Intelligent early warning system for dynamic deterioration of chronic obstructive pulmonary disease

Abstract

The invention relates to the technical field of medical care, in particular to an intelligent early warning system for dynamic deterioration of chronic obstructive pulmonary disease, which is used for solving the technical problem of inaccurate evaluation of lung ventilation capability of a user in the prior art. The system comprises a data acquisition unit used for acquiring lung function time sequence data and multi-dimensional physiological time sequence data of a target user in an observation stage; the observation phase comprises a plurality of consecutive dates; the processing unit is used for determining suspected abnormal days according to the lung function time sequence data, determining the deviation degree of the multi-dimensional physiological time sequence data of the suspected abnormal days and the multi-dimensional physiological time sequence data of the reference day and the similarity of the deviation degree of the physiological time sequence data of different dimensions, determining the abnormal days from the suspected abnormal days based on the deviation degree and the similarity, and sending the abnormal days to the user. The chronic obstructive pulmonary disease dynamic deterioration risk level of the target user is determined based on the abnormal days; and performing risk early warning based on the chronic obstructive pulmonary disease dynamic deterioration risk level of the target user.

Description

Technical Field

[0001] This invention relates to the field of healthcare technology, specifically to an intelligent early warning system for the dynamic deterioration of COPD. Background Technology

[0002] Current methods for detecting dynamic deterioration of COPD involve assessing the lung ventilation capacity of COPD patients based on the ratio of forced expiratory volume in one second (FEV1) to forced vital capacity (FVC) (FEV1 / FVC ratio). Physicians then use this ventilation capacity as a basis, combined with the patient's current physical condition and their experience, to provide early warnings of dynamic deterioration in COPD. However, current methods often fail to accurately reflect a patient's ventilation capacity due to various factors such as colds and physical exhaustion. The single FEV1 / FVC ratio cannot accurately reflect the patient's lung ventilation capacity, leading to inaccurate assessments. Summary of the Invention

[0003] In order to solve the problem of inaccurate assessment of user lung ventilation capacity in the prior art, the purpose of this invention is to provide an intelligent early warning system for dynamic deterioration of COPD, which includes a data acquisition unit and a processing unit. The data acquisition unit is used to collect time-series data on lung function and multi-dimensional physiological time-series data of the target user during the observation phase; the observation phase includes multiple consecutive dates; The processing unit is used to determine suspected abnormal days based on lung function time-series data. Suspected abnormal days are dates where the change in lung function data exceeds a first threshold. It determines the deviation of multi-dimensional physiological time-series data of suspected abnormal days from multi-dimensional physiological time-series data of reference days, as well as the similarity of deviations in different dimensions of physiological time-series data. The reference day is the date with the smallest abnormal value in lung function data and the best lung ventilation function. Based on the deviation and similarity, it identifies abnormal days from the suspected abnormal days and determines the dynamic deterioration risk level of the target user's COPD based on the abnormal days. It then provides a risk warning based on the target user's dynamic deterioration risk level of COPD.

[0004] As one possible implementation, the time-series lung function data includes forced expiratory volume in one second (FEV1) and forced vital capacity (FVC) for each day over multiple days; the processing unit is specifically used to: calculate the first ratio of FEV1 to FVC for each day; use the normalized value of the difference between the first ratio for each day and the first ratio for the previous day as the initial outlier of the daily lung function data; and record the days with initial outliers greater than a first threshold as suspected outlier days.

[0005] As one possible implementation, the processing unit is also configured to: record the date on which the initial abnormal value is less than or equal to the first threshold as the initial normal day; take consecutive adjacent initial normal days as the disease stabilization phase, select the disease stabilization phase with the most initial normal days from multiple disease stabilization phases as the reference phase; calculate the ventilation function of the target user per day in the reference phase, wherein the ventilation function is positively correlated with the first ratio and negatively correlated with the initial abnormal value; and take the initial normal day with the largest ventilation function value in the reference phase as the reference day.

[0006] As one possible implementation, the processing unit is specifically used for: fitting the target dimension physiological time-series data of the suspected abnormal day to obtain a first fitting curve, and fitting the target dimension physiological time-series data of the reference day to obtain a second fitting curve; the suspected abnormal day is one of multiple suspected abnormal days, and the target dimension physiological time-series data is physiological time-series data of one dimension of the multi-dimensional physiological time-series data; calculating the absolute value of the data difference between the first fitting curve and the second fitting curve at each time point, and the trend weight at each time point, wherein if the slope direction of the first fitting curve and the second fitting curve is the same at a time point, the trend weight is a first value, otherwise the trend weight is a second value, and the second value is greater than the first value; based on the trend weight at each time point, the absolute values ​​of the differences at that time point are weighted and summed, and the weighted sum is normalized to obtain the deviation of the target dimension of the suspected abnormal day; the average deviation of the suspected abnormal day in each dimension is taken as the deviation of the suspected abnormal day.

[0007] As one possible implementation, the processing unit is specifically used to: determine the deviation time series sequence of the first-dimensional physiological time series data and the deviation time series sequence of the second-dimensional physiological time series data; the first-dimensional physiological time series data and the second-dimensional physiological time series data are respectively physiological time series data of one dimension of multi-dimensional physiological time series data, and the first-dimensional physiological time series data and the second-dimensional physiological time series data are different; the deviation time series sequence includes the deviation at multiple different times; and the similarity between the deviation time series sequence of the first-dimensional physiological time series data and the deviation time series sequence of the second-dimensional physiological time series data is used as the similarity between the first-dimensional physiological time series data and the second-dimensional physiological time series data.

[0008] As one possible implementation, the processing unit is specifically used to: calculate the updated outlier value of the target suspected outlier day based on the deviation of the target suspected outlier day, the mean similarity of all dimensions of physiological time series data of the target suspected outlier day, and the number of days between the target suspected outlier day and the reference day. The updated outlier value is positively correlated with the deviation and the mean similarity, and negatively correlated with the number of days between the two days. If the updated outlier value of the target suspected outlier day is greater than the second threshold, then the target suspected outlier day is regarded as an outlier day.

[0009] As one possible implementation, the processing unit is specifically used to: determine at least one stage of disease deterioration for the target user based on abnormal days in the observation phase, wherein the stage of disease deterioration includes multiple consecutive abnormal days; and determine the risk level of the target user based on the severity of the deterioration stage, the ability to manage the disease, and the time interval between adjacent stages of disease deterioration, wherein the severity of the deterioration is used to characterize the severity and duration of the deterioration stage, and the ability to manage the disease is used to characterize the target user's ability to control the disease and maintain its stability.

[0010] As one possible implementation, the processing unit is specifically used for: determining the updated outlier value for each abnormal day in any stage of disease deterioration; fitting the updated outlier value of any stage of disease deterioration to obtain a third fitted curve; determining the slope normalization value for each abnormal day in any stage of disease deterioration based on the third fitted curve; multiplying the updated outlier value of any abnormal day by the slope normalization value to obtain a first product; accumulating the first products of each abnormal day to obtain a first accumulated value; calculating the time proportion of the disease deterioration stage in the observation stage; and using the product of the first accumulated value and the time proportion as the deterioration intensity of any stage of disease deterioration.

[0011] As one possible implementation, the processing unit is specifically used to: determine the severity of any deterioration stage, the normalized slope of the deterioration trend, and the duration of the stable stage following any deterioration stage, wherein the normalized slope of the deterioration trend is the normalized slope of the fitted line obtained by fitting the updated outliers of the deterioration stage; the stable stage includes multiple consecutive normal days; normal days are dates in the observation period excluding abnormal days; multiplying the severity of any deterioration stage by the corresponding normalized slope to obtain a second product; multiplying the second product by the reciprocal of the duration of the stable stage to obtain a third product; and using the negative of the third product as an exponent to calculate the power of the natural constant e to obtain the disease management capability.

[0012] As one possible implementation, the processing unit is specifically used to: fit the time series of time intervals between adjacent disease deterioration stages, and use the normalized value of the slope of the fitted straight line as the deterioration interval trend; use the product of the reciprocal of the disease management ability and the reciprocal of the deterioration interval trend as the basic risk value; use the reciprocal of the time interval between the most recent disease deterioration stage and the current time as the time-sensitive factor; multiply the basic risk value and the time-sensitive factor, and normalize the product to determine the risk level of the target user.

[0013] This invention offers the following advantages: By using the deviation of multi-dimensional physiological time-series data from a reference day and the similarity of deviations between different dimensions as a verification mechanism, this application verifies abnormalities in the user's lung function data, identifies abnormal days, and then determines the user's risk level of dynamic deterioration of COPD based on these abnormal days. Therefore, this application can effectively identify false abnormalities in lung function data caused by non-disease-related factors such as colds and physical exhaustion, thereby significantly improving the specificity and accuracy of early warning. Attached Figure Description

[0014] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a system architecture diagram of an intelligent early warning system for dynamic deterioration of COPD, provided in one embodiment of the present invention. Figure 2 This is a flowchart illustrating an intelligent early warning method for dynamic deterioration of COPD, 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 following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an intelligent early warning system for dynamic deterioration of COPD based on the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[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.

[0018] The following description, in conjunction with the accompanying drawings, details the specific scheme of the intelligent early warning system for the dynamic deterioration of COPD provided by the present invention.

[0019] Please see Figure 1 This diagram illustrates the system architecture of an intelligent early warning system for the dynamic deterioration of COPD, provided in one embodiment of the present invention. Figure 1 As shown, the intelligent early warning system for dynamic deterioration of COPD includes a data acquisition unit 101 and a processing unit 102.

[0020] The data acquisition unit is used to collect time-series data on lung function and multi-dimensional physiological time-series data of the target user during the observation phase; the observation phase includes multiple consecutive dates; The processing unit is used to determine suspected abnormal days based on lung function time-series data. Suspected abnormal days are dates where the change in lung function data exceeds a first threshold. It determines the deviation of multi-dimensional physiological time-series data of suspected abnormal days from multi-dimensional physiological time-series data of reference days, as well as the similarity of deviations in different dimensions of physiological time-series data. The reference day is the date with the smallest abnormal value in lung function data and the best lung ventilation function. Based on the deviation and similarity, it identifies abnormal days from the suspected abnormal days and determines the dynamic deterioration risk level of the target user's COPD based on the abnormal days. It then provides a risk warning based on the target user's dynamic deterioration risk level of COPD.

[0021] In one possible implementation, since the lung function data of COPD users are easily affected by short-term physiological factors, in order to improve the accuracy of early warning of dynamic deterioration of COPD users' condition, this application can obtain the user's long-term lung function data sequence to reduce the interference of non-disease factors in the short term.

[0022] Optionally, this application uses a pulmonary function testing device to collect time-series pulmonary function data of the target user during the observation phase. For example, this application can start from the user's diagnosis of chronic obstructive pulmonary disease (COPD) and use a pulmonary function testing device to collect the user's FEV1 / FVC ratio once a day to obtain the current user's FEV1 / FVC ratio sequence. Here, FEV1 is the forced expiratory volume in one second, and FVC is the forced vital capacity. The FEV1 / FVC ratio is a recognized core indicator for the diagnosis and severity grading of COPD, and is a sensitive indicator for assessing the presence of airflow obstruction. The normal value of the FEV1 / FVC ratio is usually above 80%. The higher the FEV1 / FVC ratio, the milder the degree of expiratory airflow restriction, indicating better airway patency. When the FEV1 / FVC ratio drops to <70%, it suggests possible persistent airflow restriction. This application can use smart wearable devices, such as smartwatches and smart bracelets, to collect the user's multi-dimensional physiological time-series data. For example, multidimensional physiological time-series data includes data from different dimensions such as blood oxygen saturation, heart rate, and respiratory rate.

[0023] As one implementation method, this application can use fractional scaling standardization to process the collected lung function time-series data and multi-dimensional physiological time-series data. The processed data have similar scales and distributions, which makes it easier to perform subsequent data processing.

[0024] It's important to note that COPD is a chronic disease. As treatment progresses, a user's lung function doesn't gradually improve; instead, it may remain stable or slowly decline. One of the core risks of COPD is acute exacerbations. Acute exacerbations not only lead to a rapid decline in lung function but also significantly increase hospitalization and mortality rates. To avoid this, timely intervention is crucial when early signs of deterioration appear to prevent further progression to acute exacerbations. Early deterioration of lung function in COPD patients involves small, daily abnormal fluctuations on a stable or slowly declining foundation. For example, worsening airway inflammation or sputum blockage might cause the FEV1 / FVC ratio to drop from 62% one day to 59% the next, a daily decrease of 3%. Early identification of such changes allows for timely detection of deterioration signals, preventing further worsening of the condition.

[0025] In this embodiment, the processing unit can perform preliminary screening of the collected lung function time-series data to identify suspected abnormal days. Then, it compares the multi-dimensional physiological time-series data of the suspected abnormal days and normal days to further filter out the actual abnormal days. Furthermore, based on the distribution, trends, and lung function data of the actual abnormal days, this application can further determine the user's COPD risk level, assess whether the user is experiencing early deterioration symptoms, and provide intelligent early warning of dynamic deterioration of COPD based on the risk level.

[0026] The following is a detailed explanation of the process by which the processing unit identifies suspected abnormal days: As one possible implementation, the time-series lung function data includes forced expiratory volume in one second (FEV1) and forced vital capacity (FVC) for each day over multiple days; the processing unit is specifically used to: calculate the first ratio of FEV1 to FVC for each day; use the normalized value of the difference between the first ratio of the previous day and the first ratio of the current day as the initial outlier of the daily lung function data; and record the days with initial outliers greater than a first threshold as suspected outlier days.

[0027] As an example, the first threshold is set to 0.5. Specifically, taking the FEV1 / FVC ratio collected on day d as an example, the normalized value of the difference between the FEV1 / FVC ratio on day d-1 and the FEV1 / FVC ratio on day d is calculated and recorded as the outlier of the user's lung function data on day d. .when When the value is greater than or equal to 0.5, day d is recorded as a suspected abnormal day; otherwise, it is recorded as an initial normal day. The specific value of the first threshold can be set by those skilled in the art based on clinical practice or experience, and this application will not elaborate on this.

[0028] Based on this, this application quantifies initial outliers by calculating the normalized difference between the daily FEV1 / FVC ratio and the previous day's ratio, and uses this to screen suspected abnormal days, providing an objective and quantitative preliminary anomaly identification method. This method can keenly capture subtle but continuous deterioration trends in lung function within a short period, providing reliable preliminary screening results for subsequent in-depth analysis.

[0029] The following is a detailed explanation of the process by which the processing unit determines the reference day: As one possible implementation, the processing unit is also configured to: record the date on which the initial abnormal value is less than or equal to the first threshold as the initial normal day; take consecutive adjacent initial normal days as the disease stabilization phase, select the disease stabilization phase with the most initial normal days from multiple disease stabilization phases as the reference phase; calculate the ventilation function of the target user per day in the reference phase, wherein the ventilation function is positively correlated with the first ratio and negatively correlated with the initial abnormal value; and take the initial normal day with the largest ventilation function value in the reference phase as the reference day.

[0030] Optionally, this process can be implemented as follows: Among all initial normal days, consecutive adjacent initial normal days are considered as a disease stabilization phase. The disease stabilization phase with the longest duration is designated as the reference phase. In the reference phase, the product of the target value and the corresponding FEV1 / FVC ratio is calculated. Among them, outliers in users' lung function data on normal days When the reciprocal of the value is not zero, the target value is the outlier of the lung function data of the normal user. The reciprocal of; outliers in normal user lung function data. When the reciprocal of is 0, the target value is the abnormal value of the lung function data of normal users. The sum of the reciprocal of the product and a minimum value (e.g., 0.01). For all normal days... The normal days corresponding to the maximum value in the range are denoted as reference days.

[0031] It should be noted that during the reference phase, the user's respiratory rate, heart rate, and other multi-dimensional data are all in a stable state, so the reference phase can better reflect the user's true physiological baseline. As an indicator of a user's lung function health, smaller abnormal values ​​in lung function data and better ventilation indicate a more stable condition. Therefore, the selected reference days are more reliable. The greater the deviation between the multi-dimensional data of suspected abnormal days and the multi-dimensional data of the reference days, the more unstable the condition, and the greater the probability that these suspected abnormal days are actually abnormal days.

[0032] Based on this, this application establishes a highly reliable individualized physiological baseline by comprehensively considering the stability of pulmonary function data and the optimal state of ventilation function during the longest stable phase of the disease (reference phase) and selecting the most representative reference day. This baseline truly reflects the user's optimal or typical physiological level in a stable state, providing a reliable benchmark for subsequent accurate assessment of the deviation of suspected abnormal days, thereby greatly improving the individualized accuracy of abnormal judgment.

[0033] The following details the process by which the processing unit determines the deviation of suspected abnormal days and the similarity of deviations in physiological time-series data across different dimensions: As one possible implementation, the processing unit is specifically used for: fitting the target dimension physiological time-series data of the suspected abnormal day to obtain a first fitting curve, and fitting the target dimension physiological time-series data of the reference day to obtain a second fitting curve; the suspected abnormal day is one of multiple suspected abnormal days, and the target dimension physiological time-series data is physiological time-series data of one dimension of the multi-dimensional physiological time-series data; calculating the absolute value of the data difference between the first fitting curve and the second fitting curve at each time point, and the trend weight at each time point, wherein if the slope direction of the first fitting curve and the second fitting curve is the same at a time point, the trend weight is a first value, otherwise the trend weight is a second value, and the second value is greater than the first value; based on the trend weight at each time point, the absolute values ​​of the differences at that time point are weighted and summed, and the weighted sum is normalized to obtain the deviation of the target dimension of the suspected abnormal day; the average deviation of the suspected abnormal day in each dimension is taken as the deviation of the suspected abnormal day.

[0034] As one possible implementation, the processing unit is specifically used to: determine the deviation time series sequence of the first-dimensional physiological time series data and the deviation time series sequence of the second-dimensional physiological time series data; the first-dimensional physiological time series data and the second-dimensional physiological time series data are respectively physiological time series data of one dimension of multi-dimensional physiological time series data, and the first-dimensional physiological time series data and the second-dimensional physiological time series data are different; the deviation time series sequence includes the deviation at multiple different times; and the similarity between the deviation time series sequence of the first-dimensional physiological time series data and the deviation time series sequence of the second-dimensional physiological time series data is used as the similarity between the first-dimensional physiological time series data and the second-dimensional physiological time series data.

[0035] For example, the first value is 0.1, and the second value is 1. Optionally, the process of determining the deviation and similarity of suspected abnormal days can be specifically implemented as follows: taking the w-th dimension data as an example for analysis, obtain the time series sequence of the w-th dimension data of the y-th suspected abnormal day { } and the time series sequence of the w-th dimension of the reference day { The least squares method is used to obtain the fitted curves of the two sequences.

[0036] by{ }as well as{ The absolute value of the data difference at time i in} For example, when the slopes of the tangent lines for the w-th dimension at the i-th time point on both days are either both positive or both negative, then... Weighting coefficients Record it as 0.1, and conversely, record it as 1.

[0037] Calculate the degree of deviation of the data in the w-th dimension between the y-th suspected abnormal day and the reference day at time i. ,in, Indicates the i-th time point The weight value, This indicates the degree of deviation in the trend of the w-th dimension data at the same time over two days. The larger the value, the more completely opposite the direction of change in the same dimension of data within two days, and the greater the degree of deviation.

[0038] The sequence of deviations of the w-th dimension data between the y-th suspected abnormal day and the reference day at all times is calculated. The average value of all data in this sequence is denoted as the overall deviation of the data in the w-th dimension between the y-th suspected abnormal day and the reference day. Similarly, the deviation sequence of other dimensions of data can be obtained.

[0039] Obtain the similarity of the deviation sequences of any two dimensions of data in the y-th suspected abnormal day. The similarity is denoted as the degree of deviation between the two dimensions of data. As an example, this application can obtain the Pearson correlation coefficient of the deviation sequences of any two dimensions of data, then normalize the Pearson correlation coefficient, and use the normalized result as its similarity. Since the Pearson correlation coefficient ranges from -1 to 1, the normalization method for the Pearson correlation coefficient can be: (Pearson correlation coefficient + 1) / 2.

[0040] It's important to note that the core pathological characteristic of COPD is "airflow limitation and decreased lung ventilation." When COPD users exhibit genuinely abnormal lung function data, the body activates compensatory mechanisms due to worsening airflow limitation. These mechanisms include increasing respiratory rate to compensate for insufficient ventilation and decreasing blood oxygen saturation to reduce oxygen exchange efficiency. Therefore, truly abnormal lung function data in COPD users is often accompanied by synergistic changes across various data dimensions. The more similar the deviations between any two data dimensions, the more synchronized their deviation trends and the stronger the synergy between the dimensions. For example, if an increase in respiratory rate is accompanied by a simultaneous decrease in blood oxygen saturation, it's easier to rule out non-disease-related factors. This suspected abnormal day is more likely an early sign of COPD deterioration and requires close monitoring.

[0041] Based on the above technical solution, this application, when selecting abnormal days from suspected abnormal days, not only considers the specific numerical differences between suspected abnormal days and reference days, but also introduces trend weights to quantify the consistency of the two in the data change trend (slope direction). Higher weights are assigned when trends are opposite, which can effectively capture the compensatory physiological responses of the body in a true pathological state (such as increased breathing to compensate for insufficient ventilation). Combined with the assessment of the similarity of data deviation sequences in different dimensions, the system can identify typical deterioration patterns of multi-system coordinated abnormalities, greatly enhancing its ability to distinguish between true and false abnormalities.

[0042] The following is a detailed explanation of the process by which the processing unit identifies abnormal days from suspected abnormal days: As one possible implementation, the processing unit is specifically used to: calculate the updated outlier value of the target suspected outlier day based on the deviation of the target suspected outlier day, the mean similarity of all dimensions of physiological time series data of the target suspected outlier day, and the number of days between the target suspected outlier day and the reference day. The updated outlier value is positively correlated with the deviation and the mean similarity, and negatively correlated with the number of days between the two days. If the updated outlier value of the target suspected outlier day is greater than the second threshold, then the target suspected outlier day is regarded as an outlier day.

[0043] As an example, the second threshold is set to 0.7. Optionally, the process can be implemented by: calculating the updated outlier value of the lung function data on the y-th suspected abnormal day. .Will Days with a value greater than or equal to 0.7 that are suspected to be abnormal are recorded as abnormal days, while those below this value are recorded as normal days. The specific value of the second threshold can be set by those skilled in the art based on clinical practice or experience, and this application will not elaborate on this.

[0044] As an example, updating outliers Satisfy the following formula: in, This represents all data in any two dimensions during the y-th suspected abnormal day. The mean, The larger the value, the better the synergy of the multi-dimensional data in the y-th suspected abnormal day. This represents the data of all dimensions in the y-th suspected abnormal day and the reference day. The mean, This represents the number of days between the y-th suspected abnormal day and the reference day. The smaller the interval, the more consistent the user's physical condition. The more credible it is. This represents the degree of deviation between the y-th suspected abnormal day and all dimensions of data in the reference day. The better the coordination of multi-dimensional data in the suspected abnormal day, and the greater the degree of deviation from all dimensions of data in the reference day, the more abnormal the user's multi-dimensional data appears on that day. In this case, the abnormality of the user's lung function data may be a signal of worsening condition.

[0045] Based on the above technical solution, this application calculates an updated and more reliable outlier value by considering three key factors: comprehensive deviation, multi-dimensional data synergy (mean similarity), and temporal proximity (interval in days). This allows for secondary verification and refinement of the initially screened "suspected abnormal days." Only those days exhibiting significant, synergistic, and recently occurring anomalies across multiple physiological indicators are ultimately determined to be true abnormal days. This filters out accidental factors and noise interference, providing higher-quality data for subsequent risk level assessments.

[0046] The following is a detailed explanation of the process by which the processing unit determines its risk level: As one possible implementation, the processing unit is specifically used to: determine at least one stage of disease deterioration for the target user based on abnormal days in the observation phase, wherein the stage of disease deterioration includes multiple consecutive abnormal days; and determine the risk level of the target user based on the severity of the deterioration stage, the ability to manage the disease, and the time interval between adjacent stages of disease deterioration, wherein the severity of the deterioration is used to characterize the severity and duration of the deterioration stage, and the ability to manage the disease is used to characterize the target user's ability to control the disease and maintain its stability.

[0047] Based on the above technical solution, this application can analyze the user's risk level from the perspective of the stage of disease deterioration and introduces the dynamic assessment dimension of disease management ability. This allows the application to assess the user's overall disease fluctuation pattern and self-management efficacy over a period of time, rather than just static lung function values, making the risk assessment results more reflective of individual differences and the true long-term risk trend of the user.

[0048] The following details the process by which the treatment unit determines the severity of the disease progression, its disease management capabilities, and the time intervals at each stage of disease deterioration: As one possible implementation, the processing unit is specifically used for: determining the updated outlier value for each abnormal day in any stage of disease deterioration; fitting the updated outlier value of any stage of disease deterioration to obtain a third fitted curve; determining the slope normalization value for each abnormal day in any stage of disease deterioration based on the third fitted curve; multiplying the updated outlier value of any abnormal day by the slope normalization value to obtain a first product; accumulating the first products of each abnormal day to obtain a first accumulated value; calculating the time proportion of the disease deterioration stage in the observation stage; and using the product of the first accumulated value and the time proportion as the deterioration intensity of any stage of disease deterioration.

[0049] As one possible implementation, the processing unit is specifically used to: determine the severity of any deterioration stage, the normalized slope of the deterioration trend, and the duration of the stable stage following any deterioration stage, wherein the normalized slope of the deterioration trend is the normalized slope of the fitted line obtained by fitting the updated outliers of the deterioration stage; the stable stage includes multiple consecutive normal days; normal days are dates in the observation period excluding abnormal days; multiplying the severity of any deterioration stage by the corresponding normalized slope to obtain a second product; multiplying the second product by the reciprocal of the duration of the stable stage to obtain a third product; and using the negative of the third product as an exponent to calculate the power of the natural constant e to obtain the disease management capability.

[0050] It should be noted that COPD treatment requires long-term medical intervention. Due to differences in age, living environment, and self-management ability among COPD patients, if a user does not use bronchodilators regularly, stops medication without authorization, or is exposed to triggering factors such as smoking, their condition may worsen. Once in this stage, the airway inflammation may be gradually suppressed as the user may use antibiotics or other medications to control infection, thus bringing the condition back to a stable stage. Therefore, this application can determine the user's risk level based on their relevant performance during the worsening and stable stages.

[0051] Optionally, within a preset time period, abnormal days are marked as 1, and normal days are marked as 0. This results in a sequence of 0s and 1s. A phase consisting of consecutive 1s is denoted as a disease deterioration phase, and a phase consisting of consecutive 0s is denoted as a disease stabilization phase.

[0052] Calculate the severity of the disease at the e-th stage of deterioration. : Where Y represents the number of abnormal days in the e-th stage of disease deterioration. Indicates the total number of days. This represents the update outlier on the y-th abnormal day within the e-th stage of disease deterioration. This represents the slope normalization value corresponding to the updated outlier on the y-th abnormal day in the e-th stage of disease deterioration. The larger the value, the faster the FEV1 / FVC ratio decreases, and the more rapidly the disease worsens. Unless otherwise specified, the normalization mentioned in this application uses maximum-minimum value normalization. The maximum and minimum values ​​are preset empirical extreme values ​​derived from a large amount of historical experimental data. If the calculated result exceeds the [0,1] interval, it is restricted to the [0,1] range using a truncation function (i.e., if the result is less than 0, it is taken as 0; if it is greater than 1, it is taken as 1) to eliminate the influence of outliers on the evaluation index.

[0053] In the above formula for calculating the intensity of deterioration, This represents the overall worsening effect of the e-th stage of disease progression. This indicates the proportion of time the e-th stage of disease deterioration takes up in the entire observation period. The greater the overall deterioration effect of the e-th stage of disease deterioration and the greater its proportion of time in the entire observation period, the more severe the damage to the user's respiratory function is caused by the e-th stage of deterioration, and the longer it lasts. The greater the likelihood that the user's condition will develop into an acute exacerbation, and the higher the corresponding warning risk level should be.

[0054] Obtain the updated outlier sequence for all days during the e-th stage of disease deterioration. The least squares method is used to obtain the fitted straight line for the sequence.

[0055] Calculate the user's disease management ability at the e-th stage of disease deterioration. ,in, This represents the normalized value of the slope of the fitted line for the e-th stage of disease deterioration. This indicates the severity of the deterioration in the e-th stage of disease progression. This indicates the duration of the next stable phase following the e-th stage of disease progression. The smaller the value, the more likely the user is to maintain a stable condition for a very short period after experiencing the e-th episode of disease deterioration, indicating a poorer ability to manage their condition.

[0056] In the above formula for calculating disease management ability, This indicates the degree of loss of control of the user's condition in the e-th stage of disease deterioration. The greater the degree of loss of control of the user's condition in the e-th stage of disease deterioration, and the shorter the period of stable condition, the more likely the user's condition is to quickly turn from stable to fluctuating or deteriorating again. The risk of repeated disease recurrence is high. The weaker the user's ability to manage their condition, the higher the risk level of repeated disease deterioration.

[0057] Obtain the sequence of time intervals between any two consecutive stages of disease progression. Based on this, this application obtained the severity of the disease progression, the user's disease management ability, and the time interval sequence of the disease progression stages through the above process.

[0058] The following details the process by which the processing unit determines the risk level of a target user based on the intensity of the disease progression, the user's disease management capabilities, and the time interval between adjacent disease progression stages: As one possible implementation, the processing unit is specifically used to: fit the time series of time intervals between adjacent disease deterioration stages, and use the normalized value of the slope of the fitted straight line as the deterioration interval trend; use the product of the reciprocal of the disease management ability and the reciprocal of the deterioration interval trend as the basic risk value; use the reciprocal of the time interval between the most recent disease deterioration stage and the current time as the time-sensitive factor; multiply the basic risk value and the time-sensitive factor, and normalize the product to determine the risk level of the target user.

[0059] Optional, the user's current risk level of illness. Satisfy the following formula: in, express{ The normalized value of the slope of the fitted curve. The smaller the value, the shorter the interval between two deteriorations, the shorter the period of stabilization, and the more likely the condition is to relapse and worsen, indicating a higher risk level. This indicates the user's ability to manage their condition during the most recent stage of deterioration. The higher the value, the more effective the intervention measures taken by the user during the worsening stage, and the less damage to lung function caused by the worsening condition. It represents the time interval between the most recent stage of disease deterioration and the current moment. The smaller the time interval, the greater its reference value for the risk level at the current moment.

[0060] In one possible implementation, after determining the risk level of the user's condition, the risk level of the user's condition is determined. If the risk level is greater than or equal to the third threshold, the patient is considered high-risk; if the risk level is lower... If the risk level is less than the third threshold and greater than or equal to the fourth threshold, the patient is determined to be in a medium-risk state; if the risk level is... If the value is less than the fourth threshold, the patient is determined to be in a low-risk state. As an example, the third threshold is set to 0.8, and the fourth threshold to 0.5. The specific values ​​of the third and fourth thresholds can be set by those skilled in the art based on clinical practice or experience, and this application will not elaborate on these details.

[0061] Optionally, different warnings can be issued under different risk states. For example, when a patient is determined to be in a high-risk state, the warning information sent by the system includes: basic user information (name, medical record number) and current risk level. Specific numerical values ​​and key abnormal data that trigger high risk, such as a 20% drop in FEV1 / FVC compared to the reference day, blood oxygen saturation below 90%, and respiratory rate > 24 breaths / min.

[0062] When a patient is determined to be in a medium-risk state, the system prompts the user that fluctuations in lung function may be caused by insufficient treatment adherence or environmental stimuli. Medical staff should strengthen medication guidance through telephone, online consultation, etc., such as clarifying the medication time and the correct operation of the inhaler, to prevent the condition from deteriorating further.

[0063] When a patient is determined to be in a low-risk state, the system prompts the user to strengthen daily management to prevent the risk from rising further. For example, the system reminds the user to take medication on time and in the correct dosage and prohibits the user from stopping medication without authorization.

[0064] The above provides a detailed description of the solution involved in the intelligent early warning system for dynamic deterioration of COPD provided in this application.

[0065] like Figure 2 As shown, this application also provides a flowchart illustrating an intelligent early warning method for the dynamic deterioration of COPD, as follows: Figure 2 As shown, the method includes the following steps: Step 201: Determine suspected abnormal days based on lung function time series data.

[0066] Step 202: Determine the deviation between the multidimensional physiological time series data of the suspected abnormal day and the multidimensional physiological time series data of the reference day, and the similarity of the deviation of the physiological time series data of different dimensions.

[0067] Step 203: Identify abnormal days from suspected abnormal days based on deviation and similarity.

[0068] Step 204: Determine the risk level of dynamic deterioration of COPD in the target user based on abnormal days.

[0069] Step 205: Conduct risk warning based on the dynamic deterioration risk level of the target user's COPD condition.

[0070] The specific implementation of each of the above steps can be referred to the description in the system above, and will not be repeated here.

[0071] 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.

[0072] 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 intelligent early warning system for dynamic exacerbation of chronic obstructive pulmonary disease, characterized in that, The system comprises: a data acquisition unit configured to acquire lung function time series data and multi-dimensional physiological time series data of a target user in an observation period, the observation period comprising a plurality of consecutive days; a processing unit configured to determine suspected abnormal days according to the lung function time series data, the suspected abnormal days being days in which the lung function data changes by more than a first threshold value; determine a deviation degree of the multi-dimensional physiological time series data of the suspected abnormal days from multi-dimensional physiological time series data of a reference day and a similarity of different-dimensional physiological time series data deviation degrees, the reference day being a day in which the lung function data has the smallest abnormal value and the lung ventilation function is the best; determine an abnormal day from the suspected abnormal days based on the deviation degree and the similarity, and determine a risk level of dynamic exacerbation of the target user's chronic obstructive pulmonary disease based on the abnormal day; and perform risk warning based on the risk level of dynamic exacerbation of the target user's chronic obstructive pulmonary disease.

2. The intelligent early warning system for dynamic exacerbation of COPD conditions as claimed in claim 1 wherein, The lung function time series data comprises forced expiratory volume in one second and forced vital capacity of each day in a plurality of days; and the processing unit is specifically configured to: calculate a first ratio of forced expiratory volume in one second and forced vital capacity of each day; use a normalized value of a difference between the first ratio of each day and the first ratio of the previous day as an initial abnormal value of the lung function data of each day; record a day in which the initial abnormal value is greater than the first threshold value as the suspected abnormal day.

3. The intelligent early warning system for dynamic exacerbation of COPD according to claim 2, characterized in that, The processing unit is further configured to: record a day in which the initial abnormal value is less than or equal to the first threshold value as an initial normal day; select a stable stage of the disease from a plurality of stable stages of the disease, the stable stage of the disease being a plurality of consecutive initial normal days, and select a stable stage of the disease in which the number of initial normal days is the largest as a reference stage; calculate ventilation function of the target user of each day in the reference stage, the ventilation function being positively correlated with the first ratio and negatively correlated with the initial abnormal value; record an initial normal day in which the ventilation function value is the largest in the reference stage as the reference day.

4. The intelligent early warning system for dynamic exacerbation of COPD according to claim 1, wherein, The processing unit is specifically configured to: fit target-dimensional physiological time series data of a target suspected abnormal day to obtain a first fitting curve, and fit physiological time series data of the target dimension of the reference day to obtain a second fitting curve; the target suspected abnormal day being one of the suspected abnormal days, and the target-dimensional physiological time series data being physiological time series data of one dimension of the multi-dimensional physiological time series data; calculate an absolute value of a data difference of the first fitting curve and the second fitting curve at each time point, and a trend weight at each time point, wherein if the slope directions of the first fitting curve and the second fitting curve are the same at a time point, the trend weight is a first value, and if not, the trend weight is a second value, the second value being greater than the first value; based on the trend weight at each time point, weight and sum the absolute value of the difference at the time point, and normalize the result of the weighted sum to obtain a deviation degree of the target dimension of the target suspected abnormal day; and use an average value of the deviation degrees of each dimension of the target suspected abnormal day as the deviation degree of the target suspected abnormal day.

5. The intelligent early warning system for dynamic exacerbation of COPD according to claim 4, wherein, The processing unit is specifically configured to: determine a deviation degree time sequence of the first dimension physiological time sequence data and a deviation degree time sequence of the second dimension physiological time sequence data; the first dimension physiological time sequence data and the second dimension physiological time sequence data are physiological time sequence data of one dimension in the multi-dimension physiological time sequence data, and the first dimension physiological time sequence data and the second dimension physiological time sequence data are different; the deviation degree time sequence includes deviation degrees at different time points; take a similarity between the deviation degree time sequence of the first dimension physiological time sequence data and the deviation degree time sequence of the second dimension physiological time sequence data as a similarity between the first dimension physiological time sequence data and the second dimension physiological time sequence data.

6. The intelligent early warning system for dynamic exacerbation of COPD according to claim 5, wherein, The processing unit is specifically configured to: calculate an updated abnormal value of the target suspected abnormal day based on the deviation degree of the target suspected abnormal day, an average of similarities between all dimension physiological time sequence data of the target suspected abnormal day, and a number of interval days between the target suspected abnormal day and the reference day; the updated abnormal value is positively correlated with the deviation degree and the average of the similarities, and is negatively correlated with the number of interval days; if the updated abnormal value of the target suspected abnormal day is greater than a second threshold value, the target suspected abnormal day is taken as an abnormal day.

7. The intelligent early warning system for dynamic exacerbation of COPD according to claim 2, wherein, The processing unit is specifically configured to: determine at least one condition deterioration stage of the target user based on the abnormal days in the observation stage; the condition deterioration stage includes a plurality of continuous abnormal days; determine a risk level of the target user based on a deterioration intensity of the condition deterioration stage, a condition management capability, and a time interval of adjacent condition deterioration stages; the deterioration intensity is used to represent a deterioration severity and a duration of the condition deterioration stage, and the condition management capability is used to represent an ability of the target user to control and maintain the condition.

8. The intelligent early warning system for dynamic exacerbation of COPD according to claim 7, characterized in that, The processing unit is specifically configured to: determine an updated abnormal value of each abnormal day in any condition deterioration stage; fit the updated abnormal values of the any condition deterioration stage to obtain a third fitting curve, and determine a slope normalization value of each abnormal day in the any condition deterioration stage based on the third fitting curve; multiply the updated abnormal value of any abnormal day by the slope normalization value to obtain a first product; accumulate the first products of each abnormal day to obtain a first accumulated value; calculate a time proportion of the condition deterioration stage in the observation stage; take a product of the first accumulated value and the time proportion as the deterioration intensity of the any condition deterioration stage.

9. The intelligent early warning system for dynamic exacerbation of COPD according to claim 8, characterized in that, The processing unit is specifically configured to: determine a deterioration intensity of the any condition deterioration stage, a slope normalization value of a deterioration trend, and a duration of a condition stable stage after the any condition deterioration stage; the slope normalization value of the deterioration trend is a normalization value of a slope of a fitting straight line obtained by linear fitting of the updated abnormal values of the deterioration stage; the condition stable stage includes a plurality of continuous normal days; the normal day is a date in the observation stage except the abnormal day; multiply the deterioration intensity of the any condition deterioration stage by the corresponding slope normalization value to obtain a second product; multiplying the second product by a reciprocal of a duration of the stable phase of the disease to obtain a third product; taking an exponential function with a reciprocal of the third product as an index and a natural constant e as a base as the disease management capability.

10. The intelligent early warning system for dynamic exacerbation of COPD according to claim 7, wherein, The processing unit is specifically configured to: fit a time sequence of time intervals of adjacent exacerbation phases of the disease, and take a normalized value of a slope of a fitted straight line as an exacerbation interval trend; take a product of a reciprocal of the disease management capability and a reciprocal of the exacerbation interval trend as a basic risk value; take a reciprocal of a time interval between a latest exacerbation phase closest to a current time and the current time as a time-sensitive factor; multiply the basic risk value by the time-sensitive factor, and normalize the product to determine a risk level of the target user.

Images