension of 4 dimensions and a time delay of 10 seconds, can be reconstructed into 51 high-dimensional spatial vectors, consistent with the reconstruction logic of previous MLE calculations. Spatial vectors refer to the high-dimensional vectors formed after phase space reconstruction. Each vector consists of multiple HR or SpO2 data points collected at different times, reflecting the dynamic state of the system over a period of time. For example, the reconstructed spatial vectors of HR time-series data are [82 (t=0), 83 (t=10), 81 (t=20), 84 (t=30)], containing data from 4 different collection times, corresponding to an embedding dimension of 4 dimensions. The core collection time refers to the collection time in each spatial vector that is closest to the current monitoring time, serving as the benchmark for matching the motion weights of that vector. For example, the core acquisition time of the spatial vector [82 (t=0), 83 (t=10), 81 (t=20), 84 (t=30)] is t=30 seconds, corresponding to the motion weights at that time. Minimum Data Quantity Method: This refers to a simplified method for calculating the Maximum Lyapunov Exponent (MLE). Its core is to accurately calculate system stability indicators using the fewest effective data points, adapting to the low computing power requirements of wearable devices. The Maximum Lyapunov Exponent (MLE) is a core indicator for measuring the chaotic characteristics and stability of the human oxygen regulation system, used to determine whether the system is unstable. For example, MLE > 0 indicates system instability (possibly due to a sudden, unexplained change in the central nervous system); MLE ≤ 0 indicates system stability.

[0038] Specifically, this step is the core step in determining whether the human oxygen regulation system is unstable. The core principle is to transform one-dimensional HR and SpO2 time series data into high-dimensional vectors through phase space reconstruction, and then combine motion weights to calculate MLE using the minimum data volume method to quantify system stability, providing a core judgment basis for subsequent causeless data screening and HACE early warning.

[0039] Step S104: If the MLE indicates that the human oxygen regulation system of the target monitoring object is unstable, then acquire the temperature fluctuation time series data and altitude fluctuation time series data for the first preset duration, and filter the HR time series data and SpO2 time series data based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data and the acceleration time series data to obtain the causeless HR time series data and the causeless SpO2 time series data.

[0040] Among them, temperature fluctuation time series data refers to the temperature change data sequence of the target monitoring object's environment, collected synchronously with HR and SpO2 time series data for a first preset time period, with the unit being "°C". Its core purpose is to determine whether there is an external cause such as a sudden temperature change. Altitude fluctuation time series data refers to the altitude change data sequence of the target monitoring object's location, collected synchronously with HR and SpO2 time series data for a first preset time period, with the unit being "m". Its core purpose is to determine whether there is an external cause such as a sudden altitude change.

[0041] Uninduced HR time-series data refers to HR time-series data collected under resting and stable environmental conditions after removing the influence of external factors such as exercise, sudden temperature changes, and sudden altitude changes. This data is used to accurately identify uninduced aberrations in the central oxygen regulation system. For example, after removing all data related to external factors, the retained HR time-series data is [82, 83, 81, 84, ..., 83], which is free from any external interference and accurately reflects the state of the central oxygen regulation system, meeting the needs of subsequent cluster analysis. Uninduced SpO2 time-series data refers to SpO2 time-series data collected under resting and stable environmental conditions after removing the influence of external factors such as exercise, sudden temperature changes, and sudden altitude changes. Used in conjunction with uninduced HR time-series data, it improves the accuracy of identifying uninduced aberrations in the central nervous system. For example, the retained SpO2 time-series data is [91, 90, 92, 91, ..., 92], free from external interference. It can be used in conjunction with uninduced HR time-series data to accurately capture early central abnormalities in HACE.

[0042] Specifically, when MLE indicates instability in the human oxygen regulation system, it is necessary to first distinguish the cause of instability (external triggers or central non-trigger mutations). By collecting temperature and altitude fluctuation data, combined with reused acceleration data, data that is interfered with by external triggers are removed, and non-trigger data that can truly reflect the state of the central oxygen regulation system itself is retained, providing an accurate data source for subsequent early HACE (central non-trigger mutation) screening.

[0043] From the perspective of high-altitude medicine, the core pathological mechanism of HACE (high-altitude cerebral edema) is the abnormal oxygen regulation function of the central nervous system (especially brain tissue), which leads to vasogenic edema and cytotoxic edema. Its early core feature is "central uninduced mutation", that is, after excluding external factors, the stability of the human oxygen regulation system (centrally dominated) decreases, which is manifested as hidden abnormal fluctuations in HR and SpO2 data.

[0044] Specifically, the medical mechanisms of exercise-induced stress are as follows: During strenuous exercise, the oxygen consumption of skeletal muscles increases sharply, and the sympathetic nervous system is excited, leading to an increased heart rate (HR) and a temporary decrease in SpO2. This is a normal physiological stress response and can be quickly recovered within 10-15 minutes after stopping exercise. It is unrelated to any abnormality in the central oxygen regulation system and is not an early characteristic of HACE. The medical mechanisms of sudden altitude changes are as follows: When altitude increases rapidly (≥50m / minute), the external atmospheric pressure decreases, and the alveolar oxygen partial pressure decreases, leading to passive hypoxia in the human body. The HR increases compensatorily and SpO2 decreases. This is an environmental adaptation response and can be gradually recovered within 30 minutes after the altitude stabilizes. It is not a sudden change in the central nervous system. The medical mechanisms of sudden temperature changes are as follows: When the ambient temperature changes suddenly (≥2℃ / minute), the human body's blood vessels constrict or dilate, affecting blood circulation and indirectly causing temporary fluctuations in HR and SpO2. This is a physiological response related to body temperature regulation and is not related to any abnormality in the central oxygen regulation function. Therefore, it cannot be used as an early criterion for HACE.

[0045] Step S105: Cluster the uninduced HR time series data and the uninduced SpO2 time series data respectively to obtain the dispersion of the uninduced HR time series data and the uninduced SpO2 time series data. If the larger of the two dispersions is not less than a preset dispersion threshold, a cerebral edema warning is issued.

[0046] Clustering refers to the method of grouping time-series data of HR and SpO2 without causal factors according to their data distribution characteristics (centralized or discrete) to distinguish between two states: stable data (no central nervous system mutation) and discrete data (central nervous system mutation). A simplified K-means clustering algorithm can be used, with K=2 clusters (distinguishing only between stable and discrete clusters), requiring no complex iterative optimization and adapting to the low computing power requirements of wearable devices. Dispersion refers to the degree of dispersion of the time-series data of a single indicator (HR or SpO2). The greater the dispersion, the more drastic the data fluctuation, and the worse the stability of the corresponding central oxygen regulation system. It is a core indicator for quantifying central nervous system mutations without causal factors. The preset dispersion threshold is a critical value used to determine central nervous system mutations without causal factors. It was verified and determined based on high-altitude clinical sample data and is the core judgment criterion for triggering cerebral edema warning.

[0047] Specifically, by clustering the time-series data of HR and SpO2 without cause, the data is distinguished between stable and discrete states. The dispersion of each is calculated and the degree of data fluctuation is quantified. The larger of the two dispersions is used as the judgment criterion. Combined with the preset dispersion threshold, it is determined whether there is a central uninduced mutation, thereby triggering a warning for cerebral edema.

[0048] Understandably, the core of this solution is to accurately distinguish between altitude sickness (HACE) and early-stage high-altitude cerebral edema (HACE) through a hierarchical logic of "system stability determination - external inducing factor elimination - central nervous system mutation quantification" combined with high-altitude medical pathology. Technically, it first collects multi-source time-series data such as HR, SpO2, acceleration, temperature, and altitude. It then uses phase space reconstruction and the minimum data volume method to calculate the MLE (maximum Lyapunov index) to determine whether the human oxygen regulation system is unstable. Only when instability is found, external inducing factors such as exercise and sudden environmental changes are eliminated through temperature, altitude, and acceleration data, retaining data without inducing factors. Finally, the dispersion of the uninduced HR and SpO2 data is clustered and quantified. If the dispersion meets the standard, it is determined to be early-stage HACE (central nervous system mutation without inducing factors); if it does not meet the standard, it is determined to be altitude sickness (physiological stress). This not only solves the problem of easy confusion between the two in clinical practice, but also adapts to the engineering implementation of wearable devices.

[0049] The solution provided in this application, upon receiving an alert that a target monitoring object exhibits symptoms of altitude sickness, acquires heart rate (HR) time-series data, blood oxygen saturation (SpO2) time-series data, and acceleration time-series data of the target monitoring object for a first preset duration prior to the current monitoring time. Based on the acceleration time-series data, it acquires motion weights for each acquisition time, obtains a preset embedding dimension based on the degrees of freedom of the human oxygen regulation system, and obtains a preset time delay based on the fluctuation period of the human oxygen regulation system. Based on the preset embedding dimension and the preset time delay, it reconstructs the phase space of the HR and SpO2 time-series data, and then calculates the maximum Lyapunov exponent of the reconstructed phase space using the minimum data amount method, combining the motion weights of each spatial vector in the reconstructed phase space. MLE; If the MLE indicates that the human oxygen regulation system of the target monitoring object is unstable, then acquire the temperature fluctuation time series data and altitude fluctuation time series data for the first preset duration, and based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data and the acceleration time series data, filter the HR time series data and the SpO2 time series data to obtain causeless HR time series data and causeless SpO2 time series data; cluster the causeless HR time series data and the causeless SpO2 time series data respectively to obtain the dispersion of the causeless HR time series data and the causeless SpO2 time series data respectively, and if the larger of the two dispersions is not less than a preset dispersion threshold, then issue a cerebral edema warning. This scheme effectively eliminates external interferences such as motion, temperature, and altitude by collecting heart rate, blood oxygen, and acceleration data, combined with phase space reconstruction and MLE calculation, and can accurately distinguish between ordinary altitude sickness and early high-altitude cerebral edema. At the same time, through causeless data screening and cluster dispersion determination, it achieves objective quantification of central instability state, significantly reducing false positives and false negatives. The overall method has low computational power, can run continuously in real time, is compatible with wearable devices, and can provide early, rapid, and reliable automatic early warning of high-altitude cerebral edema in high-altitude scenarios.

[0050] In one optional embodiment of this application, obtaining the motion weights at each acquisition time based on the acceleration time-series data includes: For each acquisition moment, if the acceleration timing value at the acquisition moment is not greater than a first preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 1; if the acceleration timing value at the acquisition moment is greater than the first preset acceleration threshold and not greater than a second preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 0.3; if the acceleration timing value at the acquisition moment is greater than the second preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 0; wherein, the first preset acceleration threshold is less than the second preset acceleration threshold; The motion weights of each spatial vector in the reconstructed phase space are combined to calculate the MLE of the reconstructed phase space using the minimum data amount method: Determine the nearest neighbor of each spatial vector in the reconstructed phase space, and determine multiple time steps; For each time step, obtain the Euclidean distance between each spatial vector and its nearest neighbor at that time step, take the logarithm of each Euclidean distance to obtain the logarithmic Euclidean distance, and then perform a weighted average of the logarithmic Euclidean distances corresponding to each spatial vector based on the corresponding motion weights to obtain the weighted average logarithmic Euclidean distance corresponding to that time step. The MLE is obtained by linear fitting with each time step as the independent variable and each weighted average log Euclidean distance as the dependent variable.

[0051] Specifically, the minimum emission potential (MLE) of the reconstructed phase space is calculated using the minimum data amount method in conjunction with motion weights. The MLE calculation is completed in four steps to distinguish between ordinary high-risk reactions and early pathological states of HACE. The specific steps are as follows: 1. Determine the nearest neighbor and time step of each spatial vector in the reconstructed phase space. The nearest neighbor point is determined for any spatial vector in the reconstructed phase space. Calculate its relationship with all other spatial vectors in the phase space. The Euclidean distance, the one that minimizes the Euclidean distance. Determined as The nearest neighbor points reflect the two most similar physiological states during the evolution of the oxygen regulation system, laying the foundation for subsequent distance calculations.

[0052] Time step setting: If the characteristic fluctuation period of the human oxygen regulation system is 3-5 seconds, set the time step. The time steps are set to [1,2,3,4,5]s (5 time steps in total) to ensure accurate capture of the dynamic evolution characteristics of the oxygen regulation system and avoid loss of system stability characteristics due to unreasonable time step settings.

[0053] 2. Calculate the weighted average logarithmic Euclidean distance at each time step. For each set time step The process sequentially performs Euclidean distance calculation, logarithmic transformation, and weighted averaging to obtain the weighted average logarithmic Euclidean distance for that time step. Motion weight correction is introduced throughout the process to reduce motion interference. Calculate the Euclidean distance for each time step. Each spatial vector below Calculate its nearest neighbor. Euclidean distance between The formula is:

[0054] in, For time step Below, spatial vectors Nearest Neighbor The Euclidean distance is dimensionless and is used to quantify the degree of difference in the physiological state of the oxygen regulation system corresponding to two spatial vectors; m is the embedding dimension, which is consistent with the preset parameter and corresponds to the dimension of the phase space, fitting the characteristic fluctuation period of the oxygen regulation system of 3-5s. The value of m will be explained later; for example, it can be 4. For time step Below, spatial vectors The m-th component, namely the time series values ​​of HR and SpO2 corresponding to the acquisition time, reflects the physiological state value of the oxygen regulation system in this dimension; For time step Below, nearest neighbor point The m-th component reflects the relationship with The physiological state values ​​of the most similar oxygen regulation system.

[0055] Logarithmic Euclidean distance transformation, for each Euclidean distance Taking the natural logarithm, we obtain the logarithmic Euclidean distance. This linearizes the nonlinear distance relationship, amplifies subtle differences in physiological states, improves the accuracy of subsequent linear fitting, and aligns with the computational logic of the minimum data volume method.

[0056] Weighted average calculation, for time steps The logarithmic Euclidean distance of all spatial vectors, based on their corresponding motion weights. Perform a weighted average to obtain the weighted average logarithmic Euclidean distance for that time step. The formula is:

[0057] in, For time step The corresponding weighted average logarithmic Euclidean distance is dimensionless and reflects the overall fluctuation characteristics of the oxygen regulation system at that time step. After being corrected by motion weights, it can truly reflect the system evolution law under resting conditions. For spatial vectors The corresponding motion weights; The total number of spatial vectors in the reconstructed phase space is calculated from the HR and SpO2 data acquisition length, embedding dimension, and time delay.

[0058] 3. Obtain the MLE value through linear fitting. For all time steps and its corresponding weighted average log Euclidean distance Linear fitting is performed according to the following formula:

[0059] in, is the linear fitting constant, dimensionless, and is the intercept of the fitted line, automatically generated during the fitting process, and does not affect the core judgment result of MLE.

[0060] The proposed solution employs the least squares method for linear fitting, ensuring the accuracy of the fitting results and meeting the computational requirements of the minimum data volume method. At the same time, the fitting process takes into account low computing power requirements, making it suitable for wearable device deployment scenarios.

[0061] In one optional embodiment of this application, the method further includes: Acquire a first preset number of high-altitude clinical time-series sample data, and obtain the corresponding fluctuation period and degrees of freedom of the human oxygen regulation system based on the high-altitude clinical time-series sample data; the high-altitude clinical time-series sample data includes HR time-series data and SpO2 time-series sample data; The method of obtaining a preset embedding dimension based on the degrees of freedom of the human oxygen regulation system and obtaining a preset time delay based on the fluctuation period of the human oxygen regulation system includes: Based on the degrees of freedom of the human oxygen regulation system, the Takens embedding theorem is used to determine a first traversal interval containing multiple initial embedding dimensions. The median value of the fluctuation period of the human oxygen regulation system is used as a temporary time delay. The phase space of the high-altitude clinical time series sample data is reconstructed based on each initial embedding parameter of the first traversal interval and the temporary time delay. The root mean square error of the spatial vector in each reconstructed phase space and the high-altitude clinical time series sample data is obtained. The initial embedding dimension corresponding to the minimum root mean square error is determined as the preset embedding dimension. Based on the fluctuation cycle of the human oxygen regulation system, a second traversal interval containing multiple initial time delays is determined. The phase space of the high-altitude clinical time series sample data is reconstructed based on each initial time delay of the second traversal interval and the preset embedding dimension. The root mean square error of the spatial vector in each reconstructed phase space corresponding to the high-altitude clinical time series sample data is obtained, and the initial time delay corresponding to the minimum root mean square error is determined as the preset time delay.

[0062] Specifically, a first predetermined number of high-altitude clinical time-series sample data is acquired. All sample data are collected from monitored subjects in high-altitude environments and include HR and SpO2 time-series sample data at different altitudes and under different resting states. The sample size must meet the clinical statistical significance requirement (verified in high-altitude clinical trials, the first predetermined number is no less than 500 groups) to ensure the universality of the extracted physiological characteristic parameters. Time-series feature extraction methods are used to analyze the periodic fluctuation patterns of HR and SpO2 time-series data, statistically determining the fluctuation period of the human oxygen regulation system (denoted as T; verified in clinical samples, T is 3-5 seconds). Based on the physiological regulation mechanism of the oxygen regulation system and combined with feature dimension analysis of the clinical sample data, the degrees of freedom of the human oxygen regulation system are extracted (denoted as d, d=2, corresponding to the core regulation dimension of the oxygen regulation system, avoiding increased computational complexity due to redundant dimensions).

[0063] Determining the preset embedding dimension based on the degrees of freedom of the oxygen regulation system: This step, based on Takens' embedding theorem, uses the degrees of freedom of the oxygen regulation system as a foundation, and determines the optimal preset embedding dimension through traversal screening and RMSE quantification of the goodness of fit, ensuring that the parameters match the physiological characteristics of the oxygen regulation system. Specifically, it may include the following steps: 1. Determine the first traversal interval: Based on the extracted degrees of freedom d of the human oxygen regulation system, using Takens' embedding theorem (the embedding dimension must satisfy m≥2d+1), determine the first traversal interval containing multiple initial embedding dimensions. Combining d=2, the first traversal interval is set as [5,6,7,8] (satisfying m≥2×2+1=5, taking into account physiological characteristics and computational power, and avoiding redundancy caused by excessively high dimensions). 2. Set a temporary time delay: Use the median value of the fluctuation period T of the human oxygen regulation system extracted in step 1 as the temporary time delay (denoted as...). ,like =3-5s, then =4s), providing a unified time delay parameter for phase space reconstruction, ensuring that only a single variable (initial embedding dimension) is used during the traversal and filtering process, thus improving the accuracy of the filtering results; 3. Traverse and reconstruct the phase space and calculate the root mean square error (RMSE): Select each initial embedding dimension in the first traversal interval sequentially, and combine it with the temporary time delay to reconstruct the phase space of the preprocessed high-altitude clinical time-series sample data, obtaining the reconstructed phase space corresponding to each initial embedding dimension; for each reconstructed phase space, calculate the root mean square error (RMSE) between the spatial vector within it and the original high-altitude clinical time-series sample data. ), The calculation formula is:

[0064] in, RMSE is the root mean square error, dimensionless, used to quantify the fit between the reconstructed phase space vector and the original clinical sample data. The smaller the RMSE value, the better the reconstruction effect, and the more accurately the phase space vector reflects the physiological state of the oxygen regulation system; n is the number of clinical sample data collection points, dimensionless. For the i-th original high-altitude clinical time-series sample data (HR and SpO2 time-series data); The fitted data corresponding to the spatial vector in the i-th reconstructed phase space is generated by the phase space reconstruction algorithm.

[0065] 4. Determine the preset embedding dimensions: Compare the corresponding dimensions of each initial embedding dimension. The value will be the smallest. The initial embedding dimension corresponding to the value is determined as the preset embedding dimension (denoted as m). Clinical samples have verified that the preset embedding dimension has a good fit with m=4 and perfectly matches the Takens embedding theorem and the physiological characteristics of the oxygen regulation system with degrees of freedom d=2, ensuring that the phase space reconstruction is free of redundancy and distortion.

[0066] Then, based on the extracted system fluctuation cycle and with the determined preset embedding dimension as a fixed parameter, the optimal preset time delay is determined by iterating through the selection process again and quantifying the fit using RMSE. This can include the following steps: 1. Determine the second traversal interval: Based on the extracted fluctuation period T of the human oxygen regulation system, determine the second traversal interval, which includes multiple initial time delays. If T = 3-5s, then the second traversal interval is set to [2,3,4,5]s (covering the characteristic fluctuation period of the system to ensure that the dynamic evolution characteristics of the oxygen regulation system can be captured). 2. Fixed preset embedding dimension traversal reconstruction: Select each initial time delay in the second traversal interval in turn, and combine it with the preset embedding dimension m determined above to reconstruct the phase space of the preprocessed high-altitude clinical time series sample data to obtain the reconstructed phase space corresponding to each initial time delay. 3. Calculate the RMSE of each reconstructed phase space: According to the RMSE calculation formula in the previous text, calculate the RMSE value of the spatial vector in each reconstructed phase space corresponding to the original high-altitude clinical time series sample data, and quantify the reconstruction fit of each initial time delay. 4. Determine the preset time delay: Compare the RMSE values ​​corresponding to each initial time delay, and determine the initial time delay corresponding to the minimum RMSE value as the preset time delay (denoted as τ). Validated by clinical samples, this preset time delay is set to 3s to ensure accurate capture of the 3-5s characteristic fluctuation cycle of the oxygen regulation system and avoid loss of evolutionary features due to excessively large or small time delays.

[0067] In one optional embodiment of this application, the method further includes: Based on the HR time-series data, SpO2 time-series data, and acceleration time-series data of the target monitoring object when it enters different preset altitude ranges and does not show symptoms of altitude sickness, the MLE baseline of the target monitoring object in different preset altitude ranges is obtained. Based on the time-series sample data of clinical altitude sickness but no cerebral edema in different preset altitude ranges, a second preset number of time-series samples are obtained to obtain the MLE offset threshold corresponding to different preset altitude ranges. The MLE offset threshold indicates the presence of altitude sickness but no cerebral edema in the corresponding altitude range. If the MLE is not greater than the sum of the MLE baseline and the MLE offset threshold in the corresponding altitude range, it indicates that the target monitoring object's human oxygen regulation system is not unstable, and that the target monitoring object only has altitude sickness and no cerebral edema, and an altitude sickness warning is issued. If the MLE is greater than the sum of the MLE baseline and the MLE offset threshold for the corresponding altitude range, it indicates that the target monitoring object's human oxygen regulation system is unstable, and that the target monitoring object has altitude sickness and needs further monitoring for cerebral edema.

[0068] Among them, the MLE baseline refers to the mean MLE value of the target monitoring subject in a stable oxygen regulation system during the initial time period after entering a certain preset altitude range, without altitude sickness symptoms. It is the benchmark value for the stability of the individual's oxygen regulation system in that altitude range, dimensionless, reflecting the individual adaptability of high-altitude monitoring. The MLE offset threshold refers to the maximum deviation of the MLE value from the healthy baseline in a certain preset altitude range when only altitude sickness is clinically present but without cerebral edema. It is dimensionless and indicates the upper limit of MLE fluctuation under physiological stress in that altitude range, serving as a critical reference value to distinguish between physiological and pathological states. Preset altitude range: An altitude range divided according to the high-altitude gradient and the adaptation law of the human oxygen regulation system. It is the basis for achieving differentiated judgment of different altitude environments, conforming to the gradient characteristics of the physiological effects of high-altitude hypoxic environments on the human body.

[0069] Specifically, by combining the physiological adaptation characteristics of altitude gradients in high-altitude environments, the problem of insufficient adaptability of a single MLE value determination in different altitude ranges can be solved, and the hierarchical early warning system of "altitude sickness warning - potential risk warning of cerebral edema" can be improved to adapt to the monitoring needs of wearable devices in different high-altitude scenarios.

[0070] Obtaining MLE baselines for different preset altitude ranges can involve the following steps: 1. Divide the altitude range into preset ranges: Based on the adaptation pattern of the human body's oxygen regulation system at high altitudes, divide the altitude range into preset ranges in 500m gradients. For example, divide them into 3000-3500m, 3500-4000m, 4000-4500m, and above 4500m. The gradient of each range is uniform, and the stress response of the human body to low oxygen environment at different altitudes varies. 2. Collect individual baseline time series data: During the initial time period (1-2 hours) when the target monitoring object enters each preset altitude range and does not show any symptoms of altitude sickness, according to the collection rules in step S101, simultaneously acquire HR time series data, SpO2 time series data and acceleration time series data for the first preset duration. The collection frequency is consistent with the aforementioned time series data collection frequency to ensure data dimension matching. 3. Baseline data preprocessing: Outliers are removed and noise interference is reduced from the collected baseline time series data for each altitude range to obtain baseline data with high signal-to-noise ratio, thus avoiding data interference with the accuracy of MLE calculation; 4. Calculate the individual baseline MLE value: For the preprocessed baseline data, complete phase space reconstruction (embedding dimension m=4, time delay τ=3s), motion weighting, and minimum data volume method MLE calculation. Multiple data acquisitions and calculations are performed, and the average value is determined as the MLE baseline of the target monitoring object in the corresponding preset altitude range (denoted as ). The mean calculation eliminates the error of single data fluctuations, ensuring that the baseline truly reflects the individual's stable state.

[0071] Obtaining the MLE offset threshold for different preset altitude ranges can involve the following steps: 1. Collect clinical time-series data: Obtain clinical time-series sample data for different preset altitude intervals with a second preset number (≥300 groups / altitude interval). The samples are all from monitoring subjects who only show symptoms of altitude sickness and have no pathological state of cerebral edema. The sample data includes time-series data of HR, SpO2, and acceleration, and meet the clinical statistical significance. 2. Clinical sample data preprocessing and MLE calculation: For clinical sample data in each altitude range, the MLE calculation process (phase space reconstruction + motion weights + minimum data volume method) is executed to obtain the MLE calculation value of each group of samples in the corresponding altitude range; 3. Determine the MLE offset threshold: For all clinical samples of common altitude sickness in each altitude range, calculate the maximum offset of the MLE value relative to the mean MLE of the healthy population in that altitude range. Determine this maximum offset as the MLE offset threshold for the corresponding preset altitude range (denoted as MLE). The value is the upper limit of MLE fluctuation for normal altitude sickness physiological stress in this altitude range. Exceeding this value indicates a risk of cerebral edema pathology.

[0072] Finally, based on the MLE obtained during the implementation, the altitude range was determined and compared with the corresponding altitude range. and The sums are compared to determine whether cerebral edema exists.

[0073] In one optional embodiment of this application, the HR time series data and the SpO2 time series data are filtered based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data, and the acceleration time series data to obtain causeless HR time series data and causeless SpO2 time series data: If the time series value at any acquisition time in the temperature fluctuation time series data is not less than a preset temperature threshold, the time series value at any acquisition time in the altitude fluctuation time series data is not less than a preset altitude threshold, or the time series value at any acquisition time in the acceleration time series data is not less than a third preset acceleration threshold, then the time series value of the HR time series data and the time series value of the SpO2 time series data corresponding to the acquisition time are removed.

[0074] Specifically, by using one or more of the time series data of temperature fluctuation, altitude fluctuation, and acceleration, the HR time series data and SpO2 time series data are filtered to obtain the corresponding uninduced HR time series data and uninduced SpO2 time series data.

[0075] In one optional embodiment of this application, the step of clustering the uninduced HR time-series data and the uninduced SpO2 time-series data respectively to obtain the dispersion of the uninduced HR time-series data and the uninduced SpO2 time-series data respectively includes: The midpoint of the mean range of SpO2 time series data without cause and the midpoint of the mean range of HR time series data without cause under normal clinical conditions at high altitude were taken as their respective stable cluster centers, and the midpoint of the mean range of SpO2 time series data without cause and the midpoint of the mean range of HR time series data without cause under clinical cerebral edema conditions at high altitude were taken as their respective discrete cluster centers. Based on the stable cluster centers and the discrete cluster centers, the discreteness of the uninduced HR time series data and the uninduced SpO2 time series data is obtained by using a preset K-Means clustering algorithm.

[0076] Among them, the stable cluster center refers to the midpoint of the mean range of SpO2 / HR time-series sample data under normal clinical conditions at high altitude. It is the core cluster center in K-Means clustering that represents the normal and stable state of the human oxygen regulation system, and is dimensionless ( / (times / min, %)). It serves as the reference benchmark for the normal state of data without induced factors. The discrete cluster center refers to the midpoint of the mean range of SpO2 / HR time-series sample data under cerebral edema conditions at high altitude. It is the core cluster center in K-Means clustering that represents the abnormal and unstable state of the central oxygen regulation system in the human body, and is dimensionless ( / (times / min, %)). It serves as the reference benchmark for the pathological abnormal state of data without induced factors. The preset K-Means clustering algorithm refers to the unsupervised clustering algorithm implemented based on preset stable cluster centers and discrete cluster centers. It does not require iterative optimization and directly completes the clustering of HR / SpO2 time-series data without induced factors according to the two centers, which is suitable for the low computing power requirements of wearable devices. Dispersion refers to the degree of dispersion of the distribution of HR / SpO2 time series data without causal factors relative to the corresponding cluster centers after clustering. It is dimensionless and is a core indicator for quantifying the fluctuation characteristics of data without causal factors and determining the state of cerebral edema. The higher the dispersion, the more significant the abnormality of the central oxygen regulation system.

[0077] Specifically, firstly, identifying high-altitude clinical dual-cluster centers can include the following steps: 1. Collect clinical data without induced factors at high altitude: Collect time-series data of heart rate (HR) and SpO2 without induced factors under normal clinical conditions and cerebral edema conditions at high altitude. The sample size is no less than 500 groups, and all sample data are without induced factors (excluding exercise, temperature and altitude), which meets the requirements of clinical statistical significance. 2. Statistical sample data mean range: The mean and mean range of HR and SpO2 sample data without cause under normal clinical conditions at high altitude were calculated respectively. At the same time, the mean and mean range of HR and SpO2 sample data without cause under cerebral edema conditions were calculated. The mean range was determined according to the clinical 95% confidence interval to fit the numerical distribution characteristics of physiological / pathological conditions at high altitude. 3. The midpoints of the mean ranges of HR time-series data and SpO2 time-series data under normal clinical conditions at high altitudes were respectively designated as the centers of stable HR clusters (denoted as ). SpO2 stable cluster center (denoted as) The midpoints of the mean ranges of the time-series HR data and the SpO2 data in the clinical cerebral edema state at high altitudes without any precipitating factors were respectively determined as the centers of the discrete HR clusters (denoted as ). SpO2 discrete cluster center (denoted as) The four cluster centers mentioned above are preset into the clustering algorithm as fixed benchmarks for subsequent unincentivized data clustering.

[0078] Then, K-Means clustering is performed based on the preset bi-cluster centers, which may include the following steps: 1. Cluster distance calculation: For each data point in the standardized, unincentivized HR time series data... Calculate its relationship with the HR stable cluster center. HR Discrete Cluster Center The Euclidean distance; following the same rules, calculate the Euclidean distance for each data point in the uninduced SpO2 time series data. With SpO2 stable cluster center SpO2 discrete cluster center The Euclidean distance, the formula for Euclidean distance is:

[0079] in, is the Euclidean distance between the data point and the cluster center, which is dimensionless and represents the similarity between the data point and the corresponding state cluster center; These are standardized, unincentivized data points. The standardized cluster center (stable / discrete).

[0080] 2. Clustering Assignment Determination: For each uninduced HR / SpO2 data point, the clustering assignment is determined according to the minimum distance principle—if the data point has a smaller Euclidean distance to the center of a stable cluster, it is assigned to a stable cluster; if it has a smaller Euclidean distance to the center of a discrete cluster, it is assigned to a discrete cluster. This completes the independent clustering of the uninduced HR and SpO2 time series data, and yields the clustering results for the two types of data respectively.

[0081] Finally, calculating the dispersion of the uninduced HR / SpO2 time series data can involve the following steps: 1. Determine the cluster centers: Based on the clustering results, determine the cluster centers (stable cluster centers) to which the entire HR time series data belongs without any causal factors. or discrete cluster center The cluster centers to which the entire SpO2 time series data belongs without any causal factors (stable cluster centers) or discrete cluster center ); 2. Calculate the dispersion: For the uninduced HR and SpO2 time series data, calculate the dispersion relative to the cluster centers using the following formula:

[0082] in, The dispersion of HR / SpO2 time series data without causal factors is dimensionless. The larger the value, the more dispersed the data is relative to the cluster center, and the more unstable the oxygen regulation system is. The total number of data points in the uninduced HR / SpO2 time series data; This represents the i-th data point without any known cause after standardization. The standardized data of this type is assigned to a cluster center (stable / discrete).

[0083] In one optional embodiment of this application, the method further includes: The target monitoring object is acquired at preset time intervals for a first preset duration prior to the current moment, including HR time series data, SpO2 time series data, acceleration time series data, temperature fluctuation time series data, and altitude fluctuation time series data. Based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data, and the acceleration time series data, the HR time series data and the SpO2 time series data are filtered to obtain uninduced HR time series data and uninduced SpO2 time series data. Clustering is performed on the uninduced HR time series data and the uninduced SpO2 time series data respectively to obtain the dispersion of the uninduced HR time series data and the uninduced SpO2 time series data. If the larger of the two dispersions is not less than a preset dispersion threshold, a cerebral edema warning is issued.

[0084] Specifically, in actual high-altitude scenarios, the target monitoring object may exhibit the following situation: no symptoms of altitude sickness but already showing early-stage cerebral edema. To avoid missing this situation, the following processing can be performed at preset time intervals: Based on at least one of the temperature fluctuation time-series data, the altitude fluctuation time-series data, and the acceleration time-series data, the HR time-series data and the SpO2 time-series data are filtered to obtain causeless HR time-series data and causeless SpO2 time-series data; the causeless HR time-series data and the causeless SpO2 time-series data are clustered respectively to obtain the dispersion of each of the causeless HR time-series data and the causeless SpO2 time-series data; if the larger of the two dispersions is not less than a preset dispersion threshold, a cerebral edema warning is issued.

[0085] Figure 2 A structural block diagram of a vital signs monitoring system provided in this application embodiment is shown below. Figure 2 As shown, the system includes: The time-series data acquisition module 201 is used to acquire the heart rate (HR), blood oxygen saturation (SpO2), and acceleration time-series data of the target monitoring object for a first preset duration before the current monitoring time when a warning of altitude sickness symptoms is received from the target monitoring object. The MLE acquisition module 202 is used to acquire motion weights at each acquisition time based on the acceleration time series data, acquire a preset embedding dimension based on the degrees of freedom of the human oxygen regulation system, and acquire a preset time delay based on the fluctuation period of the human oxygen regulation system; it is used to reconstruct the phase space of the HR time series data and SpO2 time series data based on the preset embedding dimension and the preset time delay, and then calculate the maximum Lyapunov exponent MLE of the reconstructed phase space using the minimum data amount method in combination with the motion weights of each spatial vector in the reconstructed phase space; wherein, the motion weight of each spatial vector in the reconstructed phase space corresponds to the motion weight of its core acquisition time; The altitude sickness screening module 203 is used to acquire temperature fluctuation time series data and altitude fluctuation time series data for the first preset duration if the MLE indicates that the human oxygen regulation system of the target monitoring object is unstable, and to filter the HR time series data and the SpO2 time series data based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data and the acceleration time series data to obtain causeless HR time series data and causeless SpO2 time series data; The cerebral edema screening module 204 is used to cluster the uninduced HR time series data and the uninduced SpO2 time series data respectively, and obtain the dispersion of the uninduced HR time series data and the uninduced SpO2 time series data respectively. If the larger of the two dispersions is not less than a preset dispersion threshold, a cerebral edema warning is issued.

[0086] The proposed solution, upon receiving an alert indicating altitude sickness symptoms in a target monitored object, acquires heart rate (HR) time-series data, blood oxygen saturation (SpO2) time-series data, and acceleration time-series data of the target monitored object for a first preset duration prior to the current monitoring time. Based on the acceleration time-series data, it acquires motion weights for each acquisition moment; based on the degrees of freedom of the human oxygen regulation system, it acquires a preset embedding dimension; and based on the fluctuation period of the human oxygen regulation system, it acquires a preset time delay. Based on the preset embedding dimension and the preset time delay, it reconstructs the phase space of the HR and SpO2 time-series data. Then, combining the motion weights of each spatial vector in the reconstructed phase space, it calculates the maximum Lyapunov exponent M of the reconstructed phase space using the minimum data amount method. If the MLE indicates that the human oxygen regulation system of the target monitoring object is unstable, then the temperature fluctuation time series data and altitude fluctuation time series data of the first preset duration are acquired, and the HR time series data and SpO2 time series data are filtered based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data and the acceleration time series data to obtain causeless HR time series data and causeless SpO2 time series data; the causeless HR time series data and causeless SpO2 time series data are clustered respectively to obtain the dispersion of the causeless HR time series data and causeless SpO2 time series data respectively; if the larger of the two dispersions is not less than a preset dispersion threshold, then a cerebral edema warning is issued. This scheme effectively eliminates external interferences such as motion, temperature, and altitude by collecting heart rate, blood oxygen, and acceleration data, combined with phase space reconstruction and MLE calculation, and can accurately distinguish between ordinary altitude sickness and early high-altitude cerebral edema. At the same time, through causeless data screening and cluster dispersion determination, it achieves objective quantification of central instability state, significantly reducing false positives and false negatives. The overall method has low computational power, can run continuously in real time, is compatible with wearable devices, and can provide early, rapid, and reliable automatic early warning of high-altitude cerebral edema in high-altitude scenarios.

[0087] In one optional embodiment of this application, obtaining the motion weights at each acquisition time based on the acceleration time-series data includes: For each acquisition moment, if the acceleration timing value at the acquisition moment is not greater than a first preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 1; if the acceleration timing value at the acquisition moment is greater than the first preset acceleration threshold and not greater than a second preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 0.3; if the acceleration timing value at the acquisition moment is greater than the second preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 0; wherein, the first preset acceleration threshold is less than the second preset acceleration threshold; The motion weights of each spatial vector in the reconstructed phase space are combined to calculate the MLE of the reconstructed phase space using the minimum data amount method: Determine the nearest neighbor of each spatial vector in the reconstructed phase space, and determine multiple time steps; For each time step, obtain the Euclidean distance between each spatial vector and its nearest neighbor at that time step, take the logarithm of each Euclidean distance to obtain the logarithmic Euclidean distance, and then perform a weighted average of the logarithmic Euclidean distances corresponding to each spatial vector based on the corresponding motion weights to obtain the weighted average logarithmic Euclidean distance corresponding to that time step. The MLE is obtained by linear fitting with each time step as the independent variable and each weighted average log Euclidean distance as the dependent variable.

[0088] In one optional embodiment of this application, obtaining the motion weights at each acquisition time based on the acceleration time-series data includes: For each acquisition moment, if the acceleration timing value at the acquisition moment is not greater than a first preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 1; if the acceleration timing value at the acquisition moment is greater than the first preset acceleration threshold and not greater than a second preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 0.3; if the acceleration timing value at the acquisition moment is greater than the second preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 0; wherein, the first preset acceleration threshold is less than the second preset acceleration threshold; The motion weights of each spatial vector in the reconstructed phase space are combined to calculate the MLE of the reconstructed phase space using the minimum data amount method: Determine the nearest neighbor of each spatial vector in the reconstructed phase space, and determine multiple time steps; For each time step, obtain the Euclidean distance between each spatial vector and its nearest neighbor at that time step, take the logarithm of each Euclidean distance to obtain the logarithmic Euclidean distance, and then perform a weighted average of the logarithmic Euclidean distances corresponding to each spatial vector based on the corresponding motion weights to obtain the weighted average logarithmic Euclidean distance corresponding to that time step. The MLE is obtained by linear fitting with each time step as the independent variable and each weighted average log Euclidean distance as the dependent variable.

[0089] In one optional embodiment of this application, the system further includes a human-like modulation system feature extraction module, used for: Acquire a first preset number of high-altitude clinical time-series sample data, and obtain the corresponding fluctuation period and degrees of freedom of the human oxygen regulation system based on the high-altitude clinical time-series sample data; the high-altitude clinical time-series sample data includes HR time-series data and SpO2 time-series sample data; The method of obtaining a preset embedding dimension based on the degrees of freedom of the human oxygen regulation system and obtaining a preset time delay based on the fluctuation period of the human oxygen regulation system includes: Based on the degrees of freedom of the human oxygen regulation system, the Takens embedding theorem is used to determine a first traversal interval containing multiple initial embedding dimensions. The median value of the fluctuation period of the human oxygen regulation system is used as a temporary time delay. The phase space of the high-altitude clinical time series sample data is reconstructed based on each initial embedding parameter of the first traversal interval and the temporary time delay. The root mean square error of the spatial vector in each reconstructed phase space and the high-altitude clinical time series sample data is obtained. The initial embedding dimension corresponding to the minimum root mean square error is determined as the preset embedding dimension. Based on the fluctuation cycle of the human oxygen regulation system, a second traversal interval containing multiple initial time delays is determined. The phase space of the high-altitude clinical time series sample data is reconstructed based on each initial time delay of the second traversal interval and the preset embedding dimension. The root mean square error of the spatial vector in each reconstructed phase space corresponding to the high-altitude clinical time series sample data is obtained, and the initial time delay corresponding to the minimum root mean square error is determined as the preset time delay.

[0090] In an optional embodiment of this application, the high-intensity rejection screening module is further used for: Based on the HR time-series data, SpO2 time-series data, and acceleration time-series data of the target monitoring object when it enters different preset altitude ranges and does not show symptoms of altitude sickness, the MLE baseline of the target monitoring object in different preset altitude ranges is obtained. Based on the time-series sample data of clinical altitude sickness but no cerebral edema in different preset altitude ranges, a second preset number of time-series samples are obtained to obtain the MLE offset threshold corresponding to different preset altitude ranges. The MLE offset threshold indicates the presence of altitude sickness but no cerebral edema in the corresponding altitude range. If the MLE is not greater than the sum of the MLE baseline and the MLE offset threshold in the corresponding altitude range, it indicates that the target monitoring object's human oxygen regulation system is not unstable, and that the target monitoring object only has altitude sickness and no cerebral edema, and an altitude sickness warning is issued. If the MLE is greater than the sum of the MLE baseline and the MLE offset threshold for the corresponding altitude range, it indicates that the target monitoring object's human oxygen regulation system is unstable, and that the target monitoring object has altitude sickness and needs further monitoring for cerebral edema.

[0091] In one optional embodiment of this application, the HR time series data and the SpO2 time series data are filtered based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data, and the acceleration time series data to obtain causeless HR time series data and causeless SpO2 time series data: If the time series value at any acquisition time in the temperature fluctuation time series data is not less than a preset temperature threshold, the time series value at any acquisition time in the altitude fluctuation time series data is not less than a preset altitude threshold, or the time series value at any acquisition time in the acceleration time series data is not less than a third preset acceleration threshold, then the time series value of the HR time series data and the time series value of the SpO2 time series data corresponding to the acquisition time are removed.

[0092] In one optional embodiment of this application, the step of clustering the uninduced HR time-series data and the uninduced SpO2 time-series data respectively to obtain the dispersion of the uninduced HR time-series data and the uninduced SpO2 time-series data respectively includes: The midpoint of the mean range of SpO2 time series data without cause and the midpoint of the mean range of HR time series data without cause under normal clinical conditions at high altitude were taken as their respective stable cluster centers, and the midpoint of the mean range of SpO2 time series data without cause and the midpoint of the mean range of HR time series data without cause under clinical cerebral edema conditions at high altitude were taken as their respective discrete cluster centers. Based on the stable cluster centers and the discrete cluster centers, the discreteness of the uninduced HR time series data and the uninduced SpO2 time series data is obtained by using a preset K-Means clustering algorithm.

[0093] In one optional embodiment of this application, the system further includes a cerebral edema prevention module, used for: The target monitoring object is acquired at preset time intervals for a first preset duration prior to the current moment, including HR time series data, SpO2 time series data, acceleration time series data, temperature fluctuation time series data, and altitude fluctuation time series data. Based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data, and the acceleration time series data, the HR time series data and the SpO2 time series data are filtered to obtain uninduced HR time series data and uninduced SpO2 time series data. Clustering is performed on the uninduced HR time series data and the uninduced SpO2 time series data respectively to obtain the dispersion of the uninduced HR time series data and the uninduced SpO2 time series data. If the larger of the two dispersions is not less than a preset dispersion threshold, a cerebral edema warning is issued.

[0094] Figure 3 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 3 As shown, the electronic device may include: a processor 310, a communication interface 320, a memory 330, and a communication bus 340. The processor 310, communication interface 320, and memory 330 communicate with each other via the communication bus 340. The processor 310 can call logical instructions in the memory 330 to execute a vital signs monitoring method. This method includes: upon receiving a warning of altitude sickness symptoms in the target monitored object, acquiring heart rate (HR) time-series data, blood oxygen saturation (SpO2) time-series data, and acceleration time-series data of the target monitored object for a first preset duration before the current monitoring time; acquiring motion weights at each acquisition time based on the acceleration time-series data; acquiring a preset embedding dimension based on the degrees of freedom of the human oxygen regulation system; acquiring a preset time delay based on the fluctuation period of the human oxygen regulation system; reconstructing the phase space of the HR time-series data and the SpO2 time-series data based on the preset embedding dimension and the preset time delay; and then calculating the maximum Lyapunov exponent (MLE) of the reconstructed phase space using the minimum data amount method, combining the motion weights of each spatial vector in the reconstructed phase space. The motion weight of each spatial vector in the reconstructed phase space corresponds to the motion weight at its core acquisition time. If the MLE indicates that the human oxygen regulation system of the target monitoring object is unstable, then the temperature fluctuation time series data and altitude fluctuation time series data of the first preset duration are acquired. Based on at least one of the temperature fluctuation time series data, altitude fluctuation time series data and acceleration time series data, the HR time series data and SpO2 time series data are filtered to obtain causeless HR time series data and causeless SpO2 time series data. The causeless HR time series data and causeless SpO2 time series data are clustered respectively to obtain the dispersion of the causeless HR time series data and causeless SpO2 time series data. If the larger of the two dispersions is not less than a preset dispersion threshold, then a cerebral edema warning is issued.

[0095] Furthermore, the logical instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0096] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the vital sign monitoring method provided by the above methods. The method includes: acquiring heart rate (HR), blood oxygen saturation (SpO2), and acceleration time-series data of the target monitored object for a first preset duration before the current monitoring time when receiving an early warning of altitude sickness symptoms in the target monitored object; acquiring motion weights at each acquisition time based on the acceleration time-series data; acquiring a preset embedding dimension based on the degrees of freedom of the human oxygen regulation system; acquiring a preset time delay based on the fluctuation period of the human oxygen regulation system; reconstructing the phase space of the HR time-series data and the SpO2 time-series data based on the preset embedding dimension and the preset time delay; and then combining the motion of each spatial vector in the reconstructed phase space. The motion weights are calculated using the minimum data volume method to obtain the maximum Lyapunov exponent (MLE) of the reconstructed phase space; wherein, the motion weight of each spatial vector in the reconstructed phase space corresponds to the motion weight at its core acquisition time; if the MLE indicates that the human oxygen regulation system of the target monitoring object is unstable, then the temperature fluctuation time series data and altitude fluctuation time series data of the first preset duration are acquired, and based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data and the acceleration time series data, the HR time series data and the SpO2 time series data are filtered to obtain the causeless HR time series data and the causeless SpO2 time series data; the causeless HR time series data and the causeless SpO2 time series data are clustered respectively to obtain the dispersion of the causeless HR time series data and the causeless SpO2 time series data respectively; if the larger of the two dispersions is not less than a preset dispersion threshold, then a cerebral edema warning is issued.

[0097] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the vital sign monitoring method provided by the above methods. This method includes: upon receiving a warning of altitude sickness symptoms in a target monitored object, acquiring heart rate (HR) time-series data, blood oxygen saturation (SpO2) time-series data, and acceleration time-series data of the target monitored object for a first preset duration prior to the current monitoring time; acquiring motion weights for each acquisition time based on the acceleration time-series data; acquiring a preset embedding dimension based on the degrees of freedom of the human oxygen regulation system; acquiring a preset time delay based on the fluctuation period of the human oxygen regulation system; reconstructing the phase space of the HR time-series data and the SpO2 time-series data based on the preset embedding dimension and the preset time delay; and then calculating the vital signs monitoring method using the minimum data amount method in combination with the motion weights of each spatial vector in the reconstructed phase space. The maximum Lyapunov exponent (MLE) of the reconstructed phase space is used; where the motion weight of each spatial vector in the reconstructed phase space corresponds to the motion weight at its core acquisition time; if the MLE indicates instability of the human oxygen regulation system of the target monitoring object, temperature fluctuation time series data and altitude fluctuation time series data of the first preset duration are acquired, and based on at least one of the temperature fluctuation time series data, altitude fluctuation time series data and acceleration time series data, the HR time series data and SpO2 time series data are filtered to obtain causeless HR time series data and causeless SpO2 time series data; the causeless HR time series data and causeless SpO2 time series data are clustered respectively to obtain the dispersion of the causeless HR time series data and causeless SpO2 time series data respectively; if the larger of the two dispersions is not less than a preset dispersion threshold, a cerebral edema warning is issued.

[0098] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0099] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for monitoring vital signs, characterized in that, include: When a warning is received that the target monitoring object has symptoms of altitude sickness, the heart rate (HR), blood oxygen saturation (SpO2), and acceleration time series data of the target monitoring object are obtained for the first preset time period before the current monitoring time. The motion weights at each acquisition time are obtained based on the acceleration time series data, the preset embedding dimension is obtained based on the degrees of freedom of the human oxygen regulation system, and the preset time delay is obtained based on the fluctuation period of the human oxygen regulation system. Based on the preset embedding dimension and the preset time delay, the HR time series data and the SpO2 time series data are reconstructed in phase space. Then, the maximum Lyapunov exponent MLE of the reconstructed phase space is calculated using the minimum data volume method, combined with the motion weights of each spatial vector in the reconstructed phase space. The motion weight of each spatial vector in the reconstructed phase space corresponds to the motion weight at its core acquisition time. If the MLE indicates that the human oxygen regulation system of the target monitoring object is unstable, then the temperature fluctuation time series data and altitude fluctuation time series data of the first preset duration are acquired, and the HR time series data and SpO2 time series data are filtered based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data and the acceleration time series data to obtain the causeless HR time series data and the causeless SpO2 time series data. Clustering is performed on the uninduced HR time series data and the uninduced SpO2 time series data respectively to obtain the dispersion of the uninduced HR time series data and the uninduced SpO2 time series data. If the larger of the two dispersions is not less than a preset dispersion threshold, a cerebral edema warning is issued.

2. The method according to claim 1, characterized in that, The process of obtaining motion weights at each acquisition time based on the acceleration time-series data includes: For each acquisition moment, if the acceleration timing value at the acquisition moment is not greater than a first preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 1; if the acceleration timing value at the acquisition moment is greater than the first preset acceleration threshold and not greater than a second preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 0.3; if the acceleration timing value at the acquisition moment is greater than the second preset acceleration threshold, then the motion weight of the HR timing value and SpO2 timing value at the acquisition moment is determined to be 0; wherein, the first preset acceleration threshold is less than the second preset acceleration threshold; The motion weights of each spatial vector in the reconstructed phase space are combined to calculate the MLE of the reconstructed phase space using the minimum data amount method: Determine the nearest neighbor of each spatial vector in the reconstructed phase space, and determine multiple time steps; For each time step, obtain the Euclidean distance between each spatial vector and its nearest neighbor at that time step, take the logarithm of each Euclidean distance to obtain the logarithmic Euclidean distance, and then perform a weighted average of the logarithmic Euclidean distances corresponding to each spatial vector based on the corresponding motion weights to obtain the weighted average logarithmic Euclidean distance corresponding to that time step. The MLE is obtained by linear fitting with each time step as the independent variable and each weighted average log Euclidean distance as the dependent variable.

3. The method according to claim 1, characterized in that, The method further includes: Acquire a first preset number of high-altitude clinical time-series sample data, and obtain the corresponding fluctuation period and degrees of freedom of the human oxygen regulation system based on the high-altitude clinical time-series sample data; the high-altitude clinical time-series sample data includes HR time-series data and SpO2 time-series sample data; The method of obtaining a preset embedding dimension based on the degrees of freedom of the human oxygen regulation system and obtaining a preset time delay based on the fluctuation period of the human oxygen regulation system includes: Based on the degrees of freedom of the human oxygen regulation system, the Takens embedding theorem is used to determine a first traversal interval containing multiple initial embedding dimensions. The median value of the fluctuation period of the human oxygen regulation system is used as a temporary time delay. The phase space of the high-altitude clinical time series sample data is reconstructed based on each initial embedding parameter of the first traversal interval and the temporary time delay. The root mean square error of the spatial vector in each reconstructed phase space and the high-altitude clinical time series sample data is obtained. The initial embedding dimension corresponding to the minimum root mean square error is determined as the preset embedding dimension. Based on the fluctuation cycle of the human oxygen regulation system, a second traversal interval containing multiple initial time delays is determined. The phase space of the high-altitude clinical time series sample data is reconstructed based on each initial time delay of the second traversal interval and the preset embedding dimension. The root mean square error of the spatial vector in each reconstructed phase space corresponding to the high-altitude clinical time series sample data is obtained, and the initial time delay corresponding to the minimum root mean square error is determined as the preset time delay.

4. The method according to claim 1, characterized in that, The method further includes: Based on the HR time-series data, SpO2 time-series data, and acceleration time-series data of the target monitoring object when it enters different preset altitude ranges and does not show symptoms of altitude sickness, the MLE baseline of the target monitoring object in different preset altitude ranges is obtained. Based on the time-series sample data of clinical altitude sickness but no cerebral edema in different preset altitude ranges, a second preset number of time-series samples are obtained to obtain the MLE offset threshold corresponding to different preset altitude ranges. The MLE offset threshold indicates the presence of altitude sickness but no cerebral edema in the corresponding altitude range. If the MLE is not greater than the sum of the MLE baseline and the MLE offset threshold in the corresponding altitude range, it indicates that the target monitoring object's human oxygen regulation system is not unstable, and that the target monitoring object only has altitude sickness and no cerebral edema, and an altitude sickness warning is issued. If the MLE is greater than the sum of the MLE baseline and the MLE offset threshold for the corresponding altitude range, it indicates that the target monitoring object's human oxygen regulation system is unstable, and that the target monitoring object has altitude sickness and needs further monitoring for cerebral edema.

5. The method according to claim 1, characterized in that, The HR time series data and SpO2 time series data are filtered based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data, and the acceleration time series data to obtain cause-free HR time series data and cause-free SpO2 time series data. If the time series value at any acquisition time in the temperature fluctuation time series data is not less than a preset temperature threshold, the time series value at any acquisition time in the altitude fluctuation time series data is not less than a preset altitude threshold, or the time series value at any acquisition time in the acceleration time series data is not less than a third preset acceleration threshold, then the time series value of the HR time series data and the time series value of the SpO2 time series data corresponding to the acquisition time are removed.

6. The method according to claim 1, characterized in that, The step of clustering the uninduced HR time-series data and the uninduced SpO2 time-series data respectively to obtain the dispersion of the uninduced HR time-series data and the uninduced SpO2 time-series data includes: The midpoint of the mean range of SpO2 time series data without cause and the midpoint of the mean range of HR time series data without cause under normal clinical conditions at high altitude were taken as their respective stable cluster centers, and the midpoint of the mean range of SpO2 time series data without cause and the midpoint of the mean range of HR time series data without cause under clinical cerebral edema conditions at high altitude were taken as their respective discrete cluster centers. Based on the stable cluster centers and the discrete cluster centers, the discreteness of the uninduced HR time series data and the uninduced SpO2 time series data is obtained by using a preset K-Means clustering algorithm.

7. The method according to claim 1, characterized in that, The method further includes: The target monitoring object is acquired at preset time intervals for a first preset duration prior to the current moment, including HR time series data, SpO2 time series data, acceleration time series data, temperature fluctuation time series data, and altitude fluctuation time series data. Based on at least one of the temperature fluctuation time series data, the altitude fluctuation time series data, and the acceleration time series data, the HR time series data and the SpO2 time series data are filtered to obtain uninduced HR time series data and uninduced SpO2 time series data. Clustering is performed on the uninduced HR time series data and the uninduced SpO2 time series data respectively to obtain the dispersion of the uninduced HR time series data and the uninduced SpO2 time series data. If the larger of the two dispersions is not less than a preset dispersion threshold, a cerebral edema warning is issued.

8. A vital signs monitoring system, characterized in that, include: The time-series data acquisition module is used to acquire the heart rate (HR), blood oxygen saturation (SpO2), and acceleration time-series data of the target monitoring object for a first preset duration before the current monitoring time when an early warning of altitude sickness symptoms is received. The MLE acquisition module is used to acquire motion weights at each acquisition time based on the acceleration time series data, acquire a preset embedding dimension based on the degrees of freedom of the human oxygen regulation system, and acquire a preset time delay based on the fluctuation period of the human oxygen regulation system. This is used to reconstruct the phase space of the HR time series data and SpO2 time series data based on the preset embedding dimension and the preset time delay, and then calculate the maximum Lyapunov exponent MLE of the reconstructed phase space using the minimum data volume method in combination with the motion weights of each spatial vector in the reconstructed phase space; wherein, the motion weight of each spatial vector in the reconstructed phase space corresponds to the motion weight of its core acquisition time. The altitude sickness screening module is used to acquire temperature fluctuation time-series data and altitude fluctuation time-series data for the first preset duration if the MLE indicates that the human oxygen regulation system of the target monitoring object is unstable. Based on at least one of the temperature fluctuation time-series data, the altitude fluctuation time-series data and the acceleration time-series data, the module filters the HR time-series data and the SpO2 time-series data to obtain causeless HR time-series data and causeless SpO2 time-series data. The cerebral edema screening module is used to cluster the uninduced HR time series data and the uninduced SpO2 time series data respectively, and obtain the dispersion of the uninduced HR time series data and the uninduced SpO2 time series data respectively. If the larger of the two dispersions is not less than a preset dispersion threshold, a cerebral edema warning is issued.

9. An electronic device 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 program, it implements the method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 7.

Images