Early warning system for acute exacerbation of chronic obstructive pulmonary disease based on wearable device

Abstract

The present application relates to the technical field of personal health risk assessment, and particularly relates to a wearable device-based early intelligent warning system for acute exacerbation of chronic obstructive pulmonary disease, which comprises a data acquisition module and a warning module, the data acquisition module is used for acquiring activity load data and heart rate data of a patient in an activity period and an adjacent rest period after the activity period based on a wearable device, the warning module is used for obtaining cumulative activity load and cumulative heart rate response from the activity load data and the heart rate data respectively, so as to determine a cardiopulmonary response delay degree, according to the cardiopulmonary response delay degree and an activity intensity feature, a comprehensive airway obstruction feature is obtained, and according to the comprehensive airway obstruction feature, early warning of acute exacerbation of chronic obstructive pulmonary disease of the patient is performed, so that the warning system can realize fine judgment, greatly reduce the false positive rate, and can capture weak pathological signals based on dynamic response, and realize continuous, non-invasive and intelligent early warning.

Description

Technical Field

[0001] This invention relates to the field of personal health risk assessment technology, specifically to an early intelligent warning system for acute exacerbations of COPD based on wearable devices. Background Technology

[0002] In the early stages of acute exacerbations, the pathological changes in patients with chronic obstructive pulmonary disease (COPD) are mainly characterized by increased inflammation and mucus secretion in the small airways, leading to mechanical instability and air trapping in the airway lumen. These pathological changes cause patients to experience end-expiratory airflow limitation and dynamic lung hyperinflation even during low-intensity exercise. Currently, daily monitoring of this disease relies primarily on patients' subjective symptom reports or regular clinical pulmonary function tests. However, subjective reports often lag behind actual physiological decline, while clinical tests, due to their low frequency and specific cooperation requirements, are difficult to capture sudden changes in the condition during daily life.

[0003] Existing wearable monitoring technologies attempt to provide early warnings using indicators such as step count, heart rate, or blood oxygen saturation. However, these indicators suffer from insufficient specificity in non-standardized daily activity scenarios. Daily activities are fragmented, short-term, and of random intensity. During low-intensity activities, weak pathological signals are easily masked by environmental noise; while during high-intensity activities, physiological increases in heart rate and shortness of breath can be easily confused with pathological airway obstruction, thus affecting the accuracy and reliability of early warnings for acute exacerbations of COPD. Summary of the Invention

[0004] To address the low accuracy of existing early warning methods for acute exacerbations of COPD, the present invention aims to provide an intelligent early warning system for acute exacerbations of COPD based on wearable devices. The specific technical solution adopted is as follows: This invention provides an early intelligent warning system for acute exacerbations of COPD based on wearable devices, comprising: The data acquisition module is used to acquire the patient's activity load data and heart rate data within a target time period based on a wearable device. The target time period includes the activity period and the adjacent resting period thereafter. The early warning module, electrically connected to the data acquisition module, is configured as follows: The cumulative activity load and cumulative heart rate response for the target time period are obtained from the activity load data and heart rate data, respectively. According to the preset percentile sequence, the load percentile time and response percentile time for each completion of the preset percentile are determined by the cumulative activity load and the cumulative heart rate response, respectively. The degree of cardiopulmonary response delay is determined by the lag time between the response percentile time and the corresponding load percentile time. Based on the cardiopulmonary response delay and activity intensity characteristics, a comprehensive airway obstruction characteristic is obtained; the activity intensity characteristic is obtained from the cumulative activity load and the duration of the activity period. Based on the comprehensive characteristics of airway obstruction, early warning of acute exacerbations of COPD can be provided for patients.

[0005] In an exemplary embodiment, the process of obtaining the target time period includes: Determine whether the patient's activity intensity exceeds a preset intensity threshold within each time period; The target time period is obtained when the activity intensity of at least a first preset number of unit time periods is greater than a preset intensity threshold, and the activity intensity of at least a second preset number of unit time periods is less than or equal to the preset intensity threshold; wherein, the last unit time period in the at least a first preset number of unit time periods is the time period preceding the first unit time period in the at least a second preset number of unit time periods, the start time of the first unit time period in the at least a first preset number of unit time periods is taken as the start time of the activity period, and the end time of the last unit time period in the at least a first preset number of unit time periods is taken as the end time of the activity period; the start time of the first unit time period in the at least a second preset number of unit time periods is taken as the start time of the rest period, and the end time of the last unit time period in the at least a second preset number of unit time periods is taken as the end time of the rest period.

[0006] In one exemplary embodiment, the active load data is the signal vector magnitude of the triaxial acceleration components; The process of obtaining the cumulative active load includes: Determine the degree to which the signal vector magnitude exceeds the gravitational acceleration at each moment during the activity period; The cumulative activity load is obtained by integrating the degree of acceleration exceeding the limit at each moment during the activity period. The process of obtaining the cumulative heart rate response includes: Determine the extent to which the heart rate exceeds the preset resting heart rate at each moment during the resting period; The cumulative heart rate response is obtained by integrating the degree of heart rate exceedance at each moment during the resting period.

[0007] In an exemplary embodiment, the process of obtaining the load percentile time includes: Based on the cumulative active load and the candidate preset percentile's index in the preset percentile sequence, a target cumulative load threshold corresponding to the candidate preset percentile is obtained; the candidate preset percentile is any preset percentile in the preset percentile sequence. The integration begins at the start of the activity period and ends at each moment of the activity period. The degree of acceleration exceeding the threshold is integrated, and the integration result is compared with the target cumulative load threshold. Integration ends when the integration result is greater than or equal to the target cumulative load threshold for the first time. The integration end time corresponding to the integration result at the end of the integration is used as the load percentile time corresponding to the candidate preset percentile.

[0008] In an exemplary embodiment, the process of obtaining the response percentile time includes: Based on the cumulative heart rate response and the candidate preset percentile's index in the preset percentile sequence, the target cumulative response threshold corresponding to the candidate preset percentile is obtained. The heart rate excess is integrated using the start time of the resting period as the start time of integration, and each moment of the resting period is used as the end time of integration. The integration result is compared with the target cumulative response threshold each time, and the integration ends when the integration result is greater than or equal to the target cumulative response threshold for the first time. The integration end time corresponding to the integration result at the end of the integration is used as the response percentile time corresponding to the candidate preset percentile.

[0009] In an exemplary embodiment, the process of obtaining the target cumulative load threshold includes: The quantile of the candidate preset percentile is obtained from the index of the candidate preset percentile in the preset percentile sequence; Calculate the product of the quantile and the cumulative active load to obtain the target cumulative load threshold corresponding to the candidate preset percentile; The process of obtaining the target cumulative response threshold includes: Calculate the product of the percentile and the cumulative heart rate response to obtain the target cumulative response threshold corresponding to the candidate preset percentile.

[0010] In an exemplary embodiment, the process of obtaining the comprehensive airway obstruction characteristics includes: The degree of cardiopulmonary response delay is used to determine the degree of expiratory response fluctuation and the degree of gas trapping at the end of expiration. Based on the degree of fluctuation in the expiratory response and the activity intensity characteristics, the expiratory resistance fluctuation characteristics are obtained; the expiratory resistance fluctuation characteristics are positively correlated with the degree of fluctuation in the expiratory response and negatively correlated with the activity intensity characteristics. By integrating the degree of gas trapping at the end of expiration and the fluctuation characteristics of expiratory resistance, the comprehensive characteristics of airway obstruction are obtained.

[0011] In an exemplary embodiment, the process of obtaining the degree of fluctuation in the expiratory response includes: Obtain the second half of the sequence in the cardiopulmonary response delay sequence corresponding to the preset percentile sequence; Determine the first-order difference absolute value sequence of the latter half of the sequence; The degree of fluctuation in the expiratory response is obtained by integrating the differences between each first-order difference absolute value in the first-order difference absolute value sequence and its corresponding local average value.

[0012] In one exemplary embodiment, the degree of gas trapping at the end of expiration is the median of the latter half of the sequence.

[0013] In one exemplary embodiment, the activity intensity characteristic is the ratio of the cumulative activity load to the duration of the activity period.

[0014] This invention offers the following advantages: It analyzes the activity period plus adjacent resting periods, treating activity as a controllable stress stimulus and resting heart rate changes as the system's response to this stimulus, thereby improving the accuracy of subsequent early warnings. By accumulating activity load and cumulative heart rate response, it determines the load percentile and response percentile times for each completed preset percentile. This allows the invention to focus not on the exact heart rate but on how heart rate changes follow activity changes. Furthermore, the degree of cardiopulmonary response delay quantifies the efficiency of the cardiopulmonary system's recovery from stress to baseline. In the early stages of acute exacerbations of COPD, airway obstruction and dynamic lung hyperinflation limit the normal decline in heart rate, causing a response lag behind the stress level. The degree of cardiopulmonary response delay effectively distinguishes between physiological and non-physiological conditions. Regarding pathology: During high-intensity activity, although heart rate and respiration physiologically increase, the response delay should remain within the normal range as long as airway function is normal. Only when pathological obstruction exists will this delay be significantly prolonged, thus enabling early warning based on the degree of cardiopulmonary response delay. Simply assessing the cardiopulmonary response delay is insufficient; therefore, combining it with activity intensity characteristics to calibrate and weight the cardiopulmonary response delay results in a more comprehensive and accurate airway obstruction profile. This profile enables early intelligent warning of acute exacerbations of COPD, allowing the warning system to make refined judgments, significantly reducing false alarm rates, and capturing weak, dynamic-response-based pathological signals, achieving continuous, seamless, and intelligent early warning. Attached Figure Description

[0015] Figure 1This is a schematic diagram of the module composition of an early intelligent warning system for acute exacerbations of COPD based on wearable devices, provided in one embodiment of the present invention. Figure 2 This is a data processing flowchart of an early warning module provided in one embodiment of the present invention. Detailed Implementation

[0016] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the specific implementation methods, structures, features, and effects of the present invention are described in detail below with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. All data and information collected in this application have been obtained with full consent.

[0018] This embodiment provides an early intelligent warning system for acute exacerbations of COPD based on wearable devices (hereinafter referred to as the system), used to provide early warning of acute exacerbations of COPD in patients with chronic obstructive pulmonary disease (hereinafter referred to as patients). Figure 1 As shown, the system includes a data acquisition module and an early warning module. The data acquisition module and the early warning module are electrically connected. The connection between the data acquisition module and the early warning module can be wired or wireless (such as, but not limited to, Bluetooth, WiFi, etc.). This embodiment collects patient physiological data through a wearable device, such as a smartwatch with real-time heart rate monitoring and motion monitoring functions. The sampling frequencies for heart rate monitoring and motion monitoring are the same and are sampled synchronously. The sampling frequency is set according to the actual detection needs, such as 25Hz. In this embodiment, a three-axis accelerometer is installed in the wearable device to realize motion monitoring. The wearable device acquires the patient's activity load data and heart rate data. The activity load data is specifically calculated from the three-axis acceleration components. The three-axis accelerometer can acquire the three axial acceleration components of the X-axis, Y-axis, and Z-axis at each moment. Based on the Euclidean norm (i.e., the magnitude of the vector), the signal vector magnitude of the three axial acceleration components of the X-axis, Y-axis, and Z-axis is calculated as the instantaneous activity feature at each moment, i.e., the activity load data at each moment.

[0019] The data acquisition module is used to identify and extract daily activity-recovery time segments that contain the complete exercise load and heart rate recovery process based on the heart rate and acceleration collected by the wearable device. The daily activity-recovery time segments are defined as target time segments. The target time segment consists of two parts: the first part is the activity period (corresponding to the exercise load stage) and the second part is the resting period (corresponding to the heart rate recovery stage). The activity period and the resting period are adjacent in time.

[0020] The following is a specific process for obtaining the target time period: A unit time period is determined, which is a unit time segment with a length much shorter than the target time period. In this embodiment, the length of the unit time period is set to 1 second (where s represents the unit of seconds). The activity intensity of each unit time period in the time series is acquired in real time. For any unit time period, the variance of the instantaneous activity characteristics at each moment within that unit time period is calculated, and the obtained variance value is taken as the activity intensity of that unit time period.

[0021] This embodiment sets a preset intensity threshold, which represents the intensity of the patient's movement in the slow walking state. The preset intensity threshold can be obtained by recording the intensity of the patient's movement in multiple slow walking processes in history, and then calculating the average value. The result is used as the preset intensity threshold.

[0022] The system determines in real time whether the activity intensity of each unit time period is greater than a preset intensity threshold. If it is greater than the preset intensity threshold, it means that the patient's activity intensity in the corresponding unit time period is higher than the intensity of slow walking, and the unit time period is determined to be in the active phase. If it is less than or equal to the preset intensity threshold, it means that the patient's activity intensity in the corresponding unit time period is lower than or equal to the intensity of slow walking, and the unit time period is determined to be in the resting phase.

[0023] This embodiment sets a first preset quantity, which represents the number of time units corresponding to the minimum exercise load duration. As a specific example, the minimum exercise load duration is 45 seconds. Since the time unit mentioned above is 1 second, the first preset quantity can be set to 45. This embodiment also sets a second preset quantity, which represents the number of time units corresponding to the minimum recovery time after exercise. As a specific example, the minimum recovery time is 60 seconds, so the second preset quantity can be set to 60.

[0024] The activity intensity of each time unit in the time sequence is determined to be greater than a preset intensity threshold. A target time period is obtained when the activity intensity of at least a first preset number of time units is greater than the preset intensity threshold, and the activity intensity of at least a second preset number of time units is less than or equal to the preset intensity threshold. Specifically, the first preset number of time units precedes the second preset number of time units, and the last time unit in the first preset number of time units precedes the first time unit in the second preset number of time units.

[0025] The start time of the first unit time period in at least a first preset number of unit time periods shall be used as the start time of the activity period. (Also the start time of the target time period), the end time of the last time period in at least the first preset number of time periods shall be used as the end time of the activity period. The start time of the rest period is defined as the beginning time of the first unit time period in at least a second preset number of unit time periods, and the end time of the rest period is defined as the end time of the last unit time period in at least a second preset number of unit time periods. (This is also the deadline for the target time period). Only the activity period and rest period obtained under the above conditions can constitute the target time period.

[0026] Therefore, when the continuous duration of the active state is greater than or equal to the minimum exercise load duration, and the continuous duration of the resting state is immediately following it, greater than or equal to the minimum recovery duration, it is determined that a daily activity-recovery time segment has occurred. The data acquisition module is used to acquire the patient's activity load data and heart rate data within the target time period. Specifically, it acquires the signal vector magnitude at each moment within the target time period and sorts them chronologically to obtain a signal vector magnitude sequence; it also acquires the heart rate at each moment within the target time period and sorts them chronologically to obtain a heart rate sequence.

[0027] It should be understood that if the duration of the active or resting period in the target time period is too short, that is, if the number of moments contained is too small, this embodiment may not process the target time period and redetermine the target time period.

[0028] The early warning module receives patient physiological data collected by the data acquisition module and then processes the received data signals. The early warning module can be a conventional data processing chip, such as a CPU (Central Processing Unit), or a System on Chip (SoC). The SoC can be configured in a medical monitoring platform or in the patient's smart terminal.

[0029] Figure 2 The implementation steps of the data processing process for the early warning module are as follows: Figure 2 As shown, the data processing steps of the early warning module include: Step S1: Obtain the cumulative activity load and cumulative heart rate response for the target time period from the activity load data and heart rate data, respectively; Step S2: According to the preset percentile sequence, determine the load percentile time and response percentile time for each preset percentile completion based on the cumulative activity load and cumulative heart rate response, respectively; Step S3: Determine the degree of cardiopulmonary response delay by the lag time between the response percentile time and the corresponding load percentile time; Step S4: Obtain comprehensive airway obstruction characteristics based on the degree of cardiopulmonary response delay and activity intensity characteristics; Step S5: Based on the comprehensive characteristics of airway obstruction, provide early warning of acute exacerbations of COPD in patients.

[0030] The following is a detailed explanation of each step.

[0031] Step S1: Obtain the cumulative activity load and cumulative heart rate response for the target time period from the activity load data and heart rate data, respectively.

[0032] It should be understood that the unit of the magnitude of the signal vector of the triaxial acceleration components is the same as the unit of gravitational acceleration, both being g (approximately equal to 9.8). Generally speaking, in most cases, due to the presence of gravity, the magnitude of the signal vector usually fluctuates around 1g (1g represents the value of one gravitational acceleration). When the patient is stationary or moving at a constant speed, the patient is only affected by the Earth's gravity and has no acceleration in other directions. Therefore, the magnitude of the signal vector of the triaxial acceleration component is the same as that of gravitational acceleration, both equal to 1g. When the patient is not moving at a constant speed (e.g., walking, swaying, etc.), in addition to gravity, the patient is also subjected to motion acceleration generated by the motion. In this case, the magnitude of the signal vector of the triaxial acceleration component is higher than that of gravitational acceleration. Furthermore, the larger the magnitude of the signal vector of the triaxial acceleration component is than that of gravitational acceleration, the more intense the patient's movement, i.e., the greater the exercise load.

[0033] Since the acceleration signal generated by exercise and the heart rate signal generated by the heart pumping blood are physical quantities of different dimensions, their amplitudes cannot be directly compared. In order to establish a mapping relationship between the two, it is necessary to calculate their cumulative amounts over a target time period. Therefore, the cumulative activity load of the target time period is obtained from the activity load data within the target time period, and the cumulative heart rate response of the target time period is obtained from the heart rate data within the target time period.

[0034] For obtaining the cumulative activity load: To determine the degree to which the signal vector magnitude of the triaxial acceleration components exceeds gravitational acceleration at each moment during the activity period (after removing the gravitational component and retaining only the pure motion acceleration component), let the number of moments in the activity period be T1, and any moment in the activity period be t1. The calculation method for the degree to which the signal vector magnitude of the triaxial acceleration components exceeds gravitational acceleration at moment t1 during the activity period is as follows: ,in, The magnitude of the signal vector representing the triaxial acceleration components at time t1. Represents gravitational acceleration. This represents the function that takes the maximum value, used to... When the value is less than 0, the acceleration excess level is set to 0.

[0035] Then, the degree of acceleration exceeding the limit at each moment during the activity period is integrated. That is, the start time of the activity period is used as the start time of integration, and the end time of the activity period is used as the end time of integration. The integration calculation is performed on the degree of acceleration exceeding the limit at each moment during the activity period, and the integrated result is the cumulative activity load of the target time period. .

[0036] For obtaining the cumulative heart rate response: This embodiment presets a resting heart rate, which represents the patient's reference heart rate in a resting state. The preset resting heart rate can be obtained by acquiring the average heart rate of the patient during multiple preset sleep periods in the early morning (e.g., from 1 a.m. to 3 a.m., used as the heart rate during deep sleep).

[0037] Determine the degree to which the heart rate exceeds the preset resting heart rate at each time point within a resting period. Let T2 be the number of time points in the resting period, and t2 be any time point within the resting period. The calculation method for the degree to which the heart rate at time t2 exceeds the preset resting heart rate is as follows: ,in, This represents the heart rate at time t2. This indicates the preset resting heart rate. Used in When the heart rate is less than 0, the heart rate excess level is set to 0.

[0038] Then, the degree of heart rate excess at each moment during the resting period is integrated. That is, the start time of the resting period is used as the start time of integration, and the end time of the resting period is used as the end time of integration. The result is used as the cumulative heart rate response for the target time period. Cumulative heart rate response It quantifies the total physiological cost to the cardiovascular system in repaying the oxygen debt generated during activity.

[0039] Meanwhile, this embodiment obtains the activity intensity characteristics of an activity period based on the cumulative activity load and the duration of the activity period (i.e., the number of moments contained within the activity period). The activity intensity characteristics characterize the average activity intensity within the activity period. In an exemplary embodiment, the activity intensity characteristic is the ratio of the cumulative activity load to the duration of the activity period.

[0040] Step S2: According to the preset percentile sequence, determine the load percentile time and response percentile time for each preset percentile completion based on the cumulative activity load and cumulative heart rate response.

[0041] Since the durations of activity and rest periods may differ, and the durations of different target time periods may also vary significantly, and since the intensity of movement changes randomly over time, directly comparing activity and rest periods in the time domain may not yield consistent dynamic characteristics. Therefore, this embodiment employs an energy cumulative percentile mapping method to transform the continuous signal based on absolute time into a discrete time sequence based on energy progress. It should be noted that if the cumulative activity load... If the data is below the preset minimum energy threshold, the data segment is deemed invalid and discarded without further percentile calculation. The preset minimum energy threshold can be set to 0. In other embodiments, the implementer can also set the threshold according to the actual situation. The current preset minimum energy threshold of 0 is obtained by long-term observation and analysis of the cumulative activity load data in the historical dataset.

[0042] First, this embodiment predetermines a percentile sequence, which includes several preset percentiles distributed from 0% to 100%. The step size of the preset percentiles is set according to the actual situation; the smaller the step size, the more preset percentiles are included. As an example, this embodiment uses a step size of 5%, that is, dividing the total energy equally with a step size of 5%. Thus, the preset percentile sequence includes 20 preset percentiles, that is, it includes 20 elements. Then, according to the preset percentile sequence, the load percentile time for each preset percentile completion is determined by the cumulative activity load, and the response percentile time for each preset percentile completion is determined by the cumulative heart rate response.

[0043] For ease of explanation, we define a candidate preset percentile as any preset percentile in the preset percentile sequence. Taking k (where k ranges from 1 to 20, and 20 represents the number of preset percentiles in the sequence) as an example, the candidate preset percentile is the k-th preset percentile in the preset percentile sequence. The following is a specific process for obtaining the load percentile time of the k-th preset percentile: Based on the cumulative activity load and the index (k) of the kth preset percentile, the target cumulative load threshold corresponding to the kth preset percentile is obtained. In an exemplary embodiment, the quantile of the kth preset percentile is obtained from the index (k / 20). Then, the target cumulative load threshold corresponding to the kth preset percentile is calculated using the following formula: ; in, This represents the target cumulative load threshold corresponding to the kth preset percentile.

[0044] Then, taking the start time of the activity period as the integration start time, and taking each moment of the activity period as the integration end time, the degree of acceleration exceeding the limit within the integration period is integrated, and the magnitude of the integration result obtained each time is compared with the target cumulative load threshold corresponding to the kth preset percentile, until the integration result is greater than or equal to the target cumulative load threshold corresponding to the kth preset percentile for the first time, and the integration ends. Thus, the integration end time of the activity period corresponding to the integration result at the end of the integration is obtained, and this integration end time is taken as the load percentile time of the kth preset percentile. Specifically, the process is as follows: Starting from the second moment of the activity period as the integration termination time, the first two moments of the activity period are used as the integration time interval. The degree of acceleration exceeding the threshold at the first two moments of the activity period is integrated. The integration result is then checked to see if it is greater than or equal to the target cumulative load threshold corresponding to the kth preset percentile. If it is, the integration ends, and the second moment of the activity period is used as the load percentile moment of the kth preset percentile. If it is less than, the first three moments of the activity period are used as the integration time interval. The degree of acceleration exceeding the threshold at the first three moments of the activity period is then integrated, and the integration result is checked to see if it is greater than or equal to the target cumulative load threshold corresponding to the kth preset percentile. If it is, the integration ends, and the third moment of the activity period is used as the load percentile moment of the kth preset percentile. If the result is less than the percentile time, the first four moments of the activity period are used as the integration time period. The acceleration exceeding the threshold at the first four moments of the activity period is integrated, and it is determined whether the integration result is greater than or equal to the target cumulative load threshold corresponding to the kth preset percentile. If it is greater than or equal to the target cumulative load threshold, the integration ends, and the fourth moment of the activity period is used as the load percentile time of the kth preset percentile. If the result is less than the threshold, the first five moments of the activity period are used as the integration time period, and the acceleration exceeding the threshold at the first five moments of the activity period is integrated. This process continues until the integration result is greater than or equal to the target cumulative load threshold corresponding to the kth preset percentile for the first time, at which point the integration ends, and the integration termination time of the activity period corresponding to the integration result at the end of the integration is used as the load percentile time of the kth preset percentile.

[0045] It should be understood that if, after traversing all moments in the activity period, the integral result is less than the target cumulative load threshold corresponding to the kth preset percentile, it means that there is no load percentile moment corresponding to the kth preset percentile. In this embodiment, no further processing will be performed, and the target time period will be re-determined.

[0046] Using the method described above for obtaining the load percentile time of the kth preset percentile, the load percentile time of each preset percentile in the preset percentile sequence is obtained. Then, the load percentile times of each preset percentile are sorted in ascending order of their index to obtain the load percentile time sequence. .

[0047] Similarly, the following is a specific process for obtaining the response percentile time of the k-th preset percentile: Based on the cumulative heart rate response and the index of the kth preset percentile, the target cumulative response threshold corresponding to the kth preset percentile is obtained. The target cumulative response threshold corresponding to the kth preset percentile is calculated using the following formula: ; in, This represents the target cumulative response threshold corresponding to the kth preset percentile.

[0048] Then, taking the start time of the resting period as the start time of integration, and taking each moment of the resting period as the end time of integration, the degree of heart rate excess within the integration period is integrated, and the magnitude of the integrated result obtained each time is compared with the target cumulative response threshold corresponding to the kth preset percentile, until the integrated result is greater than or equal to the target cumulative response threshold corresponding to the kth preset percentile for the first time, and the integration ends, thus obtaining the integration end time corresponding to the integration result at the end of the integration, and this integration end time is taken as the response percentile time corresponding to the kth preset percentile. Specifically, the integration process begins at the second moment of the resting period, which is taken as the end of the integration phase. The first two moments of the resting period are used as the integration time interval. The heart rate excess at these first two moments is integrated, and the result is checked against the target cumulative response threshold corresponding to the kth preset percentile. If the result is greater than or equal to the threshold, the integration ends, and the second moment of the resting period is taken as the response percentile moment of the kth preset percentile. If the result is less than the threshold, the first three moments of the resting period are used as the integration time interval. The heart rate excess at these first three moments is integrated, and the result is checked against the target cumulative response threshold corresponding to the kth preset percentile. If the result is greater than or equal to the threshold, the integration ends, and the third moment of the resting period is taken as the response percentile moment of the kth preset percentile. If the result is less than the percentile time, the first four moments of the resting period are used as the integration time period. The heart rate excess at the first four moments of the resting period is integrated, and it is determined whether the integration result is greater than or equal to the target cumulative response threshold corresponding to the kth preset percentile. If it is greater than or equal to the target cumulative response threshold corresponding to the kth preset percentile, the integration ends, and the fourth moment of the resting period is used as the response percentile time of the kth preset percentile. If the result is less than the percentile time, the first five moments of the resting period are used as the integration time period, and the heart rate excess at the first five moments of the resting period is integrated. This process continues until the integration result is greater than or equal to the target cumulative response threshold corresponding to the kth preset percentile for the first time, at which point the integration ends, and the integration termination time of the resting period corresponding to the integration result at the end of the integration is used as the response percentile time of the kth preset percentile.

[0049] It should be understood that if, after traversing all moments in the resting period, the integral result is less than the target cumulative response threshold corresponding to the kth preset percentile, it means that there is no response percentile moment at the kth preset percentile. In this embodiment, no further processing will be performed, and the target time period will be re-determined.

[0050] Using the method described above for obtaining the response percentile time of the kth preset percentile, the response percentile time of each preset percentile in the preset percentile sequence is obtained. Then, the response percentile times of each preset percentile are sorted in ascending order of their index to obtain the response percentile time sequence. .

[0051] After the above processing, regardless of whether the duration of the activity period and the resting period are the same, two standardized sequences with a fixed length of 20 are output: and These two sequences are strictly aligned in the energy dimension: The load percentile time at the kth preset percentile represents the cumulative active load that has reached the kth preset percentile (i.e., At that moment, and The response percentile time of the kth preset percentile represents the cumulative heart rate response that has compensated for the quantile of the kth preset percentile (i.e., The time at which the physiological debt of the k-th preset percentile is repaid is the time when the load percentile time sequence is obtained. This represents the relative time points when the cumulative activity load reaches 1 / 20, 2 / 20, ..., 20 / 20 of the total cumulative activity load; similarly, the response percentile time sequence. This represents the relative time points when the cumulative heart rate response reaches 1 / 20, 2 / 20, ..., 20 / 20 of the total cumulative heart rate response. This process transforms fragmented activities of varying duration and intensity into standardized time feature vectors that are strictly aligned in the load-response progression, providing a reliable input for subsequent kinetic analysis.

[0052] Step S3: Determine the degree of cardiopulmonary response delay by the lag time between the response percentile time and the corresponding load percentile time.

[0053] The pathological essence of early acute exacerbations of COPD is not simply a decrease in ventilation efficiency, but a combination of luminal mechanical instability caused by small airway inflammation and air trapping. Existing monitoring methods often confuse physiological tachypnea induced by exercise intensity with pathological expiratory dyspnea caused by airway obstruction. The purpose of this step is to use fluid dynamics principles to decompose airway function into macroscopic hysteresis characteristics (i.e., the degree of cardiopulmonary response delay) reflecting the degree of air trapping and microscopic obstruction characteristics reflecting the dynamic instability of airway closure. By decoupling and nonlinearly recombining these two dimensions, it is possible to specifically capture pathologically significant flow resistance abnormalities that persist even under low-load conditions during non-standardized daily activities, thereby achieving precise quantification of early airway function decline.

[0054] Airway narrowing and decreased elastic recoil can lead to delayed cardiopulmonary response due to air trapping at end-expiratory time, as well as airflow resistance oscillations caused by dynamic airway closure. Furthermore, the decreased lung elastic recoil in COPD patients results in insufficient expiratory driving pressure, leading to a prolonged retention time of metabolic products in the alveoli relative to the inspiratory work time. To quantify this macroscopic ventilatory efficiency lag, this step analyzes the percentile time series of ventilatory load. and response percentile time series The analysis was conducted to determine the lag time between each response percentile time and the corresponding load percentile time, thereby obtaining the degree of cardiopulmonary response delay and quantifying the macroscopic lag state of gas exchange.

[0055] Taking the kth preset percentile as an example, determine the load percentile time series. Load percentile time of the kth preset percentile Determine the percentile time sequence of the response. The response percentile time of the k-th preset percentile The calculation method for the cardiopulmonary response delay corresponding to the kth preset percentile is as follows: ; in, This indicates the degree of cardiopulmonary response delay corresponding to the k-th preset percentile. This represents the response percentile time of the k-th preset percentile. Lagging behind the load percentile time of the kth preset percentile The lag time, This represents the duration of the target time period. The lag time is normalized by calculating the ratio.

[0056] The cardiopulmonary response delay levels are sorted in ascending order of the percentile numbers in the preset percentile sequence to obtain a sequence of cardiopulmonary response delay levels. Therefore, the sequence of cardiopulmonary response delay corresponds to the preset percentile sequence.

[0057] The physical meaning of the calculation method for cardiopulmonary response delay is: the proportion of the cardiopulmonary system's lag time to the total time after completing the corresponding metabolic load. For healthy individuals, the cardiopulmonary response delay usually fluctuates around 0; however, for patients with gas trapping, the cardiopulmonary response delay value will increase significantly, and will show a non-linear upward trend as the k value increases (i.e., entering the end-expiratory phase or the late recovery phase).

[0058] Step S4: Based on the degree of cardiopulmonary response delay and activity intensity characteristics, obtain the comprehensive characteristics of airway obstruction.

[0059] This step involves a comprehensive analysis of the cardiopulmonary response delay and activity intensity characteristics to obtain the comprehensive airway obstruction characteristics for the target time period. In an exemplary embodiment, a specific process for obtaining these comprehensive airway obstruction characteristics is given below: Step S41: Based on the degree of cardiopulmonary response delay, obtain the degree of expiratory response fluctuation and the degree of gas trapping at the end of expiration.

[0060] In the pathological mechanism of COPD, small airway lesions cause intermittent dynamic closure and opening during end-expiration (i.e., the low lung volume phase), and this instability generates microscopic turbulent oscillations in the expiratory airflow. In contrast, the airflow in healthy individuals is laminar or smooth. However, high-intensity exercise can also induce physiological tremors of the respiratory muscles, introducing high-frequency noise into the signal. Therefore, this step extracts the degree of expiratory response fluctuation from the cardiopulmonary response delay sequence and obtains the degree of end-expiratory air trapping from the cardiopulmonary response delay sequence.

[0061] The degree of expiratory response fluctuation characterizes the fluctuation amplitude within the corresponding interval at the end of expiration. In an exemplary embodiment, the latter half of the cardiopulmonary response delay sequence is obtained. In this embodiment, since the cardiopulmonary response delay sequence includes 20 cardiopulmonary response delay levels, the latter half of the sequence is from the 11th to the 20th cardiopulmonary response delay level. It should be understood that if the number of preset percentiles in the preset percentile sequence is odd, then half of that odd number is calculated and rounded up. The cardiopulmonary response delay level corresponding to the rounded value is counted until the last cardiopulmonary response delay level, which is taken as the latter half of the sequence. It should be understood that the number of cardiopulmonary response delay levels included in the latter half of the sequence cannot be too few (e.g., it needs to be at least 3), which requires relevant limitations on the number of preset percentiles in the preset percentile sequence mentioned above.

[0062] Obtain the first-order difference absolute value sequence of the second half of the sequence. Each element in the first-order difference absolute value sequence is the absolute value of the difference between the degree of delay of two adjacent cardiopulmonary responses.

[0063] Obtain the local average value of each cardiopulmonary response delay level in the latter half of the sequence. The local window size is set to 3 when calculating the local average value. For ease of explanation, any cardiopulmonary response delay level in the latter half of the sequence is defined as the j-th cardiopulmonary response delay level, and the number of cardiopulmonary response delay levels in the latter half of the sequence is J.

[0064] The first-order absolute value of the cardiopulmonary response delay in the first-order absolute value sequence for: ,in, For the (j+1)th cardiopulmonary response delay, The degree of delay in the cardiopulmonary response is represented by the j-th term.

[0065] The local average of the degree of delay in the j-th cardiopulmonary response The calculation method is as follows: ,in, Let be the absolute value of the first-order difference of the (j+1)th cardiopulmonary response delay. This is the absolute value of the first-order difference of the (j-1)th cardiopulmonary response delay. When j=1, This is the absolute value of the difference between the first cardiopulmonary response delay level in the latter half of the sequence and its adjacent cardiopulmonary response delay level in the sequence. j takes a maximum value of J-1; when j = J-1 (i.e., the second-to-last cardiopulmonary response delay level), it is no longer calculated. That is, the local average value of the delay in the j-th cardiopulmonary response at this time. The calculation method is as follows: .

[0066] The difference between each first-order difference absolute value in the first-order difference absolute value sequence and its corresponding local mean is obtained. These differences are then fused to obtain the degree of expiratory response volatility. Therefore, the degree of expiratory response volatility characterizes the degree of deviation of the first-order difference absolute value from the local trend. The local mean is subtracted to eliminate the influence of overall sequence drift, retaining only the high-frequency oscillation components. The formula for calculating the degree of expiratory response volatility is as follows: ; in, This indicates the degree of fluctuation in the expiratory response.

[0067] The degree of gas trapping at the end of expiration is obtained from the latter half of the sequence and serves as a robust statistic for measuring the degree of gas trapping at the end of expiration. In this embodiment, the degree of gas trapping at the end of expiration is the median of the latter half of the sequence.

[0068] Step S42: Based on the degree of fluctuation in expiratory response and the characteristics of activity intensity, obtain the fluctuation characteristics of expiratory resistance.

[0069] The stronger the expiratory response fluctuation, the stronger the expiratory resistance fluctuation, and the two are positively correlated. To eliminate interference from exercise intensity, this embodiment uses activity intensity characteristics to inversely correct the expiratory response fluctuation (i.e., expiratory resistance fluctuation characteristics are inversely correlated with activity intensity characteristics) to suppress physiological noise under high load, thereby suppressing the interference of physiological tremor, and amplifying pathological fluctuations under low load, thereby amplifying pathological low-load block, thus obtaining the expiratory resistance fluctuation characteristics. This embodiment provides the following calculation method for the expiratory resistance fluctuation characteristics: ; in, This indicates the characteristics of expiratory resistance fluctuations. Let represent the logarithmic function with base e, where e represents the natural constant. Indicates the intensity characteristics of the activity. This represents a preset constant whose dimensions are opposite to those of the activity intensity characteristic. It is used to eliminate dimensions after multiplying with the activity intensity characteristic. In this embodiment, the preset constant... The value is 1.0.

[0070] In this formula, the denominator utilizes the compression property of the logarithmic function: when When the exercise is relatively vigorous (i.e., strenuous), the denominator increases significantly, thus affecting the calculus caused by the exercise. Perform numerical suppression; when At lower (i.e., mild) activity levels, the denominator changes gradually and is small in value, thus preserving or even highlighting the subtle fluctuations caused by airway lesions. The introduction of the natural constant e ensures that the denominator is always greater than 1, avoiding the risk of divergence in division values.

[0071] Step S43: Integrate the degree of gas trapping at the end of expiration and the fluctuation characteristics of expiratory resistance to obtain comprehensive airway obstruction characteristics.

[0072] A single hysteresis feature may be affected by ambient temperature or cardiovascular autonomic nervous system regulation, while a single fluctuation feature may be interfered with by physical noise such as sensor loosening. Early airway obstruction features in acute exacerbations are multidimensional, simultaneously manifesting as "slowed airflow" and "unstable airflow." To improve monitoring specificity, this embodiment nonlinearly couples the degree of end-expiratory gas trapping with expiratory resistance fluctuation features to generate a comprehensive airway obstruction feature. A high-risk value is output when both macroscopic and microscopic features are abnormal, characterizing the current degree of airway ventilation impairment.

[0073] In one exemplary embodiment, the degree of gas trapping at end-expiratory time and the expiratory resistance fluctuation characteristics are multiplicatively coupled to obtain a comprehensive airway obstruction characteristic: ; in, This indicates a comprehensive characteristic of airway obstruction. This indicates the degree of gas trapping at the end of expiration. In this formula, the term... Acting as a nonlinear gain coefficient, if the patient has only hysteresis but the airflow is stable (i.e. If the value is very small (e.g., approximately 0), then the airway obstruction syndrome degenerates into a simple hysteresis measurement; if the patient has significant airway resistance fluctuations along with the hysteresis (i.e., ... (Greater than 0), the comprehensive airway obstruction characteristics will significantly amplify the final value. This coupling mechanism ensures that the system can keenly capture the hydrodynamic signs specific to obstructive ventilation dysfunction.

[0074] Step S5: Based on the comprehensive characteristics of airway obstruction, provide early warning of acute exacerbations of COPD in patients.

[0075] The above steps have successfully eliminated the interference of exercise intensity, and the output airway obstruction comprehensive characteristics have extremely high signal-to-noise ratio and pathological specificity. The higher the value of the airway obstruction comprehensive characteristics, the greater the current degree of airway ventilation impairment, the more likely the patient is to experience an acute exacerbation of COPD, and the greater the need for early warning. Therefore, early warning of acute exacerbations of COPD can be provided for patients based on airway obstruction comprehensive characteristics.

[0076] In one exemplary embodiment, a warning threshold can be set as the standard for whether to issue an early warning. This threshold can be obtained by the doctor based on experience, or it can be obtained in the following way: Acquire comprehensive airway obstruction characteristics from a large number of historical patients, and then, based on the actual physical condition of each historical patient, determine the historical patients who require early warning for COPD exacerbations. Use the minimum value among the comprehensive airway obstruction characteristics of these historical patients as the warning threshold. Compare the comprehensive airway obstruction characteristics with the warning threshold. If the characteristic is greater than or equal to the warning threshold, an early warning signal for COPD exacerbation is output; if it is less than the warning threshold, no early warning signal for COPD exacerbation is output.

[0077] Considering that COPD patients may experience occasional airway constriction due to emotional excitement or short-term environmental stimuli (such as inhaling cold air) in their daily lives, a single suspected abnormality (i.e., a single output of an early warning signal for COPD acute exacerbation) may not be absolutely equivalent to an acute exacerbation. An acute exacerbation is a process of continuous deterioration of airway inflammation. Therefore, this embodiment can also introduce a time window confirmation mechanism to filter out occasional noise and lock in a deterministic pathological trend. Specifically: this embodiment sets a rolling time window (e.g., the past 24 hours), and determines the proportion of airway obstruction characteristics greater than or equal to the warning threshold within this rolling time window. If the proportion is greater than a first preset proportion threshold (e.g., 50%), it is determined that the patient's airway function has undergone irreversible deterioration, rather than an occasional fluctuation, and a level one warning signal is output, indicating an acute exacerbation; if the proportion is greater than or equal to a second preset proportion threshold (e.g., 30%) and less than or equal to the first preset proportion threshold, a level two warning signal is output, with the level one warning signal being more severe than the level two warning signal; if the proportion is less than the second preset proportion threshold, a temporary normal signal is output.

[0078] It should be understood that, as described above, the number of times the comprehensive airway obstruction characteristics are acquired within the rolling time window cannot be too small (e.g., at least 5 times), otherwise it may lead to misjudgment. Therefore, if the number of times the comprehensive airway obstruction characteristics are acquired within the rolling time window is too small (i.e., less than 5 times), the graded warning mechanism in the previous paragraph will no longer be executed.

[0079] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0080] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. An early intelligent warning system for acute exacerbations of COPD based on wearable devices, characterized in that, include: The data acquisition module is used to acquire the patient's activity load data and heart rate data within a target time period based on a wearable device. The target time period includes the activity period and the adjacent resting period thereafter. The early warning module, electrically connected to the data acquisition module, is configured as follows: The cumulative activity load and cumulative heart rate response for the target time period are obtained from the activity load data and heart rate data, respectively. According to the preset percentile sequence, the load percentile time and response percentile time for each completion of the preset percentile are determined by the cumulative activity load and the cumulative heart rate response, respectively. The degree of cardiopulmonary response delay is determined by the lag time between the response percentile time and the corresponding load percentile time. Based on the cardiopulmonary response delay and activity intensity characteristics, a comprehensive airway obstruction characteristic is obtained; the activity intensity characteristic is obtained from the cumulative activity load and the duration of the activity period. Based on the comprehensive characteristics of airway obstruction, early warning of acute exacerbations of COPD can be provided for patients.

2. The intelligent early warning system for acute exacerbations of COPD based on wearable devices as described in claim 1, characterized in that, The process of obtaining the target time period includes: Determine whether the patient's activity intensity exceeds a preset intensity threshold within each time period; The target time period is obtained when the activity intensity of at least a first preset number of unit time periods is greater than a preset intensity threshold, and the activity intensity of at least a second preset number of unit time periods is less than or equal to the preset intensity threshold; wherein, the last unit time period in the at least a first preset number of unit time periods is the time period preceding the first unit time period in the at least a second preset number of unit time periods, the start time of the first unit time period in the at least a first preset number of unit time periods is taken as the start time of the activity period, and the end time of the last unit time period in the at least a first preset number of unit time periods is taken as the end time of the activity period; the start time of the first unit time period in the at least a second preset number of unit time periods is taken as the start time of the rest period, and the end time of the last unit time period in the at least a second preset number of unit time periods is taken as the end time of the rest period.

3. The early intelligent warning system for acute exacerbations of COPD based on wearable devices as described in claim 1, characterized in that, The active load data is the signal vector magnitude of the triaxial acceleration components; The process of obtaining the cumulative active load includes: Determine the degree to which the signal vector magnitude exceeds the gravitational acceleration at each moment during the activity period; The cumulative activity load is obtained by integrating the degree of acceleration exceeding the limit at each moment during the activity period. The process of obtaining the cumulative heart rate response includes: Determine the extent to which the heart rate exceeds the preset resting heart rate at each moment during the resting period; The cumulative heart rate response is obtained by integrating the degree of heart rate exceedance at each moment during the resting period.

4. The early intelligent warning system for acute exacerbations of COPD based on wearable devices as described in claim 3, characterized in that, The process of obtaining the load percentile time includes: Based on the cumulative active load and the candidate preset percentile's index in the preset percentile sequence, a target cumulative load threshold corresponding to the candidate preset percentile is obtained; the candidate preset percentile is any preset percentile in the preset percentile sequence. The integration begins at the start of the activity period and ends at each moment of the activity period. The degree of acceleration exceeding the threshold is integrated, and the integration result is compared with the target cumulative load threshold. Integration ends when the integration result is greater than or equal to the target cumulative load threshold for the first time. The integration end time corresponding to the integration result at the end of the integration is used as the load percentile time corresponding to the candidate preset percentile.

5. The early intelligent warning system for acute exacerbations of COPD based on wearable devices as described in claim 4, characterized in that, The process of obtaining the percentile time of the response includes: Based on the cumulative heart rate response and the candidate preset percentile's index in the preset percentile sequence, the target cumulative response threshold corresponding to the candidate preset percentile is obtained. The heart rate excess is integrated using the start time of the resting period as the start time of integration, and each moment of the resting period is used as the end time of integration. The integration result is compared with the target cumulative response threshold each time, and the integration ends when the integration result is greater than or equal to the target cumulative response threshold for the first time. The integration end time corresponding to the integration result at the end of the integration is used as the response percentile time corresponding to the candidate preset percentile.

6. The early intelligent warning system for acute exacerbations of COPD based on wearable devices as described in claim 5, characterized in that, The process of obtaining the target cumulative load threshold includes: The quantile of the candidate preset percentile is obtained from the index of the candidate preset percentile in the preset percentile sequence; Calculate the product of the quantile and the cumulative active load to obtain the target cumulative load threshold corresponding to the candidate preset percentile; The process of obtaining the target cumulative response threshold includes: Calculate the product of the percentile and the cumulative heart rate response to obtain the target cumulative response threshold corresponding to the candidate preset percentile.

7. The early intelligent warning system for acute exacerbations of COPD based on wearable devices as described in claim 1, characterized in that, The process of obtaining the comprehensive characteristics of airway obstruction includes: The degree of cardiopulmonary response delay is used to determine the degree of expiratory response fluctuation and the degree of gas trapping at the end of expiration. Based on the degree of fluctuation in the expiratory response and the activity intensity characteristics, the expiratory resistance fluctuation characteristics are obtained; the expiratory resistance fluctuation characteristics are positively correlated with the degree of fluctuation in the expiratory response and negatively correlated with the activity intensity characteristics. By integrating the degree of gas trapping at the end of expiration and the fluctuation characteristics of expiratory resistance, the comprehensive characteristics of airway obstruction are obtained.

8. The early intelligent warning system for acute exacerbations of COPD based on wearable devices as described in claim 7, characterized in that, The process of obtaining the degree of fluctuation in the expiratory response includes: Obtain the second half of the sequence in the cardiopulmonary response delay sequence corresponding to the preset percentile sequence; Determine the first-order difference absolute value sequence of the latter half of the sequence; The degree of fluctuation in the expiratory response is obtained by integrating the differences between each first-order difference absolute value in the first-order difference absolute value sequence and its corresponding local average value.

9. The intelligent early warning system for acute exacerbations of COPD based on wearable devices as described in claim 8, characterized in that, The degree of gas trapping at the end of expiration is the median of the latter half of the sequence.

10. The intelligent early warning system for acute exacerbations of COPD based on wearable devices as described in claim 1, characterized in that, The activity intensity characteristic is the ratio of the cumulative activity load to the duration of the activity period.

Images