Breathing state evaluation model training method and system and breathing state evaluation method and system

Abstract

In the breathing state evaluation model training and breathing state evaluation method and system, breathing-related physiological parameter data is collected; the physiological parameter data comprises a blood oxygen saturation numerical value, a heart rate numerical value, a breathing carbon dioxide concentration numerical value and a breathing rate numerical value; interpreting the physiological parameter data, and giving a breathing state evaluation label corresponding to the physiological parameter data; the breathing state evaluation label comprises an IPI index; inputting the physiological parameter data and the breathing state evaluation label into an artificial intelligence training device, and training by the artificial intelligence training device to obtain a breathing state evaluation feature data set; the breathing state evaluation feature data set is matched with a corresponding artificial intelligence algorithm and is used for breathing state evaluation. The breathing state comprehensive evaluation can be directly carried out based on the complete waveform signal, and the IPI index is closer to reality.

Description

Technical Field

[0001] The present invention belongs to the field of biomedical engineering, and in particular to the field of comprehensive evaluation of respiratory status. Background Art

[0002] In the prior art, the evaluation of respiratory status is usually based on the ventilation parameters of a ventilator or an anesthesia machine. There is no respiratory status evaluation based entirely on physiological parameters.

[0003] In the prior art, there are many types of physiological parameters, including electromyographic signals, electrocardiographic signals, electroencephalographic signals, respiratory signals based on chest impedance measurement, respiratory signals based on gas concentration measurement, blood pressure signals, blood oxygen signals, etc. These signals and parameters provide rich parameter index information for clinical practice, and to some extent, they all represent some physiological states.

[0004] However, for clinicians, too many physiological parameter signals create an information burden. Especially in critical moments, attention is extremely limited. How to integrate the above-mentioned multiple physiological parameter information to provide a concise and effective evaluation information to facilitate clinical decision-making on whether to intervene in the patient's condition in a timely manner is a technical problem to be solved. Summary of the invention

[0005] The technical problem to be solved by the present invention is to avoid the above-mentioned deficiencies of the prior art, and propose a respiratory state assessment model training and a respiratory state assessment method and system; the IPI index can be given based on the blood oxygen saturation value, heart rate value, respiratory carbon dioxide concentration value, and respiratory rate value. In this application, a complete set of training methods and systems are established to obtain a respiratory state assessment feature data set, and based on the respiratory state assessment feature data set, a comprehensive respiratory state assessment can be performed to output the IPI index, providing efficient and comprehensive decision-making guidance information for clinicians.

[0006] The technical solution of the present invention to solve the above-mentioned problem is a respiratory state assessment model training method, which collects respiratory-related physiological parameter data; the above-mentioned physiological parameter data include blood oxygen saturation values, heart rate values, respiratory carbon dioxide concentration values, and respiratory rate values; interprets the above-mentioned physiological parameter data and provides a respiratory state assessment label corresponding to the above-mentioned physiological parameter data; the respiratory state assessment label includes an IPI index; the above-mentioned physiological parameter data and the respiratory state assessment label are input into an artificial intelligence training device, and a respiratory state assessment feature data set is obtained through training with the artificial intelligence training device; the above-mentioned respiratory state assessment feature data set is combined with a corresponding artificial intelligence algorithm for respiratory state assessment.

[0007] The above-mentioned respiratory state assessment model training method includes any one of the following technical features: TA1: The above-mentioned physiological parameter data include red light pulse wave signal data and infrared light pulse wave signal data; the blood oxygen saturation value and the heart rate value are calculated based on the red light pulse wave signal data and the infrared light pulse wave signal data; TA2: The above-mentioned physiological parameter data include carbon dioxide concentration waveform signal data; the respiratory carbon dioxide concentration value and the respiratory rate value are calculated based on the carbon dioxide concentration waveform signal data; TA3: The above-mentioned physiological parameter data include carbon dioxide concentration waveform signal data; the input carbon dioxide concentration waveform signal data is mainstream carbon dioxide concentration waveform signal data; TA4: The above-mentioned physiological parameter data include carbon dioxide concentration waveform signal data; the respiratory carbon dioxide concentration value and the respiratory rate value are calculated based on the carbon dioxide concentration waveform signal data; the above-mentioned respiratory carbon dioxide concentration value includes the end-tidal carbon dioxide concentration value and the end-inspiratory carbon dioxide concentration value.

[0008] In the above technical feature TA1, the steps of calculating and obtaining the blood oxygen saturation value according to the red light pulse wave signal data and the infrared light pulse wave signal data are as follows: calculating r x : Where M represents the number of samples between the minimum and maximum values ​​of the pulse rising edge of a pulse cycle in the red light pulse wave signal data. It is the integral value of the red light pulse wave at the rising edge of a pulse cycle; is the infrared light pulse wave integral value of the rising edge of a pulse cycle in the infrared light pulse wave signal data, is the effective value of the red light DC in one pulsation cycle in the red light pulse wave signal data, is the infrared light DC effective value within one pulse cycle in the infrared light pulse wave signal data, r x It is a ratio used to confirm the pulse oxygen value, which corresponds one to one with the blood oxygen saturation value SpO2(x).

[0009] The technical solution of the present invention to solve the above-mentioned problem can also be a respiratory state assessment model training system, including a data acquisition component and an artificial intelligence training device; the data acquisition component is used to collect respiratory-related physiological parameter data; the above-mentioned physiological parameter data include blood oxygen saturation values, heart rate values, respiratory carbon dioxide concentration values, and respiratory rate values; the above-mentioned physiological parameter data are interpreted to give a respiratory state assessment label corresponding to the above-mentioned physiological parameter data; the respiratory state assessment label includes an IPI index; the artificial intelligence training device performs artificial intelligence training based on the physiological parameter data and the respiratory state assessment label to obtain a respiratory state assessment feature data set; the above-mentioned respiratory state assessment feature data set is combined with a corresponding artificial intelligence algorithm for respiratory state assessment.

[0010] The technical solution of the present invention to solve the above-mentioned problem can also be a respiratory state assessment method, which obtains physiological parameter data; the physiological parameter data include blood oxygen saturation value, heart rate value, respiratory carbon dioxide concentration value, and respiratory rate value; the artificial intelligence calculation module cooperates with the respiratory state assessment feature data set to calculate the respiratory state according to the above-mentioned physiological parameter data; the above-mentioned respiratory state includes the IPI index; the respiratory state assessment feature data set is obtained through model training, and during model training, the physiological parameter data and the corresponding respiratory state assessment label are input into the artificial intelligence training device, and the respiratory state assessment feature data set is obtained through training; the respiratory state assessment label includes the IPI index.

[0011] The above-mentioned respiratory state assessment method includes any one of the following technical features: TB1: The above-mentioned physiological parameter data include red light pulse wave signal data and infrared light pulse wave signal data; the blood oxygen saturation value and the heart rate value are calculated based on the red light pulse wave signal data and the infrared light pulse wave signal data; TB2: The above-mentioned physiological parameter data include carbon dioxide concentration waveform signal data; the respiratory carbon dioxide concentration value and the respiratory rate value are calculated based on the carbon dioxide concentration waveform signal data; TB3: The above-mentioned physiological parameter data include carbon dioxide concentration waveform signal data; the input carbon dioxide concentration waveform signal data is mainstream carbon dioxide concentration waveform signal data; TB4: The above-mentioned physiological parameter data include carbon dioxide concentration waveform signal data; the respiratory carbon dioxide concentration value and the respiratory rate value are calculated based on the carbon dioxide concentration waveform signal data; the above-mentioned respiratory carbon dioxide concentration value includes the end-tidal carbon dioxide concentration value and the end-inspiratory carbon dioxide concentration value.

[0012] In the above technical feature TB1, the steps of calculating and obtaining the blood oxygen saturation value according to the red light pulse wave signal data and the infrared light pulse wave signal data are as follows: calculating r x : Where M represents the number of samples between the minimum and maximum values ​​of the pulse rising edge of a pulse cycle in the red light pulse wave signal data. It is the integral value of the red light pulse wave at the rising edge of a pulse cycle; is the infrared light pulse wave integral value of the rising edge of a pulse cycle in the infrared light pulse wave signal data, is the effective value of the red light DC in one pulsation cycle in the red light pulse wave signal data, is the infrared light DC effective value within one pulse cycle in the infrared light pulse wave signal data, r x It is a ratio used to confirm the pulse oxygen value, which corresponds one to one with the blood oxygen saturation value SpO2(x).

[0013] The technical solution of the present invention to solve the above-mentioned problem can also be a respiratory state assessment system, including a data acquisition component and a data analysis module; the data acquisition component is used to collect respiratory-related physiological parameter data; the above-mentioned physiological parameter data include blood oxygen saturation values, heart rate values, respiratory carbon dioxide concentration values, and respiratory rate values; the data analysis module includes an artificial intelligence calculation module and a respiratory state assessment feature data set, the artificial intelligence calculation module cooperates with the respiratory state assessment feature data set to calculate the respiratory state according to the above-mentioned physiological parameter data, and the above-mentioned respiratory state includes the IPI index; the respiratory state assessment feature data set is obtained through model training, and during model training, the physiological parameter data and the corresponding respiratory state assessment label are input into the artificial intelligence training device, and the respiratory state assessment feature data set is obtained through training; the respiratory state assessment label includes the IPI index.

[0014] The technical solution of the present invention to solve the above problem may also be an electronic device, including a processor and a memory storing a computer program, and when the processor executes the computer program, the above method steps are implemented.

[0015] The technical solution of the present invention to solve the above problem may also be a computer-readable storage medium, including a computer program, and when the above computer program is executed by a processor, the above method steps are implemented.

[0016] Compared with the prior art, the beneficial effects of the present invention are: the acquired red light pulse wave signal data, infrared light pulse wave signal data, and carbon dioxide concentration signal are input into the model, the model is trained, and a training data set under the set IPI index is obtained; the influence of different blood oxygen saturation values ​​SpO2, end-tidal CO2 concentration EtCO2, respiratory rate RR, and pulse rate PR parameter calculation methods on the IPI index is avoided, and the respiratory state evaluation model and respiratory state evaluation feature data set obtained based on the above method are more accurate. More original signal information and performance model training are provided, rather than just model training based on the various parameters obtained at a single moment, so that the model can be closer to reality.

[0017] Compared with the prior art, the beneficial effects of the present invention are: on the basis of waveform input, pulse rate and blood oxygen saturation value are added as input parameters, which further enriches the input of model evaluation and makes the model and training data set closer to reality.

[0018] Compared with the prior art, the beneficial effect of the present invention is that the blood oxygen saturation value SpO2 is calculated by using the ratio of the red light pulse wave signal data and the infrared light pulse wave signal data, so that the calculation of the blood oxygen saturation value SpO2 is more accurate.

[0019] Compared with the prior art, the beneficial effects of the present invention are: on the basis of waveform input, the end-tidal carbon dioxide parameters including the EtCO2 value, the end-inspiratory CO2 concentration value, i.e., the InsCO2 value and the respiratory rate are added as input parameters, which further enriches the input of the model evaluation and makes the model and the training data set closer to reality.

[0020] Compared with the prior art, the beneficial effect of the present invention is that the infrared carbon dioxide concentration signal and the carbon dioxide reference signal are used to calculate the EtCO2 value and the InsCO2 value, making the calculation of the EtCO2 value and the InsCO2 value more accurate.

[0021] Compared with the prior art, the beneficial effects of the present invention are: differentiating the infrared carbon dioxide concentration signal to obtain the rising edge and the falling edge, locating the peak and valley value position by confirming the rising edge through local search, and the infrared carbon dioxide concentration signal value corresponding to the position is the EtCO2 value and the InsCO2 value; and calculating the respiratory rate according to the peak and valley value position. The calculated EtCO2 value, InsCO2 value and respiratory rate are more accurate.

[0022] Compared with the prior art, the beneficial effect of the present invention is that comprehensive evaluation of respiratory status is performed based on the model obtained through the above training, making the comprehensive evaluation of respiratory status closer to reality.

[0023] Compared with the prior art, the beneficial effect of the present invention is that the above-mentioned training models can include multiple models, are not limited to specific models, and can also integrate multiple models to make the comprehensive evaluation of respiratory status closer to reality. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figures 1 to 4 This is a schematic diagram of the respiratory state assessment model training method. Figures 1 to 4 ;

[0025] Figures 5 and 6 This is a schematic diagram of the respiratory status assessment method. Figures 1 to 2 ;

[0026] Figure 7 It is a schematic diagram of a pulse oximeter SPO2 monitoring device;

[0027] Figure 8 It is a schematic diagram of red light pulse wave signal data and infrared light pulse wave signal data obtained by a pulse oxygen SPO2 monitoring device;

[0028] Fig. 9 It is a schematic diagram of a respiratory carbon dioxide EtCO2 monitoring device;

[0029] Fig.10It is a schematic diagram of the infrared carbon dioxide concentration signal and the carbon dioxide reference signal obtained by the respiratory carbon dioxide EtCO2 monitoring device;

[0030] Fig.11 is based on Fig.10 Schematic diagram of carbon dioxide concentration signal obtained by infrared carbon dioxide concentration signal and carbon dioxide reference signal;

[0031] Fig.12 It is a flowchart of the training method of the respiratory state assessment model;

[0032] Fig.13 It is a flowchart of the respiratory status assessment method;

[0033] Fig.14 yes Fig.13 Schematic diagram of the fuzzy logic model used in the respiratory state assessment method. DETAILED DESCRIPTION

[0034] The embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings.

[0035] The IPI (integrated pulmonary index) in the respiratory status assessment of this application is a comprehensive pulmonary index, which is specifically an integration of the four respiratory function monitoring indicators of blood oxygen saturation value SpO2, end-tidal CO2 concentration EtCO2, respiratory rate RR, and pulse rate PR, and is simplified to IPI. The IPI index uses values ​​1 to 10 to represent the different respiratory states of the patient, 1 represents severe respiratory insufficiency, and 10 represents the best respiratory state.

[0036] In the existing technology, a single pulse oxygen saturation, pulse rate, end-tidal CO2, respiratory rate and respiratory status detection are obtained, and the parameters themselves are already widely used in the domestic medical system. Each parameter represents the physiological significance of its own dimension, and it is impossible to make a holistic characterization conclusion of the comprehensive state of the lungs. Based on the values ​​of each parameter at a single moment, the correlation is not high, and the continuity of the signal is not high.

[0037] Each physiological parameter represents the state of different dimensions of human respiratory function, and each physiological parameter gives its own signal, parameter and alarm setting; and there is no correlation calculation or prompt between the parameters. It is difficult for doctors to extract useful information from a large amount of measurement data to judge the comprehensive state of the patient's lungs. It has high requirements on the doctor's professional level, information analysis ability, and time and energy for long-term observation and analysis of data. Under the current status of medical equipment monitoring, if doctors want to obtain the patient's respiratory status, they must comprehensively analyze the parameter data of multiple respiratory-related aspects to extract the patient's respiratory status. Facing multi-dimensional data analysis, it is undoubtedly a huge test for doctors' professionalism, concentration, and energy for long-term monitoring.

[0038] Studies have shown that when doctors observe more than three parameters at the same time, it becomes difficult to extract useful signals from the three parameters. Analyzing the patient's comprehensive lung status with more parameters requires extremely high professionalism and clinical experience of the doctor. If there can be further quantitative parameter calculations and the multi-dimensional respiratory-related parameters can be comprehensively quantified into an index that characterizes the comprehensive lung status, it will be conducive to the clinical application and promotion of parameters such as pulse oxygen, end-tidal CO2 and respiratory detection in the detection of comprehensive lung status. At the same time, it also reduces the requirements for the professional level of doctors and the occurrence of patient accidents.

[0039] In order to better conduct a comprehensive assessment of respiratory status, in this application, a multi-parameter dimensional analysis is performed based on four related parameters: blood oxygen saturation value SpO2, end-tidal CO2 concentration EtCO2, respiratory rate RR, and pulse rate PR, to make a comprehensive respiratory assessment. The above multiple parameters are integrated into a standard comprehensive lung function status parameter to conduct a comprehensive assessment of the respiratory status.

[0040] The present invention will realize the calculation of quantitative index for comprehensive lung status through the above-mentioned multi-parameter monitoring and parameter algorithm fusion, and then realize the evaluation of comprehensive lung status, so as to solve the above-mentioned problems and provide reliable applications.

[0041] like Figure 1 and Figure 2 In an embodiment of a respiratory state assessment model training method, respiratory-related physiological parameter data are collected; the physiological parameter data include blood oxygen saturation values, heart rate values, respiratory carbon dioxide concentration values, and respiratory rate values; the above physiological parameter data are interpreted to provide respiratory state assessment labels corresponding to the above physiological parameter data; the respiratory state assessment labels include IPI indexes; the above physiological parameter data and respiratory state assessment labels are input into an artificial intelligence training device, and a respiratory state assessment feature data set is obtained after training with the artificial intelligence training device; the above respiratory state assessment feature data set is used in conjunction with a corresponding artificial intelligence algorithm for respiratory state assessment. Fig.12 The artificial intelligence training device includes multiple different artificial intelligence data models, and different models have different weights. During training, the weights of different models can be set, and the weights of different models can be adjusted according to the situation.

[0042] like Figure 3 In an embodiment of a respiratory state assessment model training method, the above-mentioned physiological parameter data includes red light pulse wave signal data and infrared light pulse wave signal data; the blood oxygen saturation value and the heart rate value are calculated based on the red light pulse wave signal data and the infrared light pulse wave signal data.

[0043] like Figure 4In an embodiment of a respiratory state assessment model training method, the above-mentioned physiological parameter data includes carbon dioxide concentration waveform signal data; the respiratory carbon dioxide concentration value and the respiratory rate value are calculated based on the carbon dioxide concentration waveform signal data; the above-mentioned respiratory carbon dioxide concentration value includes the end-tidal carbon dioxide concentration value and the end-inspiratory carbon dioxide concentration value.

[0044] TA3: The above physiological parameter data include carbon dioxide concentration waveform signal data; the input carbon dioxide concentration waveform signal data is mainstream carbon dioxide concentration waveform signal data.

[0045] In an embodiment of a respiratory state assessment model training method, the steps of calculating and obtaining the blood oxygen saturation value according to the red light pulse wave signal data and the infrared light pulse wave signal data are as follows: calculating r x :

[0046] Where M represents the number of samples between the minimum and maximum values ​​of the pulse rising edge of a pulse cycle in the red light pulse wave signal data. It is the integral value of the red light pulse wave at the rising edge of a pulse cycle; is the infrared light pulse wave integral value of the rising edge of a pulse cycle in the infrared light pulse wave signal data, is the effective value of the red light DC in one pulsation cycle in the red light pulse wave signal data, is the infrared light DC effective value within one pulse cycle in the infrared light pulse wave signal data, r x It is a ratio used to confirm the pulse oxygen value, which corresponds one to one with the blood oxygen saturation value SpO2(x).

[0047] like Figure 1 and Figure 2 In an embodiment of a respiratory state assessment model training system, the system includes a data acquisition component and an artificial intelligence training device; the data acquisition component is used to collect respiratory-related physiological parameter data; the above-mentioned physiological parameter data include blood oxygen saturation values, heart rate values, respiratory carbon dioxide concentration values, and respiratory rate values; the above-mentioned physiological parameter data are interpreted manually or by machine to give a respiratory state assessment label corresponding to the above-mentioned physiological parameter data; the respiratory state assessment label includes an IPI index; the artificial intelligence training device performs artificial intelligence training based on the physiological parameter data and the respiratory state assessment label to obtain a respiratory state assessment feature data set; the above-mentioned respiratory state assessment feature data set is used in conjunction with a corresponding artificial intelligence algorithm for respiratory state assessment.

[0048] like Figure 5In an embodiment of a respiratory state assessment method, physiological parameter data is obtained; the physiological parameter data includes blood oxygen saturation value, heart rate value, respiratory carbon dioxide concentration value, and respiratory rate value; an artificial intelligence calculation module cooperates with a respiratory state assessment feature data set to calculate and obtain the respiratory state according to the above physiological parameter data; the above respiratory state includes an IPI index. The respiratory state assessment feature data set is obtained through model training. During model training, the physiological parameter data and the corresponding respiratory state assessment label are input into an artificial intelligence training device, and the respiratory state assessment feature data set is obtained through training; the respiratory state assessment label includes an IPI index.

[0049] like Figure 6 In an embodiment of a respiratory state assessment method, the physiological parameter data includes red light pulse wave signal data and infrared light pulse wave signal data; blood oxygen saturation value and heart rate value are calculated based on the red light pulse wave signal data and infrared light pulse wave signal data;

[0050] like Figure 6 In an embodiment of a respiratory state assessment method, the above-mentioned physiological parameter data includes carbon dioxide concentration waveform signal data; the respiratory carbon dioxide concentration value and the respiratory rate value are calculated based on the carbon dioxide concentration waveform signal data; the above-mentioned respiratory carbon dioxide concentration value includes the end-tidal carbon dioxide concentration value and the end-inspiratory carbon dioxide concentration value.

[0051] In an embodiment of a respiratory status assessment method, the steps of calculating and obtaining the blood oxygen saturation value according to the red light pulse wave signal data and the infrared light pulse wave signal data are as follows: calculating r x :

[0052] Where M represents the number of samples between the minimum and maximum values ​​of the pulse rising edge of a pulse cycle in the red light pulse wave signal data. It is the integral value of the red light pulse wave at the rising edge of a pulse cycle; is the infrared light pulse wave integral value of the rising edge of a pulse cycle in the infrared light pulse wave signal data, is the effective value of the red light DC in one pulsation cycle in the red light pulse wave signal data, is the infrared light DC effective value within one pulse cycle in the infrared light pulse wave signal data, r x It is a ratio used to confirm the pulse oxygen value, which corresponds one to one with the blood oxygen saturation value SpO2(x).

[0053] The design is aimed at the integral calculation and period recognition of pulse oxygen signal. First, it enhances the period feature recognition of pulse signal; second, it calculates the AC-DC ratio based on the above period calculation to enhance the calculation of signal characteristic parameters.

[0054] like Figure 5and Figure 6 In an embodiment of a respiratory state assessment system, the system includes a data acquisition component and a data analysis module; the data acquisition component is used to collect respiratory-related physiological parameter data; the above-mentioned physiological parameter data include blood oxygen saturation values, heart rate values, respiratory carbon dioxide concentration values, and respiratory rate values; the data analysis module includes an artificial intelligence calculation module and a respiratory state assessment feature data set, the artificial intelligence calculation module cooperates with the respiratory state assessment feature data set to calculate the respiratory state according to the above-mentioned physiological parameter data, and the above-mentioned respiratory state includes the IPI index; the respiratory state assessment feature data set is obtained through model training, and during model training, the physiological parameter data and the corresponding respiratory state assessment label are input into the artificial intelligence training device, and the respiratory state assessment feature data set is obtained through training; the respiratory state assessment label includes the IPI index.

[0055] An electronic device includes a processor and a memory storing a computer program. When the processor executes the computer program, the method steps are implemented.

[0056] A computer-readable storage medium includes a computer program. When the computer program is executed by a processor, the method steps are implemented.

[0057] The IPI index is usually calculated based on a certain model. Different models will output different IPI indexes. The model used for training is consistent with the model used for testing, which can maintain the consistency of the training model and the calculation during use.

[0058] The accuracy of the IPI index is affected by the accuracy of the calculation of the parameters of blood oxygen saturation SpO2, end-tidal CO2 concentration EtCO2, respiratory rate RR, and pulse rate PR. Under the same or similar respiratory state, different calculation methods may obtain different blood oxygen saturation values ​​SpO2, end-tidal CO2 concentration EtCO2, respiratory rate RR, and pulse rate PR, and thus the output IPI index may also be different.

[0059] The respiratory state assessment model training method based on red light pulse wave signal data, infrared light pulse wave signal data, and carbon dioxide concentration signal can directly perform comprehensive respiratory state assessment based on the complete waveform signal, avoiding the influence of different blood oxygen saturation values ​​SpO2, end-tidal CO2 concentration EtCO2, respiratory rate RR, and pulse rate PR parameter calculation methods on the IPI index. The respiratory state assessment model and respiratory state assessment feature data set obtained based on the above method are more accurate; the consistency of the obtained IPI index is better.

[0060] like Figure 7 As shown, the pulse oxygen SPO2 monitoring device can monitor the pulse oxygen SPO2 and obtain the pulse oxygen SPO2 parameters. The pulse oxygen SPO2 monitoring device usually uses a dual-wavelength spectral method and performance measurement.

[0061] like Figure 8 , a schematic diagram of the red light pulse wave signal data and infrared light pulse wave signal data obtained by the pulse oxygen SPO2 monitoring device. The pulse oxygen SPO2 monitoring device will output the pulse rate and blood oxygen saturation value. The pulse oxygen monitoring system based on dual-wavelength spectral absorption is also a prior art. Since the calculation and application of pulse oxygen parameters reflect the reaction of oxygen circulation at the end of the human circulation after breathing, there is no timely coupling application with the end-of-respiration monitoring parameters. When the pulse oxygen detection is lower than the alarm limit, the human body has already suffered a hypoxic accident and cannot be avoided in advance. It is difficult to perform a complete assessment of the respiratory state using a single pulse oxygen SPO2. However, pulse oxygen monitoring parameters play a very critical role in the evaluation of the comprehensive pulmonary state and circulation state.

[0062] like Fig. 9 , respiratory carbon dioxide EtCO2 monitoring device; can monitor respiratory carbon dioxide EtCO2 and obtain respiratory carbon dioxide EtCO2 parameters. The respiratory carbon dioxide EtCO2 monitoring device is usually carried out by dual-wavelength mid- and far-infrared spectroscopy, in which one wavelength is a band with a characteristic absorption peak of CO2, and the other band is a band without a characteristic absorption peak of CO2, which is used as a reference band. Such a comparison setting of two bands can improve the accuracy of CO2 concentration detection and avoid the influence of CO2 concentration fluctuations or other fluctuations in the environment on the measurement. The respiratory carbon dioxide EtCO2 monitoring device will output end-tidal carbon dioxide parameters including EtCO2 value, InsCO2 value and respiratory rate.

[0063] like Fig. 9 The above physiological parameter data includes carbon dioxide concentration waveform signal data; the input carbon dioxide concentration waveform signal data is mainstream carbon dioxide concentration waveform signal data. Mainstream carbon dioxide concentration waveform signal data is the carbon dioxide concentration waveform obtained in the main airway of breathing during the breathing process, so it is more closely related to the respiratory state and is used for the calculation of the IPI index and model training, so that subsequent models, data sets and calculations can be more accurate.

[0064] like Fig.10 and Fig.11 , a schematic diagram of the infrared carbon dioxide concentration signal obtained by the mainstream respiratory carbon dioxide EtCO2 monitoring device. The end-tidal carbon dioxide monitoring system based on the mid- and far-infrared spectrum is also an existing technology. The calculation and application of the end-tidal carbon dioxide parameter reflects the oxygen and carbon dioxide exchange characteristics in the lungs during breathing. It is not timely associated with the pulse oxygen calculation parameters for evaluation, and it is impossible to characterize whether the current gas exchange situation will cause hypoxia or carbon dioxide accumulation in the human body.

[0065] The respiratory carbon dioxide concentration monitoring parameter also plays an important role in the evaluation of ventilation function and respiratory gas exchange. Its response time is faster than that of pulse oxygen monitoring. In recent years, due to the advancement of algorithms, especially the development and application of artificial intelligence technology, and the basic research on clinical application needs and correlation, the quantitative evaluation of the relevant respiratory lung comprehensive status has been supported. However, the quantitative evaluation of the respiratory lung comprehensive status is usually based on the calculated blood oxygen saturation value SpO2, end-tidal CO2 concentration (EtCO2), respiratory rate RR, pulse rate PR and other single-moment parameters.

[0066] In the present application, the above-mentioned respiratory carbon dioxide concentration value includes the end-tidal carbon dioxide concentration value EtCO2 and the end-inspiratory carbon dioxide concentration value InsCO2. It is not only based on the end-tidal EtCO2 value, but also based on the end-inspiratory InsCO2 value; it provides more respiratory carbon dioxide data for the IPI index, and the abnormality of the end-inspiratory InsCO2 also indicates that the possibility of abnormal respiratory state increases.

[0067] In this application, the respiratory state assessment model training method based on red light pulse wave signal data and infrared light pulse wave signal data, as well as carbon dioxide concentration signal, can directly perform a comprehensive assessment of the respiratory state based on a complete waveform signal. Based on its lower-level signal quantity, a comprehensive assessment of the respiratory state is directly performed, so that the assessment is based on a more complete respiratory process and pulse process, so the assessment result is more consistent and closer to the real state. Not only based on single-time parameters such as blood oxygen saturation value SpO2, end-tidal CO2 concentration EtCO2, respiratory rate RR, and pulse rate PR, but also on this basis, the single-time parameters such as blood oxygen saturation value SpO2, end-tidal CO2 concentration EtCO2, respiratory rate RR, and pulse rate PR can be integrated to make the assessment result closer to the real state. The entire monitored respiratory lung comprehensive state and parameters are quantitatively evaluated, and an independent quantitative monitoring index of the comprehensive lung state is given to improve ease of use.

[0068] In this application, pulse oximetry monitoring, end-tidal carbon dioxide monitoring and comprehensive calculation of intelligent technology are combined, and through the improvement of respiratory gas monitoring and pulse oximetry monitoring, the measurement reliability of respiratory signals and pulse signals is enhanced, and a comprehensive calculation method is adopted to realize reliable calculation of IPI parameters, so as to achieve a more accurate quantitative evaluation of the comprehensive lung status.

[0069] like Fig.12As shown, a respiratory state assessment model training method and an embodiment of the respiratory state assessment method include the following steps: collecting raw data, the raw data includes respiratory-related physiological parameter data and corresponding IPI index; the physiological parameter data includes blood oxygen saturation value SPO2, heart rate value PR, end-tidal carbon dioxide concentration value EtCO2, and respiratory rate value RR; inputting the data into three different data models for training, selecting the IPI index as any one of the set values, including any one of the IPI indexes 1 to 10; under the conditions of each set IPI index value, obtaining red light pulse wave signal data and infrared light pulse wave signal data for calculating pulse blood oxygen parameters; obtaining a carbon dioxide concentration signal for calculating end-tidal carbon dioxide parameters; inputting the obtained red light pulse wave signal data, infrared light pulse wave signal data, and carbon dioxide concentration signal into the model, performing model training, and obtaining a training data set under the set IPI index; traversing multiple set IPI indexes and repeating the above steps; obtaining a respiratory state assessment feature data set.

[0070] A comprehensive algorithm model is designed, which uses pulse oxygen monitoring parameters, end-tidal carbon dioxide monitoring parameters, and respiratory flow monitoring parameters to give an IPI index of 1-10 through comprehensive calculation and threshold recognition for the evaluation of respiratory status.

[0071] The pulse oximetry parameters including pulse rate and blood oxygen saturation value are obtained; the obtained red light pulse wave signal data, infrared light pulse wave signal data, pulse rate, blood oxygen saturation value and carbon dioxide concentration signal are input into the model, the model is trained, and a training data set under the set IPI index is obtained.

[0072] The end-tidal carbon dioxide parameters obtained include EtCO2 value, InsCO2 value and respiratory rate; Step A50: input the obtained red light pulse wave signal data, infrared light pulse wave signal data, carbon dioxide concentration signal, EtCO2 value, InsCO2 value and respiratory rate into the model, perform model training, and obtain the training data set under the set IPI index.

[0073] The carbon dioxide concentration signal includes an infrared carbon dioxide concentration signal; a carbon dioxide reference signal for calculating the carbon dioxide concentration is obtained; a signal for calculating the EtCO2 value, the InsCO2 value, the respiratory rate and the carbon dioxide concentration is obtained based on the infrared carbon dioxide concentration signal and the carbon dioxide reference signal; the dual-channel carrier signal of the infrared detector, i.e., the infrared carbon dioxide concentration signal and the carbon dioxide reference signal, is integrated and demodulated to obtain the carbon dioxide concentration value:

[0074] Where N represents the number of samples in a carrier modulation cycle, Represents the ith signal sampling value of a modulation cycle of the infrared carbon dioxide concentration signal, represents the i-th reference signal sampling value of a modulation cycle of the carbon dioxide reference signal, Indicates the i-th signal sampling value of a modulation cycle when the infrared carbon dioxide concentration signal is at zero point, RefWave 0，i Represents the i-th reference signal sampling value of a modulation cycle when the carbon dioxide reference signal is at zero point; R x It is used to confirm the ratio of carbon dioxide concentration, which corresponds to carbon dioxide concentration one by one. x is the carbon dioxide concentration signal.

[0075] Design an integral calculation and period recognition of carbon dioxide signal, and use the recognition of waveform characteristic inflection points to achieve quantitative calculation of EtCO2 and respiratory gas.

[0076] The infrared carbon dioxide concentration signal is differentiated to obtain the rising edge and the falling edge. The peak and valley value positions are located by confirming the rising edge through local search. The infrared carbon dioxide concentration signal values ​​corresponding to this position are the EtCO2 value and the InsCO2 value. The respiratory rate is calculated based on the peak and valley value positions.

[0077] like Fig.12 As shown, in an embodiment of a respiratory state assessment model training method and a respiratory state assessment method, the end-tidal carbon dioxide concentration value EtCO2 and the respiratory rate value RR are first input into a data model 1 to establish a corresponding relationship;

[0078] like Fig.12 In an embodiment of a respiratory state assessment model training method and a respiratory state assessment method, the IPI index is selected as any one of the set values, which can be a set of corresponding relationships given according to an expert evaluation table. The IPI index is a specific value, and the corresponding various original signals are obtained and recorded and simultaneously input into the system for model training.

[0079] like Fig.12 In an embodiment of a respiratory state assessment model training method and a respiratory state assessment method, the end-tidal CO2 concentration EtCO2 and the respiratory rate RR can be trained based on data model 1 to obtain a data model function f1; the variables of function f1 are the carbon dioxide concentration signal, the concentration EtCO2, and the respiratory rate RR; the weight of function f1 is W1.

[0080] like Fig.12In an embodiment of a respiratory state assessment model training method and a respiratory state assessment method, based on function f1 and blood oxygen saturation value SpO2, data model 2 is used for training to obtain data model function f2; the variables of function f2 are function f1, red light pulse wave signal data, infrared light pulse wave signal data, and blood oxygen saturation value SpO2; the weight of function f1 is W2.

[0081] like Fig.12 , based on function f2 and pulse rate PR, data model three is trained to obtain data model function f3; the variables of function f3 are function f2 and pulse rate PR; the weight of function f3 is W3.

[0082] like Fig.12 In an embodiment of a respiratory state assessment model training method and a respiratory state assessment method, the above steps are repeated for different IPI indexes until a round of training is completed. In actual application, the corresponding red light pulse wave signal data and infrared light pulse wave signal data, carbon dioxide concentration signal or parameters, namely concentration EtCO2, respiratory rate RR, blood oxygen saturation value SpO2 and pulse rate PR are input to obtain the current IPI index.

[0083] like Fig.12 In an embodiment of a respiratory state assessment model training method and a respiratory state assessment method, training and assessment can be completed integratedly, and a single assessment output and input are used for regression error analysis to perform algorithm model optimization iteration.

[0084] like Fig.13 A respiratory state assessment method is provided, which obtains red light pulse wave signal data and infrared light pulse wave signal data for calculating pulse blood oxygen parameters; obtains a carbon dioxide concentration signal for calculating end-tidal carbon dioxide parameters; inputs the obtained red light pulse wave signal data, infrared light pulse wave signal data, and carbon dioxide concentration signal into a trained model to obtain an IPI index; the trained model includes a respiratory state assessment feature data set; the respiratory state assessment feature data set is obtained based on the respiratory state assessment model training method. Fig.13 The IPI in Chinese is the IPI index.

[0085] The pulse oximetry parameters including pulse rate and blood oxygen saturation value are obtained; the obtained red light pulse wave signal data, infrared light pulse wave signal data, pulse rate, blood oxygen saturation value and carbon dioxide concentration signal are input into the trained model to obtain the IPI index.

[0086] The end-tidal carbon dioxide parameters obtained include EtCO2 value, InsCO2 value and respiratory rate; Step B50: input the obtained red light pulse wave signal data, infrared light pulse wave signal data, carbon dioxide concentration signal, EtCO2 value, InsCO2 value and respiratory rate into the trained model to obtain the IPI index.

[0087] like Fig.13 and Fig.14 , designing a comprehensive algorithm model is a fuzzy logic reasoning model. Using arterial oxygen saturation SPO2, pulse rate PR, end-tidal CO2 concentration EtCO2, and respiratory rate RR, a comprehensive calculation is used to give an IPI parameter value with a calculation range of 1-10 for the evaluation of respiratory status. 1 means that the current comprehensive lung status is the most urgent and fatal, requiring immediate intervention by a doctor, and 10 means that the comprehensive lung status is healthy, using quantitative indicators to evaluate the comprehensive state of human breathing.

[0088] The establishment of the algorithm model relies on a number of clinical doctors (including respiratory doctors, anesthesiologists, and respiratory therapists) with professional knowledge of respiratory testing to conduct an evaluation questionnaire of 85 data sets based on 4 parameters (arterial oxygen saturation SPO2, pulse rate PR, end-tidal CO2 concentration EtCO2, and respiratory rate RR). The doctors evaluate the comprehensive lung function status IPI of the above 85 data sets with a score range of 1-10. Based on the score results and the corresponding four vital signs parameters, a fuzzy logic reasoning model is established based on the matlab Fuzzy Logic Toolbox as shown in formula (6):

[0089] If x1is A i1 and x2isA i2 ...x j is A ij and...x n is A in (6)

[0090] Then y i =f i (X) (7)

[0091] In formula (7), X = [x1...x j ...x n ] is the antecedent variable; A ij is x in rule i j Fuzzy set. Then we can deduce:

[0092] IPI i =f i (EtCO2, RR) (8)

[0093] like Fig.14 The fuzzy inference model of RR and EtCO2 is shown. The fuzzy inference model of EtCO2 and RR is initially established, and the IPI value derived is then Fig.12 The model training increases the influence of SPO2 on IPI measurement. Fig.12 The derived results were again Fig.14 The derivation of the human lung comprehensive status index IPI, which ranges from 1 to 10, is obtained through three calculations and four parameter inputs.

[0094] In the respiratory state assessment method and the model training method, the IPI index is selected as any one of the set values; under the numerical conditions of each set IPI index, the obtained red light pulse wave signal data, infrared light pulse wave signal data, and carbon dioxide concentration signal are input into the model to perform model training and obtain a training data set under the set IPI index; traverse multiple set IPI indexes and repeat the above steps; obtain a complete training data set for comprehensive assessment of respiratory state. Obtain red light pulse wave signal data and infrared light pulse wave signal data; obtain carbon dioxide concentration signal; input into the trained model to obtain the IPI index. The IPI index that can be directly used for comprehensive assessment of respiratory state based on the complete waveform signal is closer to reality.

[0095] The above are only embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A respiratory state assessment model training method, characterized in that: Collect data on breathing-related physiological parameters; The physiological parameter data include blood oxygen saturation value, heart rate value, respiratory carbon dioxide concentration value, and respiratory rate value; Interpreting the physiological parameter data and providing a respiratory state assessment label corresponding to the physiological parameter data; The respiratory status assessment label includes the IPI index; Inputting the physiological parameter data and the respiratory state assessment label into an artificial intelligence training device, and obtaining a respiratory state assessment feature data set through training by the artificial intelligence training device; The respiratory state assessment feature data set is combined with a corresponding artificial intelligence algorithm for use in respiratory state assessment.

2. The respiratory state assessment model training method according to claim 1, characterized in that: Include any of the following technical features: TA1: The physiological parameter data includes red light pulse wave signal data and infrared light pulse wave signal data; blood oxygen saturation value and heart rate value are calculated according to the red light pulse wave signal data and infrared light pulse wave signal data; TA2: The physiological parameter data includes carbon dioxide concentration waveform signal data; the respiratory carbon dioxide concentration value and the respiratory rate value are calculated according to the carbon dioxide concentration waveform signal data; TA3: The physiological parameter data includes carbon dioxide concentration waveform signal data; the input carbon dioxide concentration waveform signal data is mainstream carbon dioxide concentration waveform signal data; TA4: The physiological parameter data includes carbon dioxide concentration waveform signal data; the respiratory carbon dioxide concentration value and the respiratory rate value are calculated based on the carbon dioxide concentration waveform signal data; the respiratory carbon dioxide concentration value includes the end-tidal carbon dioxide concentration value and the end-inspiratory carbon dioxide concentration value.

3. The respiratory state assessment model training method according to claim 1, characterized in that: In the technical feature TA1, the steps of calculating and obtaining the blood oxygen saturation value according to the red light pulse wave signal data and the infrared light pulse wave signal data are as follows: calculating r x : Where M represents the number of samples between the minimum and maximum values ​​of the pulse rising edge of a pulse cycle in the red light pulse wave signal data. It is the integral value of the red light pulse wave at the rising edge of a pulse cycle; is the infrared light pulse wave integral value of the rising edge of a pulse cycle in the infrared light pulse wave signal data, is the effective value of the red light DC in one pulsation cycle in the red light pulse wave signal data, is the infrared light DC effective value within one pulse cycle in the infrared light pulse wave signal data, r x It is a ratio used to confirm the pulse oxygen value, which corresponds one to one with the blood oxygen saturation value SpO2(x).

4. A respiratory state assessment model training system, characterized in that: Includes data acquisition components and artificial intelligence training devices; The data acquisition component is used to collect respiratory-related physiological parameter data; The physiological parameter data include blood oxygen saturation value, heart rate value, respiratory carbon dioxide concentration value, and respiratory rate value; Interpreting the physiological parameter data and providing a respiratory state assessment label corresponding to the physiological parameter data; The respiratory status assessment label includes the IPI index; An artificial intelligence training device performs artificial intelligence training based on physiological parameter data and respiratory state assessment labels to obtain a respiratory state assessment feature data set; The respiratory state assessment feature data set is combined with a corresponding artificial intelligence algorithm for use in respiratory state assessment.

5. A method for evaluating respiratory status, characterized in that: Obtain physiological parameter data; physiological parameter data include blood oxygen saturation value, heart rate value, respiratory carbon dioxide concentration value, and respiratory rate value; The artificial intelligence calculation module cooperates with the respiratory state assessment feature data set to calculate the respiratory state according to the physiological parameter data; The respiratory status includes an IPI index; The respiratory state assessment feature data set is obtained through model training. During the model training, the physiological parameter data and the corresponding respiratory state assessment label are input into the artificial intelligence training device to obtain the respiratory state assessment feature data set through training; The respiratory status assessment tab includes the IPI index.

6. The respiratory status assessment method according to claim 5, characterized in that: Include any of the following technical features: TB1: The physiological parameter data includes red light pulse wave signal data and infrared light pulse wave signal data; blood oxygen saturation value and heart rate value are calculated according to the red light pulse wave signal data and infrared light pulse wave signal data; TB2: The physiological parameter data includes carbon dioxide concentration waveform signal data; the respiratory carbon dioxide concentration value and the respiratory rate value are calculated according to the carbon dioxide concentration waveform signal data; TB3: The physiological parameter data includes carbon dioxide concentration waveform signal data; the input carbon dioxide concentration waveform signal data is mainstream carbon dioxide concentration waveform signal data; TB4: The physiological parameter data includes carbon dioxide concentration waveform signal data; the respiratory carbon dioxide concentration value and the respiratory rate value are calculated according to the carbon dioxide concentration waveform signal data; the respiratory carbon dioxide concentration value includes the end-tidal carbon dioxide concentration value and the end-inspiratory carbon dioxide concentration value.

7. The respiratory status assessment method according to claim 6, characterized in that: In the technical feature TB1, the steps of calculating and obtaining the blood oxygen saturation value according to the red light pulse wave signal data and the infrared light pulse wave signal data are as follows: calculating r x : Where M represents the number of samples between the minimum and maximum values ​​of the pulse rising edge of a pulse cycle in the red light pulse wave signal data. It is the integral value of the red light pulse wave at the rising edge of a pulse cycle; is the infrared light pulse wave integral value of the rising edge of a pulse cycle in the infrared light pulse wave signal data, is the effective value of the red light DC in one pulsation cycle in the red light pulse wave signal data, is the infrared light DC effective value within one pulse cycle in the infrared light pulse wave signal data, r x It is a ratio used to confirm the pulse oxygen value, which corresponds one to one with the blood oxygen saturation value SpO2(x).

8. A respiratory status assessment system, characterized in that: It includes data acquisition components and data analysis modules; The data acquisition component is used to collect respiratory-related physiological parameter data; The physiological parameter data include blood oxygen saturation value, heart rate value, respiratory carbon dioxide concentration value, and respiratory rate value; The data analysis module includes an artificial intelligence calculation module and a respiratory state evaluation feature data set. The artificial intelligence calculation module cooperates with the respiratory state evaluation feature data set to calculate the respiratory state according to the physiological parameter data, and the respiratory state includes the IPI index; The respiratory state assessment feature data set is obtained through model training. During the model training, the physiological parameter data and the corresponding respiratory state assessment label are input into the artificial intelligence training device to obtain the respiratory state assessment feature data set through training; The respiratory status assessment tab includes the IPI index.

9. An electronic device comprising a processor and a memory storing a computer program, characterized in that: When the processor executes the computer program, the method steps described in any one of claims 1 to 3 and 5 to 7 are implemented.

10. A computer-readable storage medium comprising a computer program, characterized in that: When the computer program is executed by a processor, the method steps described in any one of claims 1 to 3 and 5 to 7 are implemented.

Images