A method and system for detecting human-machine discoordination based on single-port breathing waveforms

Abstract

The present invention discloses a method, system, device and computer-readable storage medium for detecting human-machine dissonance based on single-breath waveform. The method includes: obtaining single-breath waveform data of a person to be tested; the single-breath waveform data includes flow rate waveform data and airway pressure waveform data; inputting the single-breath waveform data of the person to be tested into a machine learning detection model to obtain the type of human-machine dissonance of the single-breath waveform data of the person to be tested; and outputting the type of human-machine dissonance of the single-breath waveform data of the person to be tested. The present invention can realize automatic segmentation of breaths based on real-time or offline flow rate waveforms, and effectively ensure the accuracy of breath segmentation; at the same time, it can also effectively identify and classify human-machine asynchrony or human-machine dissonance based on single-breath waveforms.

Description

Technical Field

[0001] The present invention relates to the field of respiratory monitoring, and more particularly, to a method and system for detecting human-machine dissonance based on single-breath waveforms. Background Art

[0002] Mechanical ventilation refers to the situation where when the respiratory organs cannot maintain normal gas exchange, that is, when respiratory failure occurs, a ventilator is used to replace or assist the work of the respiratory muscles. Mechanical ventilation buys treatment time and creates conditions for various types of respiratory failure caused by various reasons clinically, as well as other diseases that require respiratory function support.

[0003] Invasive mechanical ventilation treatment is an important treatment method for critically ill patients with respiratory failure. However, inappropriate invasive mechanical ventilation may cause ventilator-related lung injuries such as barotrauma and volume trauma. Appropriate monitoring during mechanical ventilation helps to avoid adverse reactions of mechanical ventilation. During the implementation of invasive mechanical ventilation monitoring, the collected flow rate and pressure waveforms are usually continuous waveforms. However, most monitoring parameters, such as peak airway pressure, plateau airway pressure, positive end-expiratory pressure, etc., are measured based on single-breath. Therefore, it is crucial to accurately identify and segment each breath in the continuous ventilator waveform data, and the segmentation result directly affects the acquisition of monitoring parameters during invasive mechanical ventilation.

[0004] During invasive mechanical ventilation treatment, the ventilator mainly relies on the opening and closing of the inspiratory valve and expiratory valve, as well as the data monitored by the flow sensor to determine inhalation and exhalation. A complete inhalation process plus a complete exhalation process is determined as a complete breath. The start of the inhalation phase is defined as the start of the breath, and the end of the exhalation phase (i.e., the start of the inhalation phase of the next breath) is defined as the end moment of the breath. Currently, there is no technology for automatically segmenting breaths based on flow waveforms; for the analysis of offline data, the methods for breath segmentation mainly include manual segmentation and segmentation based on the periodic change rules and characteristics of flow or airway pressure waveforms. The above methods mainly have the following defects: 1. The algorithms and monitoring systems built into the ventilator can perform automatic splitting of breaths, but the algorithms of different ventilator manufacturers are different and not publicly available, resulting in different results of automatic breath segmentation between different brands of ventilators, which are not comparable. In addition, the results of automatic segmentation are difficult to export, bringing difficulties to subsequent offline analysis and data processing; 2. The manual segmentation method has the highest accuracy, but it is time-consuming and laborious, and it is difficult to segment the massive amount of respiratory monitoring data generated at the bedside one by one; 3. Segmenting breaths based on the rules of periodic change rules and characteristics of flow or airway pressure can achieve a certain degree of automation, but for waveforms with perturbations, especially when there is a PVA phenomenon, the accuracy is insufficient.

[0005] The phenomenon of patient-ventilator asynchrony (PVA), which is caused by the mismatch in amplitude or phase between the patient's demand and the assistance provided by the ventilator during mechanical ventilation, can be harmful to the patient. Accurate respiratory segmentation helps in the identification and classification of PVA, thereby improving the accuracy of the PVA automatic recognition algorithm and providing the possibility to give auxiliary treatment decisions for different PVA types. And accurately identifying and detecting the types of patient-ventilator asynchrony or patient-ventilator dyssynchrony is crucial for the treatment of patients. Summary of the Invention

[0006] The present invention aims to solve at least one of the technical problems existing in the prior art. For this purpose, the present invention provides a method for detecting patient-ventilator dyssynchrony based on a single-breath waveform, which can realize automatic segmentation of breaths based on real-time or offline flow rate waveforms and effectively ensure the accuracy of respiratory segmentation; at the same time, it can also effectively identify and classify patient-ventilator asynchrony or patient-ventilator dyssynchrony based on a single-breath waveform.

[0007] The present application discloses a method for detecting patient-ventilator dyssynchrony based on a single-breath waveform, including:

[0008] Obtaining single-breath waveform data of a subject to be measured; the single-breath waveform data includes flow rate waveform data and airway pressure waveform data;

[0009] Inputting the single-breath waveform data of the subject to be measured into a machine learning detection model to obtain the type of patient-ventilator dyssynchrony of the single-breath waveform data of the subject to be measured;

[0010] Outputting the type of patient-ventilator dyssynchrony of the single-breath waveform data of the subject to be measured.

[0011] The detection method further includes: preprocessing the single-breath waveform data of the subject to be measured to obtain the preprocessed single-breath waveform data;

[0012] Optionally, the determination method of the machine learning detection model includes: inputting pre-labeled single-breath waveform data, extracting features from the pre-labeled single-breath waveform data to obtain feature data after feature extraction, and using the feature data after feature extraction to construct a machine learning detection model to obtain a machine learning detection model.

[0013] The detection method further includes: obtaining or plotting FV loop and / or PV loop waveform image data according to the preprocessed single-breath waveform data, and outputting the FV loop and / or PV loop waveform image data; the single-breath waveform data further includes volume data obtained based on flow rate data;

[0014] Feature extraction is performed on the FV loop and / or PV loop waveform image data to obtain the FV loop and / or PV loop waveform image data after feature extraction as feature data; the FV loop and / or PV loop waveform image data after feature extraction is input into a machine learning detection model to obtain the type of human-machine imbalance of the single-breath waveform data of the subject to be tested.

[0015] Optionally, the FV loop waveform image data is obtained based on flow rate data and volume data. The abscissa of the FV loop waveform image data is volume data V, and the ordinate is flow rate data F; the PV loop waveform image data is obtained based on airway pressure data and volume data. The abscissa of the PV loop waveform image data is volume data V, and the ordinate is airway pressure data P.

[0016] Optionally, the flow rate waveform data is the flow rate waveform data based on the breathing marking time; the breathing marking time is Time, and Time is set according to the sampling frequency hz of the ventilator and the data sequence number number. The calculation formula is: Time = (number - 1) / hz.

[0017] Optionally, the flow rate waveform data is the flow rate waveform data with abscissa Time and ordinate flow rate data.

[0018] Optionally, the airway pressure waveform data is the airway pressure waveform data with abscissa Time and ordinate airway pressure data.

[0019] Optionally, the volume waveform data is the volume waveform data with abscissa Time and ordinate volume data; the calculation formula for the volume data is:

[0020] The types of human-machine imbalance of the single-breath waveform data of the subject to be tested include: normal / abnormal or normal / abnormal subtypes.

[0021] The single-breath waveform data of the subject to be tested includes: single-breath waveform data segmented based on the flow rate waveform.

[0022] The method or steps for obtaining the single-breath waveform data segmented based on the flow rate waveform include:

[0023] Obtain the breathing waveform data of the subject to be tested.

[0024] Standardize the respiratory waveform data of the subject to obtain the standardized respiratory waveform data; the standardized respiratory waveform data includes: finding the points where the flow rate crosses zero and the derivative is greater than zero from the flow rate waveform data based on the respiratory marking time, and respectively recording the time indices of these points, to obtain the standardized respiratory waveform data; optionally, the preprocessing is to traverse the respiratory waveform data in sequence, at least once;

[0025] Divide the standardized respiratory waveform data into respiratory data with multiple respiratory cycles to obtain standardized respiratory data with multiple respiratory cycles; the division is to find all the points where the flow rate crosses zero and the derivative is greater than zero; perform feature extraction on the standardized respiratory data with multiple respiratory cycles to obtain the respiratory data with multiple respiratory cycles after feature extraction as feature data; input the feature data of the subject into a classification model to obtain the classification result of the feature data.

[0026] The standardized respiratory data with multiple respiratory cycles includes: respectively dividing the adjacent points to form waveforms corresponding to each breath, to obtain the respiratory data of the multiple respiratory cycles;

[0027] Optionally, respectively take the nth and (n + 1)th adjacent points, and define the interval between the nth and (n + 1)th points as a single respiratory cycle; where 1 ≤ n < N - 1, and N is the total number of time indices of all the points.

[0028] The respiratory data with multiple respiratory cycles after feature extraction as feature data includes: the respiratory time interval of a single breath, the flow rate change of a single breath, the ratio of the exhalation duration of a single breath to the total respiratory time of a single breath, and the rising slope of the exhalation waveform of a single breath;

[0029] Optionally, the respiratory time interval of a single breath is: the time interval a between the time indices corresponding to the adjacent points; the value range of a is 0s < a < 1s;

[0030] Optionally, the flow rate change of a single breath is: the flow rate change b between the time indices corresponding to the adjacent points; the value range of b is 0L / min < b < 20L / min;

[0031] Optionally, the ratio of the exhalation duration of a single breath to the total respiratory time of a single breath is: the ratio c% of the exhalation duration between the time indices corresponding to the adjacent points to the total respiratory time between the time indices corresponding to the adjacent points; the value range of c% is 20% - 80%;

[0032] Optionally, the rising slope of the exhalation waveform of the single - mouth breathing is: the rising slope d of the exhalation waveform between the time indices corresponding to the adjacent points; within the first 100 ms of the beginning of inhalation, the value range of d is 10 - 50 L / min / s.

[0033] The method or steps of inputting the characteristic data of the subject into the classification model to obtain the classification result of the characteristic data of the subject include:

[0034] Input the breathing time interval of the single breathing cycle of the subject into the classification model, and determine whether the breathing time interval of the single breathing cycle of the subject falls within the range of a; if the breathing time interval of the single breathing cycle of the subject falls within the range of a, output yes, and enter the stage of inputting the flow rate change of the single breathing cycle of the subject into the classification model; otherwise, output no and terminate the operation;

[0035] Input the flow rate change of the single breathing cycle of the subject into the classification model, and determine whether the flow rate change of the single breathing cycle of the subject falls within the range of b; if the flow rate change of the single breathing cycle of the subject falls within the range of b, output yes, and enter the stage of inputting the ratio of the exhalation duration of the single breathing cycle of the subject to the total breathing time of the single breathing cycle into the classification model; otherwise, output no and terminate the operation;

[0036] Input the ratio of the exhalation duration of the single breathing cycle of the subject to the total breathing time of the single breathing cycle into the classification model, and determine whether the ratio of the exhalation duration of the single breathing cycle of the subject to the total breathing time of the single breathing cycle falls within the range of c; if the ratio of the exhalation duration of the single breathing cycle of the subject to the total breathing time of the single breathing cycle falls within the range of c, output yes, and enter the stage of inputting the rising slope of the exhalation waveform of the single breathing cycle of the subject into the classification model; otherwise, output no and terminate the operation;

[0037] Input the rising slope of the exhalation waveform of the single breathing cycle of the subject into the classification model, and determine whether the rising slope of the exhalation waveform of the single breathing cycle of the subject falls within the range of d; if the rising slope of the exhalation waveform of the single breathing cycle of the subject falls within the range of d, output yes, and output the classification result of the breathing waveform data of the subject; otherwise, output no and terminate the operation.

[0038] A human - machine maladjustment detection device based on a single - mouth breathing waveform, the device includes: a memory and a processor;

[0039] The memory is used for storing program instructions;

[0040] The processor is used to call program instructions, and when the program instructions are executed, it is used to execute the above-mentioned human-machine imbalance detection method based on single-mouth breathing waveform.

[0041] A human-machine imbalance detection system based on single-mouth breathing waveform, comprising:

[0042] An acquisition unit, used for acquiring single-mouth breathing waveform data of the subject to be tested; the single-mouth breathing waveform data includes flow velocity waveform data and airway pressure waveform data;

[0043] A processing unit, used for inputting the single-mouth breathing waveform data of the subject to be tested into a machine learning detection model to obtain the human-machine imbalance type of the single-mouth breathing waveform data of the subject to be tested;

[0044] The classification unit is used to output the human-machine imbalance type of the single-mouth breathing waveform data of the subject.

[0045] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the above-mentioned human-machine imbalance detection method based on single-mouth breathing waveform.

[0046] The above-mentioned device is used in the extraction of respiratory parameters; optionally, after each breath is segmented based on the real-time or offline flow rate waveform, the changes in airway pressure, esophageal pressure and volume of each breath can extract many characteristic parameters for monitoring respiratory therapy.

[0047] This application has the following beneficial effects:

[0048] 1. This application innovatively discloses a method for detecting human-machine imbalance based on a single-mouth breathing waveform, which effectively ensures the accuracy of single-mouth breathing segmentation, standardizes the breathing segmentation results, and is comparable; and based on the single-mouth breathing results with high segmentation accuracy, human-machine asynchrony or human-machine imbalance is effectively identified and classified, providing the possibility of making auxiliary treatment decisions for different PVA types;

[0049] 2. The automatic breathing segmentation method proposed in this application overcomes the defects of time-consuming and labor-intensive manual segmentation and insufficient accuracy when using rules based on the periodic change law and characteristics of flow rate or airway pressure to perform breathing segmentation;

[0050] 3. The present application innovatively discloses a method for automatic respiratory segmentation based on real-time or offline flow velocity waveforms, which automatically segments the respiratory waveform into single-mouth breathing, making it convenient for doctors to quickly identify whether the patient's breathing is normal, greatly shortening the doctor's diagnosis time, and gaining precious treatment time for ICU patients; in addition, by referring to this segmentation rule, a computer program can be used to realize automatic segmentation of the respiratory waveform, facilitating the subsequent identification and monitoring of abnormal respiratory waveforms. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those skilled in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0052] Figure 1 is a schematic flowchart of a method for detecting human - machine dis - coordination based on a single - breath waveform provided by an embodiment of the present invention;

[0053] Figure 2 is a schematic diagram of a device for detecting human - machine dis - coordination based on a single - breath waveform provided by an embodiment of the present invention;

[0054] Figure 3 is a schematic flowchart of a system for detecting human - machine dis - coordination based on a single - breath waveform provided by an embodiment of the present invention;

[0055] Figure 4 is a classification flowchart of a method for detecting human - machine dis - coordination based on a single - breath waveform provided by an embodiment of the present invention;

[0056] Figure 5 is a schematic diagram of a point where the flow velocity crosses zero and the derivative is greater than zero in a method for detecting human - machine dis - coordination based on a single - breath waveform provided by an embodiment of the present invention;

[0057] Figure 6 is a schematic diagram of a time interval a in a method for detecting human - machine dis - coordination based on a single - breath waveform provided by an embodiment of the present invention;

[0058] Figure 7 is a schematic diagram of a flow velocity change b in a method for detecting human - machine dis - coordination based on a single - breath waveform provided by an embodiment of the present invention;

[0059] Figure 8 is a schematic diagram of an exhalation waveform rising slope d in a method for detecting human - machine dis - coordination based on a single - breath waveform provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] In order to enable those skilled in the art to better understand the solutions of the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention.

[0061] In some of the processes described in the specification, claims, and the above-mentioned drawings of the present invention, a plurality of operations appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order in which they appear herein or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent a sequence, and do not limit that "first" and "second" are of different types.

[0062] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present invention.

[0063] Figure 1 It is a schematic flowchart of a human-machine dissonance detection method based on a single-breath waveform provided by an embodiment of the present invention. Specifically, the method includes the following steps:

[0064] 101: Obtain the single-breath waveform data of the person to be tested; the single-breath waveform data includes flow rate waveform data and airway pressure waveform data;

[0065] In one embodiment, the single-breath waveform data is at least flow rate waveform data and airway pressure waveform data, and information such as ventilator setting parameters and patient disease status may be added during specific use;

[0066] In one embodiment, the detection method further includes: preprocessing the single-breath waveform data of the person to be tested to obtain the preprocessed single-breath waveform data; the preprocessing method is the conventional preprocessing of machine learning / deep learning training data, including processes such as data cleaning, data filling, and data standardization, and unifying the data format.

[0067] In one embodiment, the single-breath waveform data of the person to be tested includes: single-breath waveform data segmented based on the flow rate waveform;

[0068] The method or steps for obtaining the single-breath waveform data segmented based on the flow rate waveform include:

[0069] Obtain the respiratory waveform data of the subject to be measured; the respiratory waveform data includes: respiratory flow rate data based on the respiratory marking time; the respiratory flow rate data based on the respiratory marking time is flow rate waveform data with the abscissa being Time (s) and the ordinate being Flow (L / min); the respiratory marking time is Time (s), and Time is set according to the sampling frequency hz of the ventilator and the serial number number of the data, and the calculation formula is: Time = (number - 1) / hz; number is the serial number of the sampling point.

[0070] Perform normalization processing on the respiratory waveform data of the subject to be measured to obtain the normalized respiratory waveform data; the normalized respiratory waveform data includes: from the flow rate waveform data based on the respiratory marking time, find the points where the flow rate crosses zero and the derivative is greater than zero, and record the time indices of these points respectively, to obtain the normalized respiratory waveform data; optionally, the preprocessing is to traverse the respiratory waveform data in sequence, at least once.

[0071] Slice the normalized respiratory waveform data into respiratory data with multiple respiratory cycles to obtain the normalized respiratory data with multiple respiratory cycles; the slicing is to find all the points where the flow rate crosses zero (one point is less than or equal to 0 and the other point is greater than or equal to 0) and the derivative is greater than zero; this point is as Figure 5 shown by the circle; perform feature extraction on the normalized respiratory data with multiple respiratory cycles to obtain the feature-extracted respiratory data with multiple respiratory cycles as feature data; input the feature data of the subject into the classification model to obtain the classification result of the feature data.

[0072] The normalized respiratory data with multiple respiratory cycles includes: respectively slice the adjacent points to form waveforms corresponding to each breath, to obtain the respiratory data of the multiple respiratory cycles.

[0073] Each breath is a single breath, and the definition of a single breath: a complete inhalation phase plus a complete exhalation phase; the inspiratory flow rate (Flow) changing from negative to positive is recorded as the start of a breath; the end of the breath is the previous sampling point before the start of the next breath.

[0074] Optionally, respectively take the nth and (n + 1)th adjacent points, and define the interval between the nth and (n + 1)th points as a single respiratory cycle; where, 1 ≤ n < N - 1, and N is the total number of time indices of all the points.

[0075] The feature data of the breathing data with multiple breathing cycles after feature extraction includes: the breathing time interval of a single breath, the flow rate change of a single breath, the ratio of the exhalation duration of a single breath to the total breathing time of a single breath, and the rising slope of the exhalation waveform of a single breath.

[0076] Optionally, the breathing time interval of a single breath is: the time interval a between the time indices corresponding to the adjacent points, that is, the time interval between two adjacent circles, as shown by the arrow in Figure 6 ; the value range of a is 0s < a < 1s; preferably, 0s < a < 0.6s; the normal breathing rate is about 8 - 20 breaths per minute, and in pathological conditions, the patient's breathing rate can reach 30 - 60 breaths per minute. Therefore, the value range of the breathing time a is 0s < a < 1s, and the extreme case where the breathing rate > 100 times is extremely rare. Therefore, the preferred value range of a is 0s < a < 0.6s.

[0077] Optionally, the flow rate change of a single breath is: the flow rate change b between the time indices corresponding to the adjacent points; as shown by the arrow in Figure 7 ; the value range of b is 0L / min < b < 20L / min; preferably, 0L / min < b < 10L / min; the inspiratory flow rate of a normal person is 40 - 60L / min, and that of a child is 5 - 10L / min. Therefore, the value range of the flow rate change b during a single breath is 0L / min < b < 20L / min. Considering that some patients have insufficient inspiratory strength, the preferred value range of b is 0L / min < b < 10L / min.

[0078] Optionally, the ratio of the exhalation duration (the time when the flow rate is negative) of a single breath to the total breathing time of a single breath (the time between two adjacent circles) is: the ratio c% of the exhalation duration between the time indices corresponding to the adjacent points to the total breathing time between the time indices corresponding to the adjacent points; the value range of c% is 20% - 80%; preferably, c is 30% - 50%; according to the definition of double triggering, the exhalation time between two inspiratory cycles is less than half of the average inspiratory time. The value range of c is 20% - 80%, and the preferred value range of c is 30% - 50%. This setting can well avoid the influence of double or multiple triggering on the results; in clinical tachypnea and abnormal breathing, when breathing twice in a short time, the accuracy of calculating a single breath by this segmentation method is better; a judgment is made in the middle, and how to deal with it with the reference to effective assistance in case of abnormal situations.

[0079] Optionally, the rising slope of the exhalation waveform of a single breath is: the rising slope d of the exhalation waveform between the time indices corresponding to the adjacent points; as shown by the arrow in Figure 8as shown in the square box; within the first 100 ms of inhalation, the value range of d is 10 - 50 L / min / s; d is preferably 5 - 25 L / min / s; the flow rate trigger setting value is about 1 - 5 L / min. Within the first 100 ms of inhalation, the value range of d is about 10 - 50 L / min / s. Considering part of the trigger delay, the preferred range of d is 5 - 25 L / min / s.

[0080] The method or steps of inputting the characteristic data of the person to be tested into the classification model to obtain the classification result of the characteristic data of the person to be tested include:

[0081] Input the breathing time interval of the single breathing cycle of the person to be tested into the classification model, and judge whether the breathing time interval of the single breathing cycle of the person to be tested falls within the range of a; if the breathing time interval of the single breathing cycle of the person to be tested falls within the range of a, output yes, and enter the stage of inputting the flow rate change of the single breathing cycle of the person to be tested into the classification model; otherwise, output no and terminate the operation; a in this step is denoted as 1 in the appendix Figure 4 and is recorded as 1 in the appendix;

[0082] Input the flow rate change of the single breathing cycle of the person to be tested into the classification model, and judge whether the flow rate change of the single breathing cycle of the person to be tested falls within the range of b; if the flow rate change of the single breathing cycle of the person to be tested falls within the range of b, output yes, and enter the stage of inputting the ratio of the expiratory duration of the single breathing cycle of the person to be tested to the total breathing time of the single breathing cycle into the classification model; otherwise, output no and terminate the operation; b in this step is denoted as 2 in the appendix Figure 4 and is recorded as 2 in the appendix;

[0083] Input the ratio of the expiratory duration of the single breathing cycle of the person to be tested to the total breathing time of the single breathing cycle into the classification model, and judge whether the ratio of the expiratory duration of the single breathing cycle of the person to be tested to the total breathing time of the single breathing cycle falls within the range of c; if the ratio of the expiratory duration of the single breathing cycle of the person to be tested to the total breathing time of the single breathing cycle falls within the range of c, output yes, and enter the stage of inputting the rising slope of the expiratory waveform of the single breathing cycle of the person to be tested into the classification model; otherwise, output no and terminate the operation; c in this step is denoted as 3 in the appendix Figure 4 and is recorded as 3 in the appendix;

[0084] Input the rising slope of the exhalation waveform of the single respiratory cycle of the subject to be measured into the classification model, and determine whether the rising slope of the exhalation waveform of the single respiratory cycle of the subject to be measured falls within the range of d; in the case where the rising slope of the exhalation waveform of the single respiratory cycle of the subject to be measured falls within the range of d, output "yes", and output the classification result of the respiratory waveform data of the subject to be measured; otherwise, output "no" and terminate the operation. d in this step is recorded as 4 in the appendix Figure 4 as shown in

[0085] In one embodiment, the respiratory waveform data is respiratory waveform data with a continuous waveform signal.

[0086] In one embodiment, the respiratory waveform data further includes: esophageal pressure waveform data based on the respiratory marking time; the esophageal pressure waveform data is esophageal pressure waveform data with the abscissa being Time and the ordinate being the esophageal pressure data Pes (cmH2O).

[0087] 102: Input the single-breath waveform data of the subject to be measured into the machine learning detection model to obtain the type of human-machine imbalance of the single-breath waveform data of the subject to be measured.

[0088] Optionally, the determination method of the machine learning detection model includes: inputting pre-annotated single-breath waveform data, extracting features from the pre-annotated single-breath waveform data to obtain feature data after feature extraction, and using the feature data after feature extraction to construct a machine learning detection model to obtain a machine learning detection model. The pre-annotated single-breath waveform data is single-breath waveform data with manual annotation results.

[0089] The detection method further includes: obtaining or drawing FV loop and / or PV loop waveform image data according to the preprocessed single-breath waveform data, and outputting the FV loop and / or PV loop waveform image data; the single-breath waveform data further includes volume data obtained based on flow rate data.

[0090] Extract features from the FV loop and / or PV loop waveform image data to obtain the FV loop and / or PV loop waveform image data after feature extraction as feature data; input the FV loop and / or PV loop waveform image data after feature extraction into the machine learning detection model (deep learning detection model) to obtain the type of human-machine imbalance of the single-breath waveform data of the subject to be measured.

[0091] Optionally, the FV loop waveform image data is obtained based on flow rate data and volume data, the abscissa of the FV loop waveform image data is volume data V, and the ordinate is flow rate data F; the PV loop waveform image data is obtained based on airway pressure data and volume data, the abscissa of the PV loop waveform image data is volume data V, and the ordinate is airway pressure data P.

[0092] Optionally, the flow rate waveform data is flow rate waveform data based on the breathing marking time; the breathing marking time is Time, and Time is set according to the sampling frequency hz of the ventilator and the sequence number number of the data. The calculation formula is: Time = (number - 1) / hz.

[0093] Optionally, the flow rate waveform data is flow rate waveform data with Time as the abscissa and flow rate data as the ordinate;

[0094] Optionally, the airway pressure waveform data is airway pressure waveform data with Time as the abscissa and airway pressure data as the ordinate;

[0095] Optionally, the volume waveform data is volume waveform data with Time as the abscissa and volume data as the ordinate; the calculation formula for the volume data is:

[0096] In one embodiment, the breathing waveform data is breathing waveform data with a continuous waveform signal.

[0097] The determination method of the classification model includes:

[0098] Obtain the breathing waveform data of normal people;

[0099] Perform normalization processing on the breathing waveform data to obtain the normalized breathing waveform data; divide the normalized breathing waveform data into breathing data with multiple breathing cycles to obtain normalized breathing data with multiple breathing cycles;

[0100] Perform feature selection or feature extraction on the normalized breathing data with multiple breathing cycles to obtain the breathing data with multiple breathing cycles after feature selection or feature extraction as feature data;

[0101] Use the method of machine learning to perform feature extraction on the feature data to obtain the feature data after feature extraction, and use the feature data after feature extraction to construct a classification model to obtain the constructed classification model.

[0102] 103: Output the type of man-machine mismatch of the single-breath waveform data of the subject.

[0103] The type of man-machine mismatch of the single-breath waveform data of the subject includes: normal / abnormal or normal / abnormal subtype.

[0104] The detection method further includes determining a breathing abnormality detection result based on the output of the type of man-machine mismatch.

[0105] Figure 2 Yes Schematic diagram of the human-machine dissonance detection device based on single-port breathing waveform provided by an embodiment of the present invention. The device includes: a memory and a processor;

[0106] The memory is used to store program instructions;

[0107] The processor is used to call the program instructions. When the program instructions are executed, it is used to execute the above-mentioned human-machine dissonance detection method based on single-port breathing waveform.

[0108] Figure 3 Yes Schematic flowchart of the human-machine dissonance detection system based on single-port breathing waveform provided by an embodiment of the present invention, including:

[0109] An acquisition unit 301, configured to acquire single-port breathing waveform data of a person to be tested; the single-port breathing waveform data includes flow rate waveform data and airway pressure waveform data;

[0110] A first processing unit 302, configured to input the single-port breathing waveform data of the person to be tested into a machine learning detection model to obtain the type of human-machine dissonance of the single-port breathing waveform data of the person to be tested;

[0111] A classification unit 303, configured to output the type of human-machine dissonance of the single-port breathing waveform data of the person to be tested.

[0112] A computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the above-mentioned human-machine dissonance detection method based on single-port breathing waveform.

[0113] The verification result of this verification embodiment shows that assigning fixed weights to the indications can moderately improve the performance of this method compared to the default settings.

[0114] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0115] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be in an electrical, mechanical, or other forms.

[0116] The unit described as a separation component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0117] In addition, each functional unit in various embodiments of the present invention may be integrated in a processing unit, may exist separately as individual physical units, or two or more units may be integrated in one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0118] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and this program can be stored in a computer-readable storage medium. The storage medium may include: read-only memory (ROM, Read Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk, etc.

[0119] Those of ordinary skill in the art can understand that all or part of the steps in implementing the methods of the above embodiments can be completed by instructing relevant hardware through a program, and the said program can be stored in a computer-readable storage medium. The storage medium mentioned above may be a read-only memory, magnetic disk or optical disk, etc.

[0120] The above has introduced in detail a computer device provided by the present invention. For those of ordinary skill in the art, according to the idea of the embodiments of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation on the present invention.

Claims

1. A method for detecting human - machine dis - coordination based on single - breath waveform, comprising: Obtaining the single - breath waveform data of the person to be tested; The single - breath waveform data includes flow - rate waveform data, airway - pressure waveform data, and volume waveform data; Pre - processing the single - breath waveform data of the person to be tested to obtain pre - processed single - breath waveform data; The abscissa of the flow velocity waveform data is , and the ordinate is the flow velocity data; the abscissa of the airway pressure waveform data is , and the ordinate is the airway pressure data. The volume waveform data has the as the abscissa and volume data as the ordinate; Obtaining or plotting FV - loop and / or PV - loop waveform image data based on the pre - processed single - breath waveform data; extracting features from the FV - loop and / or PV - loop waveform image data, and inputting the feature - extracted FV - loop and / or PV - loop waveform image data into a machine - learning detection model to obtain the type of human - machine dis - coordination of the single - breath waveform data of the person to be tested; The FV - loop waveform image data is obtained based on flow - rate data and volume data. The abscissa of the FV - loop waveform image data is volume data V, and the ordinate is flow - rate data F; the PV - loop waveform image data is obtained based on airway - pressure data and volume data. The abscissa of the PV - loop waveform image data is volume data V, and the ordinate is airway - pressure data P.

2. The method for detecting human-machine maladjustment based on a single-port breathing waveform according to claim 1, wherein The determination method of the machine - learning detection model includes: inputting pre - labeled single - breath waveform data with manual annotation results, extracting features from the pre - labeled single - breath waveform data with manual annotation results to obtain feature - extracted feature data, and using the feature - extracted feature data to construct a machine - learning detection model to obtain the machine - learning detection model.

3. The method for detecting human-machine imbalance based on a single-port breathing waveform according to claim 1, wherein, The flow rate waveform data is flow rate waveform data based on the breathing marking time; the breathing marking time is Time, and Time is set according to the sampling frequency hz of the ventilator and the data sequence number number. The calculation formula is: .

4. The method for detecting human-machine maladjustment based on single-port breathing waveform according to claim 1, wherein The calculation formula for the volume waveform data is: Volume(t)= .

5. The method for detecting human-machine maladjustment based on a single-port breathing waveform according to claim 1, wherein The type of human - machine dis - coordination includes: normal / abnormal or normal / abnormal subtype.

6. The method for detecting human-machine discoordination based on a single-port breathing waveform according to claim 3, characterized in that, The single - breath waveform data of the person to be tested is: single - breath waveform data segmented based on the flow - rate waveform.

7. The method for detecting human-machine imbalance based on a single-port breathing waveform according to claim 6, characterized in that, The method or steps for obtaining the single - breath waveform data segmented based on the flow - rate waveform include: Obtaining the breath waveform data of the person to be tested; Finding the points where the flow - rate crosses zero and the derivative is greater than zero in the breath waveform data, and respectively recording the time indices of these points to obtain normalized breath waveform data; Segmenting the normalized breath waveform data into breath data with multiple breath cycles to obtain normalized breath data with multiple breath cycles; the segmentation is to find all points where the flow - rate crosses zero and the derivative is greater than zero; Extracting features from the normalized breath data with multiple breath cycles to obtain feature - extracted breath data with multiple breath cycles as feature data; inputting the feature data of the person to be tested into a classification model to obtain the classification result of the feature data; the feature - extracted breath data with multiple breath cycles as feature data includes: the breath time interval of a single breath, the flow - rate change of a single breath, the ratio of the exhalation duration of a single breath to the total breath time of a single breath, and the rising slope of the exhalation waveform of a single breath.

8. The method for detecting human-machine maladjustment based on a single-port breathing waveform according to claim 7, characterized in that, The pre - processing is to traverse the breath waveform data in order, at least once.

9. The method for detecting human-machine maladjustment based on a single-port breathing waveform according to claim 7, wherein, The normalized breath data with multiple breath cycles includes: respectively segmenting adjacent points to form waveforms corresponding to each breath, and obtaining the breath data of the multiple breath cycles.

10. The method for detecting human-machine imbalance based on single-port breathing waveform according to claim 9, characterized in that, The method for obtaining respiratory data of the respiratory cycle includes: respectively taking the nth and (n + 1)th adjacent points, and defining the interval between the nth and (n + 1)th points as a single respiratory cycle; where 1 ≤ n < N - 1, and N is the total number of time indices of all the points.

11. The method for detecting human-machine imbalance based on single-port breathing waveform according to claim 10, characterized in that, The respiratory time interval of a single breath is: the time interval a between the time indices corresponding to the adjacent points; the value range of a is 0s < a < 1s.

12. The method for detecting human-machine dissonance based on a single-port breathing waveform according to claim 11, wherein The flow rate change of a single breath is: the flow rate change b between the time indices corresponding to the adjacent points; the value range of b is 0 L / min < b < 20 L / min.

13. The method for detecting human-machine imbalance based on a single-port breathing waveform according to claim 12, characterized in that, The ratio of the exhalation duration of a single breath to the total respiratory time of a single breath is: the ratio c % of the exhalation duration between the time indices corresponding to the adjacent points to the total respiratory time between the time indices corresponding to the adjacent points; the value range of c % is 20% - 80%.

14. The method for detecting human-machine maladjustment based on a single-port breathing waveform according to claim 13, wherein The rising slope of the exhalation waveform of a single breath is: the rising slope d of the exhalation waveform between the time indices corresponding to the adjacent points; within the first 100 ms of the beginning of inhalation, the value range of d is 10 - 50 L / min / s.

15. The method for detecting human-machine imbalance based on single-port breathing waveform according to claim 14, characterized in that, The method or steps for inputting the characteristic data of the subject into the classification model to obtain the classification result of the characteristic data include: Input the respiratory time interval of a single respiratory cycle of the subject into the classification model, and determine whether the respiratory time interval of a single respiratory cycle of the subject falls within the range of a; if the respiratory time interval of a single respiratory cycle of the subject falls within the range of a, output yes, and enter the stage of inputting the flow rate change of a single respiratory cycle of the subject into the classification model; otherwise, output no and terminate the operation; Input the flow rate change of a single respiratory cycle of the subject into the classification model, and determine whether the flow rate change of a single respiratory cycle of the subject falls within the range of b; if the flow rate change of a single respiratory cycle of the subject falls within the range of b, output yes, and enter the stage of inputting the ratio of the exhalation duration of a single respiratory cycle of the subject to the total respiratory time of a single respiratory cycle into the classification model; otherwise, output no and terminate the operation; Input the ratio of the exhalation duration of a single respiratory cycle of the subject to the total respiratory time of a single respiratory cycle into the classification model, and determine whether the ratio of the exhalation duration of a single respiratory cycle of the subject to the total respiratory time of a single respiratory cycle falls within the range of c; if the ratio of the exhalation duration of a single respiratory cycle of the subject to the total respiratory time of a single respiratory cycle falls within the range of c, output yes, and enter the stage of inputting the rising slope of the exhalation waveform of a single respiratory cycle of the subject into the classification model; otherwise, output no and terminate the operation; Input the rising slope of the exhalation waveform of the single respiratory cycle of the subject to be measured into the classification model, and determine whether the rising slope of the exhalation waveform of the single respiratory cycle of the subject to be measured falls within the range of d; in the case where the rising slope of the exhalation waveform of the single respiratory cycle of the subject to be measured falls within the range of d, output "yes", and output the classification result of the respiratory waveform data of the subject to be measured; otherwise, output "no" and terminate the operation.

16. A human-machine dissonance detection device based on a single-port breathing waveform, the device comprising: A memory and a processor; The memory is used for storing program instructions; The processor is used for calling the program instructions, and when the program instructions are executed, it is used for executing the method for detecting human-machine maladjustment based on a single-port respiratory waveform according to any one of claims 1-15.

17. A system for detecting human-machine maladjustment based on a single-port respiratory waveform, comprising: An acquisition unit, configured to acquire the single-port respiratory waveform data of the subject to be measured; The single-port respiratory waveform data includes flow rate waveform data, airway pressure waveform data, and volume waveform data; A processing unit, configured to preprocess the single-port respiratory waveform data of the subject to be measured to obtain preprocessed single-port respiratory waveform data; The abscissa of the flow rate waveform data is , and the ordinate is the flow rate data; the abscissa of the airway pressure waveform data is , and the ordinate is the airway pressure data; The volume waveform data has the as the abscissa and volume data as the ordinate; A classification unit, configured to obtain or draw FV loop and / or PV loop waveform image data according to the preprocessed single-port respiratory waveform data; extract features from the FV loop and / or PV loop waveform image data, and input the FV loop and / or PV loop waveform image data after feature extraction into a machine learning detection model to obtain the type of human-machine maladjustment of the single-port respiratory waveform data of the subject to be measured; the FV loop waveform image data is obtained based on flow rate data and volume data, the abscissa of the FV loop waveform image data is volume data V, and the ordinate is flow rate data F; the PV loop waveform image data is obtained based on airway pressure data and volume data, the abscissa of the PV loop waveform image data is volume data V, and the ordinate is airway pressure data P.

18. A computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the method for detecting human-machine maladjustment based on a single-port respiratory waveform according to any one of claims 1-15 above.

Images