Respiratory training device and method for pneumology department

Abstract

The invention provides a respiratory training device and method for the pneumology department, and the method comprises the steps: determining a flow volume loop curve through the respiratory flow and tidal volume of a target user during respiratory training, and obtaining the feature deviation degree of the flow volume loop curve; frequency variability of the target user during respiratory training is determined, and respiratory variation values of all respiratory state parameters when the target user is in the abnormal respiratory state are determined through the feature deviation degree and the frequency variability; determining dynamic characteristics of different breathing state parameters when the target user breathes normally, and determining a characteristic distribution domain of all breathing state parameters when the target user is in the normal breathing state according to the dynamic characteristics of the different breathing state parameters; and dynamically adjusting the breathing training strategy of the target user according to the breathing variation value and the feature distribution domain. By means of the scheme, the perception accuracy of the breathing training device on the breathing state of the target user can be enhanced, and therefore the training effect is improved.

Description

Technical Field

[0001] The present application relates to the technical field of respiratory training, and more specifically, to a respiratory training device and method for respiratory medicine. Background Art

[0002] A breathing training device is a medical device used to assist or improve respiratory function. It is widely used in rehabilitation training for respiratory diseases, postoperative recovery, exercise or psychological relaxation, etc. It guides patients to perform regular breathing exercises to enhance lung capacity, improve respiratory control, increase the strength of respiratory muscles, and in some cases relieve anxiety and stress. Common breathing training devices include: handheld spirometer, respiratory resistance trainer, virtual reality breathing training system, biofeedback breathing training device and intelligent breathing training device.

[0003] Existing breathing training devices mainly rely on immersive scenes and biofeedback technology to guide and adjust the patient's breathing pattern. By combining with respiratory sensors, breathing training devices can monitor the patient's breathing frequency, depth and other parameters, and convert the collected data into real-time feedback. However, existing biofeedback technology is usually presented in a visual or auditory way, and does not fully combine the interaction of other senses, resulting in the breathing training device being unable to accurately identify the breathing state of the target user during breathing training, and thus unable to perform precise training based on the patient's breathing state. Therefore, how to enhance the accuracy of the breathing training device's perception of the target user's breathing state, thereby improving the training effect, has become a difficult problem faced by the industry. Summary of the invention

[0004] The present application provides a respiratory training device and method for respiratory medicine, which can enhance the accuracy of the respiratory training device in sensing the respiratory state of a target user, thereby improving the training effect.

[0005] In a first aspect, the present application provides a breathing training method for respiratory medicine, comprising: The respiratory flow and tidal volume of the target user during breathing training are monitored in real time through a respiratory sensor; determining a flow-volume loop curve when a target user breathes according to the respiratory flow and the tidal volume, identifying abnormal features of the flow-volume loop curve based on a long short-term memory network, and thereby obtaining a characteristic deviation of the flow-volume loop curve when the target user performs breathing training; Determine a respiratory frequency sequence when the target user performs respiratory training, perform trend decomposition on the respiratory frequency sequence, obtain frequency variability when the target user performs respiratory training, and then determine respiratory variation values ​​of all respiratory state parameters when the target user is in an abnormal respiratory state through the characteristic deviation and the frequency variability; Acquire breathing data of the target user in a normal breathing state, extract dynamic features of different breathing state parameters of the target user in normal breathing from the breathing data based on multi-scale time series decomposition, and then determine the characteristic distribution domain of all breathing state parameters of the target user in a normal breathing state through the dynamic features of the different breathing state parameters; The breathing training of the target user is dynamically adjusted according to the breathing variation value and the characteristic distribution domain.

[0006] In some embodiments, determining the flow-volume loop curve of the target user's breathing according to the respiratory flow and the tidal volume specifically includes: Extract the respiratory flow in each respiratory cycle, and then obtain the respiratory flow sequence of the target user; Extract the tidal volume in each respiratory cycle, and then obtain the tidal volume sequence of the target user; A flow-volume loop curve when the target user breathes is determined according to the respiratory flow sequence and the tidal volume sequence.

[0007] In some embodiments, the abnormal feature recognition of the flow-volume loop curve based on the long short-term memory network, and then obtaining the characteristic deviation of the flow-volume loop curve when the target user performs breathing training specifically includes: Dividing the flow volume loop curve into time windows to obtain flow volume curve segments in each time window; A long short-term memory network model for identifying abnormal features is constructed based on all flow-volume curve segments; The flow volume curve segment in the current time window is input into the long short-term memory network model to obtain the characteristic deviation when the target user is in an abnormal breathing state in the current breathing cycle.

[0008] In some embodiments, performing trend decomposition on the respiratory frequency sequence to obtain the frequency variability of the target user during the respiratory training specifically includes: Extracting all local extreme values ​​of the target user's breathing frequency from the breathing frequency sequence; Fitting envelopes to all local extreme values ​​to obtain an average envelope of the respiratory frequency sequence; The respiratory frequency sequence is subjected to empirical mode decomposition based on the average envelope to obtain frequency variability of the target user when performing respiratory training.

[0009] In some embodiments, extracting dynamic features of different breathing state parameters of the target user during normal breathing from the breathing data based on multi-scale time series decomposition specifically includes: Performing time series decomposition on the respiratory data to obtain a frequency time-frequency diagram of the target user's respiratory frequency changes, a flow time-frequency diagram of the respiratory flow changes, and a tidal time-frequency diagram of the tidal volume changes; Determine the frequency dynamic characteristics of the target user's breathing according to the frequency time-frequency graph; Determining the dynamic characteristics of the tidal volume when the target user breathes according to the tidal time-frequency diagram; Determine the dynamic characteristics of the flow rate when the target user breathes according to the flow time-frequency graph; The dynamic characteristics of different breathing state parameters of the target user during normal breathing are determined based on the frequency dynamic characteristics, the tidal volume dynamic characteristics, and the flow dynamic characteristics.

[0010] In some embodiments, determining the characteristic distribution domain of all breathing state parameters when the target user is in a normal breathing state through the dynamic characteristics of the different breathing state parameters specifically includes: Determine the dimension-reduced feature time series of all breathing state parameters of the target user during normal breathing based on all dynamic features; Determine the central trend and feature change value of all dynamic features of the target user during normal breathing based on the dimension reduction feature time series; The characteristic distribution domain of all breathing state parameters when the target user is in a normal breathing state is determined according to the central trend and the characteristic variation value.

[0011] In some embodiments, the respiratory state parameters are physiological indicators that reflect different respiratory states of the target user, including: respiratory flow, tidal volume and respiratory frequency.

[0012] In a second aspect, the present application provides a breathing training device for respiratory medicine, which uses the above method to perform breathing training, and the breathing training device includes: A monitoring module, used to monitor the respiratory flow and tidal volume of the target user in real time during respiratory training through a respiratory sensor; a processing module, configured to determine a flow-volume loop curve when a target user breathes according to the respiratory flow and the tidal volume, and to identify abnormal features of the flow-volume loop curve based on a long short-term memory network, thereby obtaining a characteristic deviation of the flow-volume loop curve when the target user performs breathing training; The processing module is used to determine the respiratory frequency sequence of the target user when performing respiratory training, perform trend decomposition on the respiratory frequency sequence, obtain the frequency variability of the target user when performing respiratory training, and then determine the respiratory variation values ​​of all respiratory state parameters when the target user is in an abnormal respiratory state through the characteristic deviation and the frequency variability; The processing module is used to obtain the breathing data of the target user in the normal breathing state, extract the dynamic characteristics of different breathing state parameters of the target user in the normal breathing state from the breathing data based on multi-scale time series decomposition, and then determine the characteristic distribution domain of all breathing state parameters of the target user in the normal breathing state through the dynamic characteristics of the different breathing state parameters; An execution module is used to dynamically adjust the breathing training of the target user according to the breathing variation value and the characteristic distribution domain.

[0013] In a third aspect, the present application provides a computer device, comprising a memory and a processor, wherein the memory stores codes, and the processor is configured to obtain the codes and execute the above-mentioned breathing training method for respiratory medicine.

[0014] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned breathing training method for respiratory medicine is implemented.

[0015] The technical solution provided by the embodiments disclosed in this application has the following beneficial effects: In the respiratory training device and method for respiratory medicine provided by the present application, the respiratory flow and tidal volume of the target user during respiratory training are first monitored in real time by a respiratory sensor; the flow-volume loop curve of the target user during breathing is determined according to the respiratory flow and the tidal volume, and the abnormal characteristics of the flow-volume loop curve are identified based on the long short-term memory network, so as to obtain the characteristic deviation of the flow-volume loop curve when the target user is performing respiratory training; the respiratory frequency sequence of the target user during respiratory training is determined, and the trend of the respiratory frequency sequence is decomposed to obtain the frequency variability of the target user during respiratory training, and then the respiratory variation values ​​of all respiratory state parameters when the target user is in an abnormal respiratory state are determined through the characteristic deviation and the frequency variability; the respiratory data of the target user in a normal respiratory state is obtained, and the dynamic characteristics of different respiratory state parameters of the target user during normal breathing are extracted from the respiratory data based on multi-scale time series decomposition, and then the characteristic distribution domain of all respiratory state parameters when the target user is in a normal respiratory state is determined through the dynamic characteristics of the different respiratory state parameters; the respiratory training of the target user is dynamically adjusted according to the respiratory variation value and the characteristic distribution domain.

[0016] It can be seen that the present application dynamically adjusts the target user's breathing training according to the breathing variation value and the characteristic distribution domain; first, the breathing variation value is determined to obtain the degree to which the target user's breathing state changes over time during the breathing training process. A low breathing variation value indicates that the target user's breathing is too slow or rigid, which may be related to respiratory muscle weakness, lung insufficiency or neuromuscular coordination problems. A high breathing variation value indicates excessive fluctuations in the breathing process, which may be related to factors such as airway instability, respiratory muscle incoordination, and anxiety. By determining the breathing variation value, the target user's respiratory health status can be evaluated, so as to formulate a personalized breathing training plan according to different breathing states; then, determining the characteristic distribution domain can obtain the range of possible fluctuations of the characteristic values ​​representing the target user's breathing state parameters under normal breathing conditions. , the characteristic distribution domain is a dynamic range that is adjusted according to the changes in the physiological state of the target user. It can help evaluate whether the target user's breathing state is normal or in an abnormal fluctuation state, and solve the problem that the breathing training device has limited response ability to abnormal breathing conditions (such as shortness of breath or insufficient breathing) and cannot accurately perform targeted training according to the patient's breathing state; finally, when the respiratory variation value is greater than the upper limit threshold of the characteristic distribution domain, the respiratory electrical stimulation training module is started to perform respiratory electrical stimulation training on the target user, and when the respiratory variation value is less than the lower limit threshold of the characteristic distribution domain, the respiratory resistance training module is started to perform respiratory resistance training on the target user; in summary, based on the above scheme, the perception accuracy of the respiratory training device on the target user's breathing state can be enhanced, thereby improving the training effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is an exemplary flow chart of a breathing training method for respiratory medicine according to some embodiments of the present application; Figure 2 is an exemplary flow chart of determining dynamic features according to some embodiments of the present application; Figure 3 It is a block diagram of structural elements for determining a feature distribution domain according to some embodiments of the present application; Figure 4 is a schematic structural diagram of a respiratory training device for respiratory medicine according to some embodiments of the present application; Figure 5 It is a diagram of the internal structure of a computer device for implementing a breathing training method for respiratory medicine according to some embodiments of the present application. DETAILED DESCRIPTION

[0018] In order to better understand the technical solution of the present application, the technical solution of the present application will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0019] refer to Figure 1, which is an exemplary flow chart of a breathing training method for respiratory medicine according to some embodiments of the present application. The breathing training method 100 for respiratory medicine mainly includes the following steps: In step 101, the respiratory flow and tidal volume of the target user during breathing training are monitored in real time by a respiratory sensor.

[0020] It should be noted that respiratory flow refers to the amount of gas flowing per unit time when the target user breathes; tidal volume refers to the amount of air inhaled or exhaled by the target user with each breath; in specific implementation, the respiratory flow and tidal volume of the target user during breathing training can be monitored and collected in real time through the respiratory sensors (such as thermistor sensors, pressure sensors, flow sensors, etc.) in the breathing training device.

[0021] In addition, it should be noted that both respiratory flow and tidal volume are respiratory state parameters that reflect different respiratory states of the target user. By real-time monitoring of the numerical changes in the target user's respiratory state parameters, it can be detected whether the target user has an abnormal breathing state during breathing training; respiratory flow can reflect the intensity of the target user's breathing; tidal volume can reflect the depth of the target user's breathing.

[0022] In step 102, a flow-volume loop curve of the target user when breathing is determined according to the respiratory flow and the tidal volume, and abnormal features of the flow-volume loop curve are identified based on a long short-term memory network to obtain a characteristic deviation of the flow-volume loop curve when the target user performs breathing training.

[0023] In some embodiments, determining the flow-volume loop curve of the target user's breathing according to the respiratory flow and the tidal volume can be implemented by the following steps: Extract the respiratory flow in each respiratory cycle, and then obtain the respiratory flow sequence of the target user; Extract the tidal volume in each respiratory cycle, and then obtain the tidal volume sequence of the target user; A flow-volume loop curve when the target user breathes is determined according to the respiratory flow sequence and the tidal volume sequence.

[0024] It should be noted that, in the present application, the flow-volume loop curve represents a curve that describes the relationship between gas flow and lung volume during the breathing process. The flow-volume loop curve can display the dynamic changes of the target user from maximum inhalation to maximum exhalation during the breathing cycle, and can be used to detect the functional status of the respiratory system.

[0025] In specific implementation, the respiratory flow in each breathing cycle is extracted, and then the respiratory flow sequence of the target user is obtained, which can be achieved in the following way, that is: the parameter value of the respiratory flow in each breathing cycle can be used as the respiratory flow value, and all the obtained respiratory flow values ​​are arranged in chronological order to obtain the respiratory flow sequence of the target user; wherein, the respiratory flow sequence is a time series composed of the respiratory flow of the target user in each breathing cycle, which reflects the change of the airflow rate of the target user during the breathing training process.

[0026] In specific implementation, the tidal volume in each respiratory cycle is extracted, and then the tidal volume sequence of the target user is obtained, which can be achieved in the following way, that is: the parameter value of the tidal volume in each respiratory cycle can be used as the tidal value, and all the obtained tidal values ​​are arranged in chronological order to obtain the tidal volume sequence of the target user; wherein the tidal volume sequence is a time series composed of the tidal volume of the target user in each respiratory cycle, which reflects the changing trend of the air intake and output of the target user during the breathing training process.

[0027] In a specific implementation, determining the flow-volume loop curve of the target user's breathing according to the respiratory flow sequence and the tidal volume sequence can be achieved in the following manner, namely: a two-dimensional coordinate axis can be initialized, the respiratory flow value in the respiratory flow sequence is used as the ordinate value, the tidal volume value in the tidal volume sequence is used as the abscissa value, and the respiratory flow value and the tidal volume value at each time point are connected in sequence according to the time sequence to generate a complete curve, and then the curve is used as the flow-volume loop curve of the target user's breathing.

[0028] In some embodiments, the abnormal feature recognition of the flow-volume loop curve based on the long short-term memory network, and then obtaining the characteristic deviation of the flow-volume loop curve when the target user performs breathing training can be achieved by the following steps: Dividing the flow volume loop curve into time windows to obtain flow volume curve segments in each time window; A long short-term memory network model for identifying abnormal features is constructed based on all flow-volume curve segments; The flow volume curve segment in the current time window is input into the long short-term memory network model to obtain the characteristic deviation when the target user is in an abnormal breathing state in the current breathing cycle.

[0029] It should be noted that in the present application, the characteristic deviation represents the degree of difference between the characteristic values ​​of the target user's current respiratory flow and tidal volume and their normal range. The characteristic deviation quantifies the deviation between the current respiratory state and the user's normal respiratory state, which helps to judge the severity of abnormal breathing. The characteristic deviation can be used to further determine the respiratory variability value of the target user during respiratory training, and then decide whether electrical stimulation training or respiratory resistance training is needed to correct respiratory abnormalities.

[0030] In specific implementation, the flow-volume loop curve is divided into time windows, and the flow-volume curve segments under each time window are obtained in the following manner, namely: a breathing cycle of the user can be taken as a time window, and the flow-volume loop curve within the time window range is taken as a flow-volume curve segment, and then the flow-volume loop curve is divided according to each time window, thereby obtaining the flow-volume curve segments under each time window, wherein the flow-volume curve segment represents a curve segment that describes the relationship between the respiratory flow and the tidal volume in a complete breathing cycle.

[0031] It should be noted that the long short-term memory network model is an improved recursive neural network, which is specially designed to process and predict long-term dependencies in sequence data and can retain information over a long time span. Therefore, it is very suitable for analyzing patterns and features in time series data, such as speech recognition, natural language processing, and anomaly detection. In this application, the long short-term memory network model is used for anomaly detection in respiratory parameter monitoring. The abnormal features of the target user's breathing process can be identified by analyzing the time series of respiratory state parameters (for example, segmented data of the flow volume loop curve), and the corresponding respiratory training module is triggered by the captured abnormal features to help the user adjust the breathing state; In specific implementation, constructing a long short-term memory network model for identifying abnormal features based on all flow-volume curve segments can be achieved in the following manner, namely: a long short-term memory network model can be initialized, the input layer of the long short-term memory network model is set to input the flow-volume curve segment of the target user's current breathing cycle, and the output layer of the long short-term memory network model is set to output the mean value of the curve deviation between the flow-volume curve segment when the current breathing cycle is in an abnormal state and all the flow-volume curve segments in the memory unit of the long short-term memory network model, and then the flow-volume curve segments under all time windows are used to train the long short-term memory network model to obtain a trained long short-term memory network model that can identify abnormal features.

[0032] In specific implementation, the flow volume curve segment under the current time window is input into the long short-term memory network model, and the characteristic deviation when the target user is in an abnormal breathing state in the current breathing cycle is obtained. This can be achieved in the following way, namely: the flow volume curve segment under the current time window can be used as the input of the input layer of the long short-term memory network model, the long short-term memory network model is executed, and the output value of the output layer in the long short-term memory network model is used as the characteristic deviation when the target user is in an abnormal breathing state.

[0033] In step 103, the respiratory frequency sequence of the target user during respiratory training is determined, and the respiratory frequency sequence is trend decomposed to obtain the frequency variability of the target user during respiratory training, and then the respiratory variation values ​​of all respiratory state parameters when the target user is in an abnormal respiratory state are determined through the characteristic deviation and the frequency variability.

[0034] In some embodiments, determining the breathing frequency sequence of the target user during breathing training may be implemented by the following steps: The breathing frequency of the target user during breathing training is monitored in real time through a breathing sensor; The respiratory frequency in each respiratory cycle is extracted, and then the respiratory frequency sequence of the target user is obtained.

[0035] In specific implementation, real-time monitoring of the breathing frequency of the target user during breathing training through a breathing sensor can be achieved in the following manner, namely: real-time monitoring of the breathing frequency of the target user during breathing training through a breathing sensor can be performed by using a breathing sensor (such as a thermistor sensor, a pressure sensor, a flow sensor, etc.) in a breathing training device to monitor and collect the breathing frequency of the target user during breathing training in real time; wherein, the breathing frequency refers to the number of breaths per unit time when the target user breathes.

[0036] It should be noted that the breathing frequency is also one of the breathing state parameters that reflects the different breathing states of the target user.

[0037] In specific implementation, the respiratory frequency in each respiratory cycle is extracted, and then the respiratory frequency sequence of the target user is obtained, which can be achieved in the following way, namely: the parameter value of the respiratory frequency in each respiratory cycle can be used as the respiratory frequency value, and all the obtained respiratory frequency values ​​are arranged in chronological order to obtain the respiratory frequency sequence of the target user; wherein, the respiratory frequency sequence is a time series composed of the respiratory frequency of the target user in each respiratory cycle, which reflects the stability of the breathing rhythm of the target user during the breathing training process.

[0038] In some embodiments, trend decomposition of the respiratory frequency sequence to obtain the frequency variability of the target user during respiratory training may be achieved by the following steps: Extracting all local extreme values ​​of the target user's breathing frequency from the breathing frequency sequence; Fitting envelopes to all local extreme values ​​to obtain an average envelope of the respiratory frequency sequence; The respiratory frequency sequence is subjected to empirical mode decomposition based on the average envelope to obtain frequency variability of the target user when performing respiratory training.

[0039] It should be noted that, in the present application, frequency variability is an indicator for quantifying the instability of the target user's breathing frequency during the breathing training process. The greater the frequency variability, the more unstable the target user's breathing frequency during the breathing training process. The smaller the frequency variability, the more stable the target user's breathing frequency during the breathing training process. In a specific implementation, all local extreme values ​​of the target user's respiratory frequency can be extracted from the respiratory frequency sequence in the following manner, namely: each respiratory frequency value in the respiratory frequency sequence can be traversed, and if the respiratory frequency value is simultaneously greater than the respiratory frequency values ​​before and after it, the respiratory frequency value is marked as a local maximum value; if the respiratory frequency value is simultaneously less than the respiratory frequency values ​​before and after it, the respiratory frequency value is marked as a local minimum value; and then all local maximum values ​​and local minimum values ​​are taken as local extreme values ​​to obtain all local extreme values ​​of the target user's respiratory frequency; wherein the local extreme value refers to a respiratory frequency value that reaches a maximum or minimum value relative to its adjacent respiratory frequency values ​​within a local range of the respiratory frequency sequence, including a local maximum value and a local minimum value.

[0040] It should be noted that, in the present application, the average envelope refers to a dynamic baseline that reflects the overall trend of the target user's breathing frequency. The average envelope can effectively capture the changes in the breathing frequency in the time dimension and help remove high-frequency noise. In specific implementation, the envelope is fitted to all extreme values ​​to obtain the average envelope of the breathing frequency sequence, which can be achieved in the following manner, namely: all local maxima can be fitted into a smooth curve through an existing data fitting method (for example, a linear interpolation fitting method), and the smooth curve is used as the upper boundary line of the target user's breathing frequency, and all local minima can be fitted into a smooth curve through an existing data fitting method (for example, a linear interpolation fitting method), and the smooth curve is used as the lower boundary line of the target user's breathing frequency, and then the average values ​​of the upper boundary line and the lower boundary line at each time point are calculated, and the smooth curve represented by all average values ​​is used as the average envelope, thereby obtaining the average envelope of the breathing frequency sequence.

[0041] In a specific implementation, the respiratory frequency sequence is subjected to empirical mode decomposition based on the average envelope to obtain the frequency variability of the target user during breathing training, which can be achieved in the following manner, namely: the respiratory frequency sequence can be subjected to empirical mode decomposition based on the average envelope to obtain multiple intrinsic mode functions, and then each intrinsic mode function is subjected to Hilbert transform to obtain the instantaneous frequency of each intrinsic mode function, and the standard deviation and the average value of all instantaneous frequencies are calculated, and then the quotient of the standard deviation and the average value is used as the frequency variability of the target user during breathing training.

[0042] In some embodiments, determining the respiratory variation values ​​of all respiratory state parameters when the target user is in an abnormal respiratory state by using the characteristic deviation and the frequency variability can be implemented by the following steps: determining a weight ratio between the characteristic deviation and the frequency variability; The characteristic deviation and the frequency variability are weightedly combined by the weight ratio to obtain a breathing variability value when the target user performs breathing training.

[0043] It should be noted that in the present application, the respiratory variability is a parameter that measures the degree of change in the characteristic value of the respiratory state parameter of the target user under the respiratory training state. The respiratory variability measures the stability of the target user's breathing based on the fluctuation of respiratory state parameters such as respiratory frequency, tidal volume, and respiratory flow. A low respiratory variability indicates that the target user's breathing is too slow or rigid, which may be related to respiratory muscle weakness, pulmonary insufficiency, or neuromuscular coordination problems. A high respiratory variability indicates excessive fluctuations in the breathing process, which may be related to factors such as airway instability, respiratory muscle incoordination, and anxiety. In specific implementation, determining the weight ratio between the characteristic deviation and the frequency variability can be achieved in the following manner, namely: the weight ratio between the characteristic deviation and the frequency variability can be set to 1.5 (i.e., 3:2). In other embodiments, it can also be set to other ratios, which are not specifically limited here; wherein the weight ratio is the ratio of the weight coefficient of the characteristic deviation to the weight coefficient of the frequency variability.

[0044] In specific implementation, the characteristic deviation and the frequency variability are weightedly combined by the weight ratio to obtain the respiratory variability value of the target user when performing respiratory training. This can be achieved in the following manner, namely: the weight ratio can be first divided by the sum of the weight ratio and 1, and the quotient obtained is used as the weight coefficient of the characteristic deviation, and 1 is divided by the sum of the weight ratio and 1, and the quotient obtained is used as the weight coefficient of the frequency variability, and then the product of the characteristic deviation and its weight coefficient is added to the product of the frequency variability and its weight coefficient, and the sum obtained is used as the respiratory variability value of the target user when performing respiratory training, wherein the sum of the weight coefficient of the characteristic deviation and the weight coefficient of the frequency variability is 1.

[0045] In step 104, the breathing data of the target user in a normal breathing state is obtained, and the dynamic characteristics of different breathing state parameters of the target user in normal breathing are extracted from the breathing data based on multi-scale time series decomposition, and then the characteristic distribution domain of all breathing state parameters of the target user in a normal breathing state is determined through the dynamic characteristics of the different breathing state parameters.

[0046] It should be noted that, in the present application, respiratory data is a data set that reflects the changes in physiological parameters during the user's breathing process; the normal breathing state is a breathing pattern in which the target user breathes stably and regularly in a natural state without external interference or abnormal pressure; in specific implementation, when the target user is in a resting state (for example, sleeping state or lying state), a high-precision respiratory sensor (for example, a flow sensor, a chest strap-type respiratory sensor, etc.) can be used to collect the target user's respiratory flow, tidal volume, respiratory frequency and other physiological parameters within a preset collection time period, and the collection of the values ​​of all the collected physiological parameters is used as the respiratory data of the target user in a normal breathing state, wherein the preset collection time period can be ten minutes, and can also be set to other times in other embodiments, which is not specifically limited here.

[0047] In some embodiments, reference Figure 2 , which is an exemplary flow chart of determining dynamic features according to some embodiments of the present application. In the present application, the dynamic features of different breathing state parameters of the target user during normal breathing are extracted from the breathing data based on multi-scale time series decomposition, which can be implemented by the following steps: First, in step 1041, the respiratory data is time-series decomposed to obtain a frequency time-frequency diagram of the target user's respiratory frequency changes, a flow time-frequency diagram of the respiratory flow changes, and a tidal time-frequency diagram of the tidal volume changes; Secondly, in step 1042, the frequency dynamic characteristics of the target user's breathing are determined according to the frequency time-frequency graph; Then, in step 1043, the dynamic characteristics of the tidal volume when the target user breathes are determined according to the tidal time-frequency diagram; Next, in step 1044, the flow dynamic characteristics of the target user's breathing are determined according to the flow time-frequency graph; Finally, in step 1045, the dynamic characteristics of different breathing state parameters of the target user during normal breathing are determined based on the frequency dynamic characteristics, the tidal volume dynamic characteristics, and the flow dynamic characteristics.

[0048] It should be noted that, in the present application, the frequency time-frequency diagram is an image showing the changes in respiratory frequency in the time and frequency dimensions; the flow time-frequency diagram is an image showing the changes in respiratory flow in the time and frequency dimensions; the tidal time-frequency diagram is an image showing the changes in tidal volume in the time and frequency dimensions. In specific implementation, the respiratory data is time-series decomposed to obtain the frequency time-frequency diagram of the target user's respiratory frequency changes, the flow time-frequency diagram of the respiratory flow changes, and the tidal time-frequency diagram of the tidal volume changes. This can be achieved in the following way, namely: first, the time series data composed of the parameter values ​​of the target user's respiratory frequency can be extracted from the respiratory data, and the time series data can be time-series decomposed through the existing time series decomposition method (for example: wavelet transform) to obtain the frequency time-frequency diagram of the target user's respiratory frequency changes, and then the flow time-frequency diagram of the target user's respiratory flow changes and the tidal time-frequency diagram of the tidal volume changes can be obtained using the same method as the frequency time-frequency diagram.

[0049] In addition, it should be noted that, in the present application, the frequency dynamic feature is a feature that measures the changing trend of the respiratory frequency over time. The frequency dynamic feature reflects the dynamic changes of the respiratory frequency at different time points, and can reveal the stability and rhythmicity of the respiratory frequency and the response to external stimuli or training. In specific implementation, the frequency dynamic feature of the target user's breathing can be determined according to the frequency time-frequency graph. The following method can be used, namely: the existing peak detection algorithm (for example: derivative method) can be used to locate the frequency peak in each breathing cycle in the frequency time-frequency graph, and the time series composed of all frequency peaks is used as the frequency dynamic feature of the target user's breathing.

[0050] In specific implementation, determining the dynamic characteristics of the tidal volume when the target user breathes according to the tidal time-frequency diagram can be achieved in the following manner, namely: the existing peak detection algorithm (for example, the derivative method) can be used to locate the tidal volume peaks of exhalation and inhalation in each breathing cycle in the tidal time-frequency diagram, and the mean of the tidal volume peaks of exhalation and inhalation is calculated as the tidal volume mean, and the time series composed of all tidal volume means is used as the dynamic characteristics of the tidal volume when the target user breathes; it should be noted that in the present application, the dynamic characteristics of tidal volume are characteristics that measure the changing trend of tidal volume over time. The dynamic characteristics of tidal volume can help measure the breathing depth, breathing control and adaptability of the respiratory system of the target user, and are widely used in scenarios such as breathing training, health monitoring and abnormal breathing detection.

[0051] It should be noted that, in the present application, the flow dynamic characteristics are characteristics that measure the time-varying characteristics of the inhalation and exhalation flow during the breathing process, and the flow dynamic characteristics reflect the dynamic performance of the respiratory system by observing the speed of gas flow during the breathing cycle; in specific implementation, the flow dynamic characteristics of the target user's breathing can be determined based on the flow time-frequency diagram in the following manner, namely: the existing peak detection algorithm (for example: derivative method) can be used to locate the peak flow of exhalation and inhalation in each breathing cycle in the flow time-frequency diagram, the mean of the peak flow of exhalation and inhalation is calculated as the flow mean, and the time series composed of all flow means is used as the flow dynamic characteristics of the target user's breathing.

[0052] In addition, it should be noted that the dynamic characteristics are characteristics that reflect the overall breathing rhythm and changing trends of the target user during normal breathing; in specific implementation, the dynamic characteristics of different respiratory state parameters of the target user during normal breathing are determined based on the frequency dynamic characteristics, the tidal volume dynamic characteristics and the flow dynamic characteristics. The following method can be used to achieve this, namely: the frequency dynamic characteristics are used as the dynamic characteristics of the breathing frequency of the target user during normal breathing, the tidal volume dynamic characteristics are used as the dynamic characteristics of the tidal volume of the target user during normal breathing, and the flow dynamic characteristics are used as the dynamic characteristics of the respiratory flow of the target user during normal breathing, thereby obtaining the dynamic characteristics of different respiratory state parameters of the target user during normal breathing.

[0053] In some embodiments, determining the characteristic distribution domain of all breathing state parameters when the target user is in a normal breathing state through the dynamic characteristics of the different breathing state parameters can be achieved by the following steps: Determine the dimension-reduced feature time series of all breathing state parameters of the target user during normal breathing based on all dynamic features; Determine the central trend and feature change value of all dynamic features of the target user during normal breathing based on the dimension reduction feature time series; The characteristic distribution domain of all breathing state parameters when the target user is in a normal breathing state is determined according to the central trend and the characteristic variation value.

[0054] In specific implementation, the dynamic features of all respiratory state parameters of the target user during normal breathing can be determined based on all dynamic features in the following manner, namely: all dynamic features can be combined into a multidimensional dynamic feature vector, and the existing data dimensionality reduction method (for example: principal component analysis method) can be used to synthesize multiple time series in the multidimensional dynamic feature vector into a feature time series, and then the feature time series is used as the reduced dimensionality feature time series of all respiratory state parameters of the target user during normal breathing; wherein the reduced dimensionality feature time series is a one-dimensional vector that reflects the overall breathing rhythm and changing trend of the target user during normal breathing.

[0055] It should be noted that, in the present application, the central tendency is the central position of the characteristic value distribution of the target user's breathing state parameters under normal breathing state; the characteristic change value is a parameter that measures the degree of change of the characteristic values ​​of all breathing state parameters of the target user under normal breathing state; in specific implementation, the central tendency and characteristic change value of all dynamic features of the target user when breathing normally are determined based on the reduced dimension feature time series, which can be achieved in the following way, namely: the average value of all characteristic values ​​in the dynamic feature can be calculated first, and the average value is used as the central tendency of the dynamic feature when the target user breathes normally, and then the standard deviation of all characteristic values ​​in the dynamic feature is calculated, and the standard deviation is used as the characteristic change value of all dynamic features when the target user breathes normally.

[0056] In specific implementation, determining the characteristic distribution domain of all respiratory state parameters when the target user is in a normal breathing state based on the central trend and the characteristic change value can be achieved in the following manner, namely: an interval can be initialized, and the sum of the central trend and the characteristic change value is used as the upper threshold of the interval, and the difference between the central trend and the characteristic change value is used as the lower threshold of the interval, and then the interval is used as the characteristic distribution domain of all respiratory state parameters when the target user is in a normal breathing state; it should be noted that in the present application, the characteristic distribution domain represents the range in which the characteristic values ​​of the respiratory state parameters of the target user in a normal breathing state may fluctuate. The characteristic distribution domain is a dynamic range that is adjusted according to changes in the individual physiological state, which can help evaluate whether the target user's breathing state is normal or whether it is in an abnormal fluctuation state.

[0057] In step 105, the breathing training of the target user is dynamically adjusted according to the breathing variation value and the characteristic distribution domain.

[0058] In some embodiments, dynamically adjusting the breathing training of the target user according to the breathing variation value and the characteristic distribution domain can be implemented by the following steps: When the respiratory variation value is greater than the upper threshold of the characteristic distribution domain, starting the respiratory electrical stimulation training module to perform respiratory electrical stimulation training on the target user; When the respiratory variation value is less than the lower threshold of the characteristic distribution domain, the respiratory resistance training module is started to perform respiratory resistance training on the target user.

[0059] It should be noted that in the present application, respiratory electrical stimulation training is a training method for adjusting respiratory function by applying electrical stimulation to respiratory-related muscles; respiratory resistance training is a training method for exercising respiratory muscles by applying a certain resistance during inhalation and exhalation.

[0060] In specific implementation, starting the respiratory electrical stimulation training module to conduct respiratory electrical stimulation training for the target user can be achieved in the following manner, namely: the respiratory muscles can be electrically stimulated by connecting electrode patches placed on respiratory muscles such as the diaphragm, rectus abdominis or intercostal muscles, and the respiratory state parameters can be monitored in real time during the electrical stimulation process, and the intensity and frequency of the electrical stimulation can be dynamically adjusted to gradually stabilize the breathing of the target user.

[0061] It should be noted that when the respiratory variation value is greater than the upper threshold of the characteristic distribution domain, it means that the fluctuation of the respiratory state parameters of the target user is too drastic, and the target user is in a state of shortness of breath or difficulty breathing, indicating that the target user may have problems such as poor respiratory nerve control, airway obstruction, anxiety, etc. Respiratory electrical stimulation training can apply electrical stimulation to the respiratory muscles (such as diaphragm, abdominal muscles, etc.) through electrical stimulation equipment to help users restore a stable respiratory rhythm, improve fluctuations in respiratory flow and tidal volume, and reduce respiratory variation values.

[0062] In specific implementation, starting the respiratory resistance training module to perform respiratory resistance training on the target user can be achieved in the following manner, namely: a respiratory training device with adjustable resistance (for example, a respiratory resistance mask or a handheld respiratory trainer) can be used to guide the target user to perform deep inhalation and exhalation, and slowly apply resistance to the respiratory muscles starting from a lower resistance throughout the entire breathing cycle, and monitor the respiratory state parameters in real time to dynamically adjust the intensity range of the resistance training.

[0063] It should be noted that when the respiratory variation value is less than the lower threshold of the characteristic distribution domain, it means that the fluctuation of the respiratory state parameters of the target user is too slow, and the target user is in a state of slow breathing or insufficient breathing, indicating that the target user's respiratory system may have some degree of functional deficiency, such as respiratory muscle weakness, lung insufficiency or neuromuscular coordination problems. Respiratory resistance training can impose a certain amount of airflow resistance during inhalation or exhalation, forcing the user's respiratory muscles to work under greater resistance, thereby enhancing the strength and endurance of the respiratory muscles, making the breathing process more flexible and adaptable, and increasing the respiratory variation value.

[0064] In addition, in another aspect of the present application, in some embodiments, the present application provides a breathing training device for respiratory medicine, which uses the above method to perform breathing training, referring to Figure 4 , which is a schematic diagram of the structure of a respiratory training device for respiratory department according to some embodiments of the present application, the respiratory training device 400 for respiratory department includes: a monitoring module 401, a processing module 402 and an execution module 403, which are respectively described as follows: Monitoring module 401, in this application, monitoring module 401 is mainly used to monitor the respiratory flow and tidal volume of the target user during breathing training in real time through a respiratory sensor; Processing module 402, in the present application, the processing module 402 is mainly used to determine the flow volume loop curve when the target user breathes according to the respiratory flow and the tidal volume, and to identify abnormal features of the flow volume loop curve based on the long short-term memory network, so as to obtain the characteristic deviation of the flow volume loop curve when the target user performs breathing training; It should be noted that the processing module 402 in the present application is also used to determine the respiratory frequency sequence of the target user when performing respiratory training, perform trend decomposition on the respiratory frequency sequence, obtain the frequency variability of the target user when performing respiratory training, and then determine the respiratory variation values ​​of all respiratory state parameters when the target user is in an abnormal respiratory state through the characteristic deviation and the frequency variability; In addition, it should be noted that the processing module 402 in the present application is also used to obtain the breathing data of the target user in the normal breathing state, extract the dynamic characteristics of different breathing state parameters of the target user in the normal breathing state from the breathing data based on multi-scale time series decomposition, and then determine the characteristic distribution domain of all breathing state parameters when the target user is in the normal breathing state through the dynamic characteristics of the different breathing state parameters; Execution module 403, in the present application, execution module 403 is mainly used to dynamically adjust the breathing training of the target user according to the breathing variation value and the characteristic distribution domain.

[0065] Each module in the respiratory training device for respiratory department can be implemented in whole or in part by software, hardware and their combination. Each module can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to each module.

[0066] In addition, in one embodiment, the present application provides a computer device, which may be a server, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data on a breathing training method for respiratory medicine. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a breathing training method for respiratory medicine is implemented.

[0067] Those skilled in the art will understand that Figure 5The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0068] In one embodiment, a computer device is also provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps in the above-mentioned breathing training method embodiment for respiratory department are implemented.

[0069] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which, when executed by a processor, implements the steps in the above-mentioned breathing training method embodiment for respiratory medicine.

[0070] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program includes computer instructions, the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps in the above-mentioned breathing training method embodiment for respiratory department.

[0071] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0072] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0073] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A breathing training method for respiratory medicine, characterized in that: The steps include: The respiratory flow and tidal volume of the target user during breathing training are monitored in real time through a respiratory sensor; Determine a flow-volume loop curve when the target user breathes according to the respiratory flow and the tidal volume, identify abnormal features of the flow-volume loop curve based on a long short-term memory network, and then obtain a characteristic deviation of the flow-volume loop curve when the target user performs breathing training; Determine a respiratory frequency sequence when the target user performs respiratory training, perform trend decomposition on the respiratory frequency sequence, obtain frequency variability when the target user performs respiratory training, and then determine respiratory variation values ​​of all respiratory state parameters when the target user is in an abnormal respiratory state through the characteristic deviation and the frequency variability; Acquire breathing data of the target user in a normal breathing state, extract dynamic features of different breathing state parameters of the target user in normal breathing from the breathing data based on multi-scale time series decomposition, and then determine the characteristic distribution domain of all breathing state parameters of the target user in a normal breathing state through the dynamic features of the different breathing state parameters; The breathing training of the target user is dynamically adjusted according to the breathing variation value and the characteristic distribution domain.

2. The method according to claim 1, characterized in that Determining the flow volume loop curve when the target user breathes according to the respiratory flow and the tidal volume specifically includes: Extract the respiratory flow in each respiratory cycle, and then obtain the respiratory flow sequence of the target user; Extract the tidal volume in each respiratory cycle, and then obtain the tidal volume sequence of the target user; A flow-volume loop curve when the target user breathes is determined according to the respiratory flow sequence and the tidal volume sequence.

3. The method according to claim 1, characterized in that The abnormal feature recognition of the flow volume loop curve is performed based on the long short-term memory network, and then the characteristic deviation of the flow volume loop curve when the target user performs breathing training is obtained, which specifically includes: Dividing the flow volume loop curve into time windows to obtain flow volume curve segments in each time window; A long short-term memory network model for identifying abnormal features is constructed based on all flow-volume curve segments; The flow volume curve segment in the current time window is input into the long short-term memory network model to obtain the characteristic deviation when the target user is in an abnormal breathing state in the current breathing cycle.

4. The method according to claim 1, characterized in that The trend decomposition of the respiratory frequency sequence is performed to obtain the frequency variability of the target user during the respiratory training, specifically including: Extracting all local extreme values ​​of the target user's breathing frequency from the breathing frequency sequence; Fitting envelopes to all local extreme values ​​to obtain an average envelope of the respiratory frequency sequence; The respiratory frequency sequence is subjected to empirical mode decomposition based on the average envelope to obtain frequency variability of the target user when performing respiratory training.

5. The method according to claim 1, characterized in that Extracting dynamic features of different breathing state parameters of the target user during normal breathing from the breathing data based on multi-scale time series decomposition specifically includes: Performing time series decomposition on the respiratory data to obtain a frequency time-frequency diagram of the target user's respiratory frequency changes, a flow time-frequency diagram of the respiratory flow changes, and a tidal time-frequency diagram of the tidal volume changes; Determine the frequency dynamic characteristics of the target user's breathing according to the frequency time-frequency graph; Determining the dynamic characteristics of the tidal volume when the target user breathes according to the tidal time-frequency diagram; Determine the dynamic characteristics of the flow rate when the target user breathes according to the flow time-frequency graph; The dynamic characteristics of different breathing state parameters of the target user during normal breathing are determined based on the frequency dynamic characteristics, the tidal volume dynamic characteristics, and the flow dynamic characteristics.

6. The method according to claim 1, characterized in that Determining the characteristic distribution domain of all breathing state parameters when the target user is in a normal breathing state through the dynamic characteristics of the different breathing state parameters specifically includes: Determine the dimension-reduced feature time series of all breathing state parameters of the target user during normal breathing based on all dynamic features; Determine the central trend and feature change value of all dynamic features of the target user during normal breathing based on the dimension reduction feature time series; The characteristic distribution domain of all breathing state parameters when the target user is in a normal breathing state is determined according to the central trend and the characteristic variation value.

7. The method according to claim 1, characterized in that The respiratory state parameters are physiological indicators that reflect different respiratory states of the target user, including: respiratory flow, tidal volume and respiratory frequency.

8. A breathing training device for respiratory department, which adopts the method according to any one of claims 1 to 7 to perform breathing training, characterized in that: The breathing training device comprises: A monitoring module, used to monitor the respiratory flow and tidal volume of the target user in real time during respiratory training through a respiratory sensor; a processing module, configured to determine a flow-volume loop curve when a target user breathes according to the respiratory flow and the tidal volume, and to identify abnormal features of the flow-volume loop curve based on a long short-term memory network, thereby obtaining a characteristic deviation of the flow-volume loop curve when the target user performs breathing training; The processing module is used to determine the respiratory frequency sequence of the target user when performing respiratory training, perform trend decomposition on the respiratory frequency sequence, obtain the frequency variability of the target user when performing respiratory training, and then determine the respiratory variation values ​​of all respiratory state parameters when the target user is in an abnormal respiratory state through the characteristic deviation and the frequency variability; The processing module is further used to obtain breathing data of the target user in a normal breathing state, extract dynamic features of different breathing state parameters of the target user in normal breathing from the breathing data based on multi-scale time series decomposition, and then determine the feature distribution domain of all breathing state parameters of the target user in a normal breathing state through the dynamic features of the different breathing state parameters; The execution module is used to dynamically adjust the breathing training of the target user according to the breathing variation value and the characteristic distribution domain.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the breathing training method for respiratory medicine according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the breathing training method for respiratory medicine according to any one of claims 1 to 7 are implemented.

Images