WO2023102820A1 - Dispositif médical et procédé de reconnaissance d'état de ventilation - Google Patents

Dispositif médical et procédé de reconnaissance d'état de ventilation Download PDF

Info

Publication number
WO2023102820A1
WO2023102820A1 PCT/CN2021/136756 CN2021136756W WO2023102820A1 WO 2023102820 A1 WO2023102820 A1 WO 2023102820A1 CN 2021136756 W CN2021136756 W CN 2021136756W WO 2023102820 A1 WO2023102820 A1 WO 2023102820A1
Authority
WO
WIPO (PCT)
Prior art keywords
waveform data
target object
respiratory waveform
recognition
data
Prior art date
Application number
PCT/CN2021/136756
Other languages
English (en)
Chinese (zh)
Inventor
李响
刘京雷
黄志文
万聪颖
周小勇
Original Assignee
深圳迈瑞生物医疗电子股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳迈瑞生物医疗电子股份有限公司 filed Critical 深圳迈瑞生物医疗电子股份有限公司
Priority to CN202180103623.XA priority Critical patent/CN118159322A/zh
Priority to PCT/CN2021/136756 priority patent/WO2023102820A1/fr
Publication of WO2023102820A1 publication Critical patent/WO2023102820A1/fr

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61M—DEVICES FOR INTRODUCING MEDIA INTO, OR ONTO, THE BODY; DEVICES FOR TRANSDUCING BODY MEDIA OR FOR TAKING MEDIA FROM THE BODY; DEVICES FOR PRODUCING OR ENDING SLEEP OR STUPOR
    • A61M16/00—Devices for influencing the respiratory system of patients by gas treatment, e.g. mouth-to-mouth respiration; Tracheal tubes

Definitions

  • the present invention relates to the technical field of medical equipment, and more particularly relates to a medical equipment and a ventilation state identification method.
  • the inspiratory trigger can be set as a flow rate trigger, when the flow rate exceeds the trigger sensitivity (such as 2L/min), it will switch to the inspiratory phase, or when the pressure is triggered, the airway pressure is lower than the positive end expiratory pressure (Positive End Expiratory Pressure, PEEP) - trigger When the sensitivity (such as 2cmH2O) is changed to the gas phase.
  • the trigger sensitivity such as 2L/min
  • PEEP Positive End Expiratory Pressure
  • the expiratory trigger sensitivity is generally a percentage of the inspiratory peak flow rate, for example, when the inspiratory flow rate drops to 25% of the inspiratory peak flow rate, the ventilator switches to the expiratory phase. Since the inhalation or exhalation trigger sensitivity is set by the doctor based on experience, clinically there are cases where the sensitivity setting does not match the patient's needs, resulting in the occurrence of man-machine incoordination. For example, inspiratory trigger delay, invalid trigger, expiratory switching advance or switching delay, etc. In addition, according to the patient's own differences, water accumulation in the pipeline, excessive resistance, and low compliance may also occur during mechanical ventilation.
  • the present application provides a new method for identifying medical equipment and ventilation status.
  • the present invention has been made to solve at least one of the above-mentioned problems.
  • the present invention provides a medical device on the one hand, and the medical device includes:
  • a processor configured to execute the program instructions stored in the memory, so that the processor performs the following steps:
  • the respiratory waveform data includes basic respiratory waveform data and auxiliary waveform data
  • Recognition processing is performed on the respiratory waveform data of the target subject based on the trained recognition model to determine a recognition result of the target subject's current ventilation state, where the recognition result includes normal ventilation or abnormal ventilation events.
  • the respiratory waveform data includes basic respiratory waveform data and auxiliary waveform data
  • Recognition processing is performed on the respiratory waveform data of the target subject based on the trained recognition model to determine a recognition result of the target subject's current ventilation state, where the recognition result includes normal ventilation or abnormal ventilation events.
  • Another aspect of the present application also provides a medical device, the device comprising:
  • a processor configured to execute the program instructions stored in the memory, so that the processor performs the following steps:
  • the respiratory waveform data includes basic respiratory waveform data and/or auxiliary waveform data
  • the trained recognition model is obtained based on training by a machine learning method.
  • the trained recognition model includes a trained neural network model
  • the processor is further configured to: use deep learning to Method training to obtain the trained recognition model, the historical data information of the patient's respiratory waveform data includes the historical data information of the basic respiratory waveform data and/or the historical data information of the auxiliary waveform data.
  • the processor performs identification processing on the respiratory waveform data and the patient data based on a trained identification model, and determines the identification result of the current ventilation state of the target object, including:
  • the respiratory waveform data is input into the trained neural network model to obtain network output results;
  • the patient data of the target object is input into the trained neural network model, and the patient data of the target object is transformed
  • a parameter matrix is obtained through mapping; in the trained neural network model, information fusion is performed on the parameter matrix and the output result of the network to obtain the recognition result.
  • the processor performs identification processing on the respiratory waveform data and the patient data, and determines the identification result of the current ventilation state of the target object, including: preprocessing the respiratory waveform data of the target object , to obtain the processed respiratory waveform data, wherein the preprocessing includes normalization processing and/or filtering processing; based on the trained recognition model, the processed respiratory waveform data and the patient data of the target object Perform recognition processing to determine the recognition result of the current ventilation state of the target object.
  • the processor performs recognition processing on the respiratory waveform data and the patient data based on a trained recognition model, and determining the recognition result of the current ventilation state of the target object includes: judging the respiratory waveform Whether the quality of the data meets the preset condition; when it is determined that the quality of the respiratory waveform data meets the preset condition, input the respiratory waveform data and the patient data into the trained identification model for identification processing, and determine the target The recognition result of the current ventilation state of the subject; when it is determined that the quality of the respiratory waveform data does not meet the preset conditions, the respiratory waveform data and the patient data are not input into the trained recognition model for recognition processing, and it is determined The identification result of the current ventilation state of the target object.
  • the processor performs recognition processing on the respiratory waveform data and the patient data based on a trained recognition model, and determines the recognition result of the current ventilation state of the target object, including: After identifying and processing the respiratory waveform data of the target object and the patient data of the target object by the identification model, the expert system is used to identify the respiratory waveform data of the target object and the patient data to determine the target An identification result of the subject's current ventilation state.
  • the processor performing obtaining patient data of the target object includes: obtaining electronic medical record information of the target object; obtaining at least part of patient data of the target object based on the electronic medical record information; or, Acquiring the patient data of the target object input by the user through the human-computer interaction interface.
  • the patient data includes at least one of the following information: patient's medical record information, mechanical ventilation parameters, monitoring parameters, blood gas value parameters, ultrasound parameters, disease severity score, fluid balance information
  • patient The medical record information includes at least one of the following information: age, height, weight, BMI index, comorbidity information, genetic disease information, drug treatment information
  • mechanical ventilation parameters include at least one of the following information: tidal volume, Minute ventilation, respiratory rate, inspiratory time, PEEP, mean airway pressure, inspired oxygen concentration, pressure support level, maximum inspiratory pressure, maximum expiratory pressure, end-tidal carbon dioxide, lung compliance, airway resistance, mechanical energy, Work of breathing, driving pressure, peak pressure, plateau pressure
  • the monitoring parameters include at least one of the following parameters: blood oxygen, heart rate, blood pressure, mean arterial pressure, central venous pressure, temperature
  • blood gas value parameters include the following information At least one of: partial pressure of carbon dioxide in arterial blood, partial pressure of oxygen in arterial blood, pH of arterial blood gas
  • the disease severity score includes at least one
  • the processor is configured to: select a corresponding recognition mode according to an instruction input by the user; when the recognition mode is the first recognition mode, perform a search on the respiratory waveform data of the target object and the obtained recognition model based on the trained recognition model. Perform identification processing on the patient data of the target object, and determine the identification result of the current ventilation state of the target object;
  • the expert system is used to identify the respiratory waveform data of the target object and the patient data of the target object, and determine the identification result of the current ventilation state of the target object;
  • the recognition mode is the third recognition mode, use the trained recognition model and the expert system to perform recognition processing on the respiratory waveform data of the target object and the patient data of the target object, and determine that the target object is currently The recognition result of the ventilation state.
  • the processor is further configured to: when it is determined that there is an abnormal ventilation event, output prompt information.
  • the medical device further includes a display, and the display is used to acquire and display the prompt information.
  • the display is configured to display the prompt information in a preset display manner
  • the preset display manner includes one or more of the following manners: highlight display, additional symbol display, differentiated color, Differentiated shading and flickering.
  • the medical device includes a display, buttons are arranged on a display interface of the display, and the processor is configured to: when a user instruction input by a user through the buttons is acquired, control the display to display the prompt information.
  • the medical device further includes an alarm device, which is used to give an alarm prompt when the number or frequency of abnormal ventilation events within a preset time meets a preset condition.
  • an alarm device which is used to give an alarm prompt when the number or frequency of abnormal ventilation events within a preset time meets a preset condition.
  • the processor is further configured to: when the recognition result is an abnormal ventilation event, according to the recognition result, automatically adjust the setting parameters of the ventilator for ventilating the target object.
  • the processor is further configured to: when the recognition result is an abnormal ventilation event, according to the recognition result, output adjustment suggestion information of setting parameters of a ventilator for ventilating the target object.
  • the medical device includes a display, and the display is used to acquire and display the adjustment suggestion information.
  • the processor is further configured to: when the recognition result is an abnormal ventilation event, evaluate the degree of lung injury of the target subject according to the recognition result and the patient data of the target subject; and the medical
  • the device also includes a display for displaying the assessment of the extent of the lung injury.
  • the processor assessing the lung injury degree of the target subject according to the recognition result and the patient data of the target subject includes: based on the tidal volume in the recognition result and the patient data of the target subject The extent of lung injury in the subject is assessed.
  • the basic respiratory waveform data includes one or more of the following waveform data: pressure, flow velocity or volume waveform; the auxiliary waveform data includes one or more of the following waveforms: transpulmonary pressure, esophagus pressure, intragastric pressure, or diaphragmatic electromyography.
  • the abnormal ventilation event includes one or more of the following events: invalid trigger, double inhalation, false trigger, too small flow rate, reverse trigger, early switching, delayed switching, slow pressure rise, pressure rise Too fast, endogenous positive end-expiratory pressure ventilation, short inspiratory time, long inspiratory time, water accumulation in the circuit, excessive resistance, low compliance, inverse ratio ventilation.
  • the medical equipment is a ventilator, an anesthesia machine, a monitor or a central station.
  • the monitor when the medical device is a monitor, the monitor includes a communication interface, and the ventilation interface is configured to communicate with a ventilator for ventilating the target object.
  • Another aspect of the present application also provides a ventilation state recognition method, the method comprising:
  • the respiratory waveform data includes basic respiratory waveform data and/or auxiliary waveform data
  • the trained recognition model is obtained based on training by a machine learning method.
  • the trained identification model includes a trained neural network model, and the trained identification model uses a deep learning method based on the historical data information of the patient data of the previous patient and the respiratory waveform data of the previous patient.
  • the historical data information of the previous patient's respiratory waveform data obtained through training includes historical data information of basic respiratory waveform data and/or historical data information of auxiliary waveform data.
  • performing recognition processing on the respiratory waveform data of the target object and the patient data of the target object based on the trained recognition model, and determining the recognition result of the current ventilation state of the target object includes: judging the Whether the quality of the respiratory waveform data of the target object meets the preset condition; when it is determined that the quality of the respiratory waveform data of the target object meets the preset condition, input the respiratory waveform data of the target object and the patient data of the target object into the
  • the trained recognition model is used for recognition processing to determine the recognition result of the current ventilation state of the target object; when it is determined that the respiratory waveform data of the target object does not meet the preset conditions, the respiratory waveform data and the patient
  • the data is input to the trained recognition model for recognition processing, and the recognition result of the current ventilation state of the target object is determined.
  • the patient data includes at least one of the following information: patient's medical record information, mechanical ventilation parameters, monitoring parameters, blood gas value parameters, ultrasound parameters, disease severity score, fluid balance information
  • patient The medical record information includes at least one of the following information: age, height, weight, BMI index, comorbidity information, genetic disease information, drug treatment information
  • mechanical ventilation parameters include at least one of the following information: tidal volume, Minute ventilation, respiratory rate, inspiratory time, PEEP, mean airway pressure, inspired oxygen concentration, pressure support level, maximum inspiratory pressure, maximum expiratory pressure, end-tidal carbon dioxide, lung compliance, airway resistance, mechanical energy, Work of breathing, driving pressure, peak pressure, plateau pressure
  • the monitoring parameters include at least one of the following parameters: blood oxygen, heart rate, blood pressure, mean arterial pressure, central venous pressure, temperature
  • blood gas value parameters include the following information At least one of: partial pressure of carbon dioxide in arterial blood, partial pressure of oxygen in arterial blood, pH of arterial blood gas
  • the disease severity score includes at least one
  • the method further includes: when it is determined that there is an abnormal ventilation event, outputting prompt information.
  • the method also includes:
  • the result of the assessment of the degree of lung injury is displayed on a monitor.
  • the evaluating the lung injury degree of the target object according to the identification result and the patient data of the target object includes:
  • a degree of lung injury of the target subject is assessed based on the identification result and the tidal volume in the patient data of the target subject.
  • the basic waveform and auxiliary waveform data of the target object can be recognized based on the trained recognition model, and the recognition result of the current ventilation state of the target object can be determined. Therefore, it is possible Improve the accuracy of identifying abnormal ventilation events in the aspect of man-machine synchronization, and assist doctors to make judgments on man-machine confrontation events, thereby reducing the occurrence of man-machine confrontation and improving the breathing comfort and safety of mechanically ventilated patients.
  • the medical device can identify and process the respiratory waveform data and patient data of the target object based on the trained recognition model, and determine the recognition result of the current ventilation state of the target object. Therefore, it can improve the human-machine synchronization aspect.
  • the accuracy of identifying abnormal ventilation events can assist doctors to make judgments on man-machine confrontation events, thereby reducing the occurrence of man-machine confrontation and improving the breathing comfort and safety of mechanically ventilated patients.
  • Fig. 1 shows a schematic block diagram of a medical device in one embodiment of the present invention
  • Figure 2 shows a schematic block diagram of a ventilator in one embodiment of the present invention
  • Fig. 3 shows the schematic flowchart when the processor in one embodiment of the present invention is identified through a neural network model
  • Fig. 4 shows the schematic flowchart when the processor in one embodiment of the present invention is identified by a neural network model and an expert system
  • Fig. 5 shows a schematic flow chart when the processor in one embodiment of the present invention recognizes through a neural network model and an expert system in combination with patient data;
  • Fig. 6 shows a schematic flow chart when the processor in one embodiment of the present invention recognizes through multiple neural network models
  • Fig. 7 shows a schematic flow chart when the processor in one embodiment of the present invention recognizes through multiple neural network models and expert systems
  • Fig. 8 shows a schematic flow chart when the processor in one embodiment of the present invention recognizes through multiple neural network models and expert systems combined with patient data;
  • Fig. 9 shows a flowchart of a method for identifying abnormal ventilation in an embodiment of the present invention.
  • Fig. 10 shows a schematic flow chart when a processor in another embodiment of the present invention performs recognition through a neural network model
  • Fig. 11 shows a schematic flow chart when the processor in another embodiment of the present invention recognizes through the neural network model
  • Fig. 12 shows a schematic flow chart when the processor in an embodiment of the present invention recognizes through a neural network model and an expert system
  • Fig. 13 shows a schematic flow chart when the processor in one embodiment of the present invention performs identification through a neural network model and an expert system combined with patient data;
  • Fig. 14 shows a flowchart of a method for identifying abnormal ventilation in another embodiment of the present invention.
  • the present application provides a medical device, including: a memory for storing executable program instructions; a processor for executing the program instructions stored in the memory, so that the processor performs the following steps: acquiring the target object The current respiratory waveform data, wherein the respiratory waveform data includes basic respiratory waveform data and auxiliary waveform data; based on the trained identification model, the respiratory waveform data of the target object is recognized and processed to determine the recognition result of the current ventilation state of the target object, and the recognition result Includes normal or abnormal ventilation events.
  • the medical device can identify and process the basic waveform and auxiliary waveform data of the target object based on the trained recognition model, and determine the recognition result of the current ventilation state of the target object. Therefore, it can improve the detection of abnormal ventilation events in terms of man-machine synchronization
  • the accuracy of recognition assists doctors to make judgments on man-machine confrontation events, thereby reducing the occurrence of man-machine confrontation and improving the breathing comfort and safety of mechanically ventilated patients.
  • the medical equipment in this application may be a ventilator, an anesthesia machine, a monitor, or a central station.
  • the medical device when it is a monitor or a central station, it includes a communication interface, which can communicate with a ventilator, so as to obtain various data output from the ventilator or anesthesia machine, such as respiratory waveform data, various Setting parameter information, etc.
  • respiratory waveform data refers to waveform data related to respiration.
  • the central station can directly obtain various data output by the ventilator or anesthesia machine through its communication interface, or the central station can also be connected to a monitor, and the monitor is connected to the respiratory system.
  • the ventilator or anesthesia machine may also include a communication interface through which a monitor or a central station is communicatively connected to obtain various monitoring parameters of the ventilated patient from the monitor or the central station, such as blood oxygen ( SpO2), heart rate (HR), blood pressure (BP), mean arterial pressure (MAP), central venous pressure (CVP), temperature (T), etc.
  • a monitor or a central station is communicatively connected to obtain various monitoring parameters of the ventilated patient from the monitor or the central station, such as blood oxygen ( SpO2), heart rate (HR), blood pressure (BP), mean arterial pressure (MAP), central venous pressure (CVP), temperature (T), etc.
  • the medical device 100 of the present application includes one or more processors 50 , a display 70 , a memory 60 , and a communication interface. These components are interconnected by a bus system and/or other form of connection mechanism (not shown). It should be noted that the components and structure of the medical device 100 shown in FIG. 1 are only exemplary rather than limiting, and the medical device 100 may also have other components and structures as required.
  • the memory 60 is used to store various data and executable programs generated during the use of relevant medical equipment, for example, to store system programs of medical equipment, various application programs or algorithms for realizing various specific functions.
  • One or more computer program products may be included, and computer program products may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory.
  • the volatile memory may include random access memory (RAM) and/or cache memory (cache), etc., for example.
  • Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, and the like.
  • the processor 50 may be a central processing unit (CPU), a graphics processing unit (GPU), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other forms of processing with data processing capabilities and/or instruction execution capabilities. unit, and can control other components in the medical device to perform desired functions.
  • processor 50 can include one or more embedded processors, processor cores, microprocessors, logic circuits, hardware finite state machines (FSMs), digital signal processors (DSPs), graphics processing units (GPUs) or their combination.
  • FSMs hardware finite state machines
  • DSPs digital signal processors
  • GPUs graphics processing units
  • the medical equipment can also include a human-computer interaction device, which can include a display 70, which is used to display the respiratory waveform data when the ventilator is ventilating the patient, display the patient's status information, the recognition result of the ventilation status, and various prompt information Or alarm information, etc., the specific content displayed can include text, charts, numbers, colors, waveforms, characters, etc., used to intuitively display various information.
  • the human-computer interaction device may also include an input device, through which the medical staff can set various parameters, select and control the display interface of the display 70, etc., to realize information interaction between man and machine.
  • the display 70 can also be a touch display.
  • the human-computer interaction interface may refer to the display interface of the display 70 .
  • the medical device can be a ventilator, which is an artificial mechanical ventilation device used to assist or control the breathing movement of the patient to achieve gas exchange in the lungs and reduce the work of breathing for the patient to facilitate respiratory function. recovery.
  • the ventilator can also include a breathing interface 211 (i.e., a patient interface), an air source interface 212, a breathing circuit (i.e., a breathing circuit), a breathing assistance device (i.e., an airflow providing device), for A ventilation detection device for detecting ventilation parameters such as respiratory waveform data, a processor 50, a memory 60, a display 70, and the like.
  • the ventilation detection device (not shown) is arranged on the breathing circuit or the patient interface, and is used to detect various respiratory waveform data, etc., which may include the flow rate of the ventilation flow, airway pressure, respiratory rate, tidal volume, inspiratory time, respiratory rate, etc. system or lung compliance, etc. It should be noted that the detection of the respiratory waveform data can be obtained by direct detection, or can be obtained by calculation after certain basic parameters are detected.
  • the breathing circuit selectively communicates the gas source interface 212 with the patient's breathing system.
  • the breathing circuit includes an expiratory branch 213a and an inspiratory branch 213b.
  • the expiratory branch 213a is connected between the respiratory interface 211 and the exhaust port 213c, and is used to guide the patient's exhaled air to the exhaust port. 213c.
  • the exhaust port 213c can lead to the external environment, and also can be in a dedicated gas recovery device for the channel.
  • the gas source interface 212 is used to connect with the gas source (not shown in the figure), and the gas source is used to provide gas, and the gas can generally adopt oxygen and air, etc.; in some embodiments, the gas source can adopt compressed gas cylinders or
  • the central air supply source supplies air to the ventilator through the air source interface 212.
  • the types of air supply include oxygen O2 and air, etc.
  • the air source interface 212 can include pressure gauges, pressure regulators, flow meters, pressure reducing valves and air-oxygen Conventional components such as proportional regulation protection devices are used to control the flow of various gases (such as oxygen and air).
  • the inspiratory branch 213b is connected between the respiratory interface 211 and the air source interface 212, and is used to provide oxygen or air for the patient. the patient's lungs.
  • the respiratory interface 211 is used to connect the patient to the breathing circuit.
  • the gas exhaled by the patient can also be introduced to the exhaust port 213c through the expiratory branch 213a;
  • the breathing interface 211 may be a nasal cannula or a face mask worn on the mouth and nose, or the breathing interface 211 may also be a nasal mask, a nasal cannula, or a tracheal tube.
  • the breathing assistance device is connected with the gas source interface 212 and the breathing circuit, and controls the gas provided by the external gas source to be delivered to the patient through the breathing circuit; in some embodiments, the breathing assistance device may include an exhalation controller 214a and an inhalation controller 214b, The exhalation controller 214a is arranged on the exhalation branch 213a, and is used for turning on the exhalation branch 213a or closing the exhalation branch 213a according to the control instruction, or controlling the flow rate or pressure of the gas exhaled by the patient.
  • the exhalation controller 214a may include one or more of devices capable of controlling flow or pressure, such as an exhalation valve, a one-way valve, a flow controller, and a PEEP valve.
  • the suction controller 214b is arranged on the suction branch 213b, and is used for turning on the suction branch 213b or closing the suction branch 213b according to a control command, or controlling the flow rate or pressure of the output gas.
  • the inhalation controller 214b may include one or more of devices capable of controlling flow or pressure, such as an exhalation valve, a one-way valve, or a flow controller.
  • the processor 50 can also be used to execute instructions or programs, control the various control valves in the breathing assistance device, the air source interface 212 and/or the breathing circuit, or process the received data to generate the The required calculation or judgment results, or generate visualization data or graphics, and output the visualization data or graphics to the display 70 for display.
  • the processor 50 of the medical device is used to execute the program instructions stored in the memory, so that the processor 50 performs the following steps: acquire the current respiratory waveform data of the target object, wherein the respiratory waveform data includes basic respiratory waveform data and auxiliary waveform data; The respiratory waveform data of the target object is identified and processed based on the trained identification model, and the identification result of the target object's current ventilation state is determined, and the identification result includes normal ventilation or abnormal ventilation events.
  • the respiratory waveform data refers to data related to respiration, which may be a waveform or data capable of characterizing a waveform.
  • the basic respiratory waveform data includes one or more of the following waveform data: pressure, flow rate, or volume.
  • the auxiliary waveform data includes one or more of the following waveforms: transpulmonary pressure, esophageal pressure, intragastric pressure or diaphragm electromyography, or other waveform data that can reflect the respiratory state of the patient.
  • the above-mentioned respiratory waveform data can be obtained based on the detection of the ventilation detection device used to detect ventilation parameters such as respiratory waveform data of the ventilator. Esophageal pressure is measured by inserting a catheter for measuring intragastric pressure into the patient's esophagus, and intragastric pressure can be measured by inserting a catheter for measuring intragastric pressure into the patient's stomach.
  • one or more of the above respiratory waveform data can also be calculated based on one or more data, for example, the transpulmonary pressure is the difference between the alveolar pressure and the intrathoracic pressure, where the airway pressure can be measured Characterize the alveolar pressure, measure the esophageal pressure to estimate the intrathoracic pressure, and then calculate the transpulmonary pressure based on the airway pressure and intrathoracic pressure.
  • the volume can be obtained by integrating the flow velocity.
  • new auxiliary waveform data can also be calculated based on one or more waveform data in respiratory waveform data (such as basic waveform data and auxiliary waveform data).
  • Pressure, esophageal pressure, intragastric pressure or other auxiliary waveform data that can reflect the respiratory state of the patient, such as combining the waveforms of pressure, flow rate, volume and esophageal pressure to derive new auxiliary waveform data.
  • the trained recognition model may be trained by machine learning methods, including but not limited to methods based on decision trees, random forests, Bayesian learning, deep learning, and the like.
  • the trained recognition model can be obtained by using deep learning to train the neural network model, that is, the trained recognition model includes a trained neural network model.
  • the neural network model includes but is not limited to CNN multi-layer Convolutional neural network, RNN cyclic neural network or CNN+RNN network, wherein the trained recognition model is obtained by using deep learning method training based on the historical data information of the respiratory waveform data of the previous patients, and the history of the respiratory waveform data of the previous patients
  • the data information includes historical data information of basic respiratory waveform data and historical data information of auxiliary waveform data.
  • the process of training the neural network model may include the following steps: First, establish a waveform database, and the samples in the database contain historical data information of previous patients' respiratory waveform data, such as including basic Historical data information of respiratory waveform data, historical data information of basic respiratory waveform data includes historical data information of pressure, historical data information of flow velocity and historical data information of volume information, and also includes historical data information of basic respiratory waveform data and esophageal pressure Different combinations of historical data information of auxiliary waveform information such as intragastric pressure or diaphragm EMG.
  • the database can contain training set and test set.
  • the database also includes a verification set (also called a tuning set), the training set adjusts the weight of each neuron in order to train the parameters of the neural network;
  • the verification set also called a tuning set
  • the hyperparameters such as the number of network layers, the number of neurons in each layer, and the learning rate, etc.
  • the test set is for the final evaluation of the performance of the model, and the generalization ability of the trained neural network model is given. Make a review.
  • the training set, validation set (tuning set) and test set together constitute the respiratory waveform, which are mutually exclusive and evenly distributed.
  • the duration of the respiratory waveform data of each sample in the database can include at least one or several complete respiratory cycles, wherein the database contains both normal waveforms and abnormal ventilation waveforms, and the abnormal ventilation waveforms are included in the total waveform data
  • the proportion is within the preset threshold range.
  • the probability of occurrence of abnormal ventilation waveforms can be simulated in order to ensure that the sample distribution conforms to the actual clinical situation, so that the trained recognition model For example, the neural network model can still give reasonable predictions and judgments when the information is incomplete.
  • each sample has its own label, whether it is a normal waveform or an abnormal ventilation waveform. These labels can be given by multiple clinical experts.
  • the neural network model is trained using deep learning methods to obtain the Before training the recognition model, the first data processing is performed on the historical data information of the respiratory waveform data to obtain abnormal waveforms with missing waveforms, so as to obtain more abnormal waveforms, and then use the deep learning method to train the neural network model , when obtaining the trained recognition model, put the abnormal waveforms with missing waveforms into the database, so as to expand the samples in the database, and then enhance the adaptability of the neural network model to the missing waveforms.
  • the first data processing method includes but is not limited to: data masking, adding noise, removing part of the waveform or removing any type of waveform, and the like.
  • the respiratory waveform data of the samples in the database can also be preprocessed to improve the recognition accuracy of the neural network model.
  • the preprocessing methods include But not limited to data normalization, taking pressure, flow velocity, volume and esophageal pressure as examples, the data can be normalized based on the following formula:
  • P 0 (t), F 0 (t), V 0 (t), Pes 0 (t) is the pressure, flow rate, volume and esophageal pressure at time t
  • P(t), F(t) , V(t), Pes(t) are normalized pressure, flow rate, volume and esophageal pressure, It can be an average value, or a fixed value can be given based on experience
  • P std , F std , V std , and Pes std are standard deviations.
  • the normalization method of other auxiliary waveform information is basically the same as that of esophageal pressure.
  • the neural network model can be trained by a deep learning algorithm to complete the mapping from respiratory waveform data (as well as basic waveform data and auxiliary waveform data) to abnormal ventilation event types.
  • the processor 50 performs recognition processing on the respiratory waveform data based on the trained recognition model to determine the recognition result of the current ventilation state of the target object, including: preprocessing the respiratory waveform data of the target object to obtain the processed Respiratory waveform data; based on the trained identification model, the processed respiratory waveform data is identified and processed to determine the identification result of the current ventilation state of the target object.
  • the preprocessing method includes, but is not limited to, one or more of data normalization and filtering processing, wherein when performing data normalization processing, its process is generally the same as the data normalization in the training process described above.
  • the method of normalization is the same and will not be described again here.
  • the processor 50 performs recognition processing on the respiratory waveform data of the target subject based on the trained recognition model, and determines the recognition result of the current ventilation state of the target subject, including: judging whether the quality of the auxiliary waveform data and the basic waveform data meets the requirements of Preset conditions; when it is determined that the quality of the auxiliary waveform data and the basic waveform data meets the preset conditions, the auxiliary waveform data and basic respiratory waveform data that meet the requirements are input into the trained recognition model for recognition processing, and the current ventilation status of the target object is determined recognition results; when it is determined that the quality of the auxiliary waveform data and the basic waveform data does not meet the preset conditions, the auxiliary waveform data and the basic respiratory waveform data are not input into the trained recognition model for recognition processing.
  • the accuracy of the recognition result can be guaranteed through quality judgment, and the problem of outputting wrong results or not outputting results due to unqualified quality of respiratory waveform data can be avoided.
  • the processor 50 can judge whether the quality of the auxiliary waveform data and the basic waveform meets the preset condition by using methods such as time domain method, frequency domain method or waveform template.
  • the time domain method refers to the judgment in the time domain by the preset characteristics of the waveform, such as maximum inspiratory pressure (MIP), average airway pressure, slope or time of pressure rise, inspiratory time, maximum flow rate,
  • MIP maximum inspiratory pressure
  • Vt tidal volume
  • each preset parameter indicator corresponds to its normal range, and the waveform quality can be judged by these parameters.
  • the waveform quality is good, otherwise Identified as unqualified; for the frequency domain method: transform the waveform into the frequency domain for spectrum analysis, get the proportion of different frequencies, compare it with the spectrum of the normal waveform, and judge it by the threshold value, for example, when the proportion of the high frequency part When the preset threshold is reached, it is considered that the waveform has too much noise and the quality is unqualified, and when the proportion of the high-frequency part is lower than the preset threshold, the quality of the waveform is judged to be qualified, or, for some waveforms, when the proportion of the low-frequency part When the ratio reaches the preset threshold, it can be judged that the waveform has too much noise and the quality is unqualified.
  • a waveform template can be established according to the normal waveform (for example, a pressure template can be established according to multiple normal pressure waveforms), and the respiratory waveform data scheduled to be input into the trained recognition model is compared with its corresponding waveform template, and the whole cycle The degree of difference is calculated, and the quality is judged by the threshold. For example, when the degree of difference is greater than the preset threshold, it is judged that the quality of the respiratory waveform data is unqualified; when the degree of difference is lower than or equal to the preset threshold, the quality of the respiratory waveform data is judged Quality standards.
  • the quality of part of the basic waveform data in the multiple basic waveform data and part of the auxiliary waveform data in the multiple auxiliary waveform data meets the preset
  • the basic waveform data and auxiliary waveform data that meet the preset conditions can also be input into the trained recognition model for recognition.
  • the number of trained recognition models can be one, for example, the trained recognition model is a trained neural network model, and the processor 50 is also used to: use the basis of the target object The respiratory waveform data and/or the auxiliary waveform data of the target object are fused to obtain fused data; the fused data is input into the trained recognition model for recognition processing, and the recognition result of the current ventilation state of the target object is determined.
  • the trained recognition model is a trained neural network model
  • the processor 50 is also used to: use the basis of the target object
  • the respiratory waveform data and/or the auxiliary waveform data of the target object are fused to obtain fused data; the fused data is input into the trained recognition model for recognition processing, and the recognition result of the current ventilation state of the target object is determined.
  • different types and quantities of basic respiratory waveform data and/or different types of auxiliary waveform data can be combined to form multiple types of combined waveform data, such as basic respiratory waveform data A combination of at least two types of waveform data in the basic respiratory waveform data, or a combination of at least one type of waveform data in the basic respiratory waveform data and at least one type of waveform data in the auxiliary waveform data, or at least two types of auxiliary waveform data A combination of waveform data.
  • Fusion processing refers to the fusion of each combination of waveform data (the combined waveform data includes at least two types of waveform data), such as size alignment, etc., and the purpose of fusion processing is to determine the final input to the trained Identify the format of the model.
  • the identification results of the current ventilation state include but are not limited to normal ventilation or abnormal ventilation events, wherein the abnormal ventilation events include one or more of the following events: invalid trigger, double inspiratory (specifically, double inspiratory Can include double triggering and breath superimposition), false triggering, too small flow rate, reverse triggering, early switching, delayed switching, slow pressure rise, too fast pressure rise, intrinsic positive end-expiratory pressure ventilation, short inspiratory time, Long inspiratory time, water in the circuit, excessive resistance, low compliance, inverse ratio ventilation, etc., or other identifiable abnormal ventilation events may also be included.
  • the abnormal ventilation events include one or more of the following events: invalid trigger, double inspiratory (specifically, double inspiratory Can include double triggering and breath superimposition), false triggering, too small flow rate, reverse triggering, early switching, delayed switching, slow pressure rise, too fast pressure rise, intrinsic positive end-expiratory pressure ventilation, short inspiratory time, Long inspiratory time, water in the circuit, excessive resistance, low compliance, inverse ratio ventilation, etc., or other
  • the processor 50 performs recognition processing on the respiratory waveform data of the target object based on the trained recognition model, and determines the recognition result of the current ventilation state of the target object, including: recognizing the breathing waveform data of the target object based on the trained recognition model
  • the waveform data is identified and processed to determine the identification result of the current ventilation state of the target object and the corresponding reliability of the identification result (also referred to as the confidence level herein).
  • the confidence level also referred to as the confidence level herein.
  • the number of trained recognition models can also be multiple, for example, when the trained recognition models include multiple trained neural network models, as shown in Figure 6, the respiratory waveform data includes multiple combinations At least one of the waveform data, each trained neural network model corresponds to a combined waveform data, optionally, the combined waveform data includes one or more of the following combinations: at least one type of basic respiratory waveform data, at least A combination of one type of auxiliary waveform data and at least one type of basic respiratory waveform data, for example, the basic respiratory waveform data includes one or more of the following waveform data: pressure, flow rate or volume, and the auxiliary waveform data includes one of the following waveforms or more: transpulmonary pressure, esophageal pressure, intragastric pressure or diaphragm electromyography, then at least one type of basic respiratory waveform data can be pressure, flow rate or solvent, and can also include any combination of two types of pressure, flow rate and solvent and their The combination of the three types, the combination of at least one type of auxiliary waveform data and at least one type of
  • the processor 50 is also used to: obtain at least one combined waveform data based on the obtained basic respiratory waveform data and auxiliary waveform data, and input various combined waveform data into its corresponding trained neural network Recognition is performed in the model to obtain a set of recognition results, wherein the recognition results may include normal ventilation or abnormal ventilation events, and correspond to the reliability of the recognition results.
  • the user can obtain one or more recognition results, and for one or more recognition results that may be different or partly the same and partly different, the user can choose the credible result according to his own experience, or he can reliability to determine the patient's current ventilation status.
  • the processor 50 performs recognition processing on the respiratory waveform data of the target object, and determines the recognition result of the current ventilation state of the target object, and further includes: performing recognition processing on the respiratory waveform data of the target object based on the trained recognition model After that, use the expert system to identify and process the respiratory waveform data of the target object to determine the recognition result of the current ventilation state of the target object.
  • the expert system performs recognition processing (for example, using the knowledge base of the expert system for reasoning) based on the output value of the trained neural network model combined with respiratory waveform data to determine the recognition result of the patient's current ventilation state
  • recognition processing for example, using the knowledge base of the expert system for reasoning
  • the expert system is used to perform identification processing according to the output value of each trained neural network model combined with respiratory waveform data, so as to determine the identification result of the patient's current ventilation state.
  • the processor 50 performs identification processing on the respiratory waveform data of the target object using an expert system, including: acquiring patient data of the target object; and identifying processing on the respiratory waveform data and patient data using an expert system, for example, as shown in FIG. 5 and Fig. 8, the recognition result (including the type of recognized recognition result) and the confidence P of the trained recognition model (such as a neural network model) for recognition output are input to the expert system, and the expert system is used to obtain The patient data and respiratory waveform data are identified, the knowledge base in the expert system is used for reasoning, and the output results of the neural network model are combined to give the final identification result and the final confidence level P1.
  • an expert system including: acquiring patient data of the target object; and identifying processing on the respiratory waveform data and patient data using an expert system, for example, as shown in FIG. 5 and Fig. 8, the recognition result (including the type of recognized recognition result) and the confidence P of the trained recognition model (such as a neural network model) for recognition output are input to the expert system, and the expert system is used
  • the patient data may be input by the user through the human-computer interaction interface, or may also be obtained by the processor 50 by obtaining information such as the patient's electronic medical record.
  • the patient data may include one or more of the following data: patient's medical record information, mechanical ventilation parameters, monitoring parameters, blood gas value parameters, ultrasound parameters, disease severity score, fluid balance information, etc.
  • the patient's Medical record information includes at least one of the following information: age, height, weight, BMI index (where BMI index is weight (kg) divided by height (m) squared), comorbidity information, genetic disease information, drug treatment information ( Such as antibiotic type), etc.
  • mechanical ventilation parameters include at least one of the following information: tidal volume (Vt), minute ventilation (MV), respiratory rate (RR), inspiratory time, PEEP, mean airway pressure, inspired oxygen Concentration (FiO2), pressure support level (PS level), maximum inspiratory pressure (MIP), maximum expiratory pressure (MEP), end-tidal carbon dioxide (ETCO2), lung compliance
  • Vt
  • an expert system is a computer program designed to model the problem-solving ability of human experts. It is an intelligent computer program system, which contains the knowledge and experience of one or more experts in the field, and can use the knowledge and problem-solving methods of human experts to solve problems in this field.
  • the expert system is pre-stored in the memory, and the processor 50 to be used when needed.
  • the expert system in this article combines the professional knowledge, experience and problem-solving methods of many ICU clinicians to identify, judge and process the respiratory waveform, and further confirm the identification results of the neural network model, so that the obtained identification results can be obtained. It is more accepted and recognized by users, and makes the accuracy of recognition results higher.
  • the processor 50 is configured to: select a corresponding recognition mode according to an instruction input by the user; The data is identified and processed to determine the identification result of the current ventilation state of the target object; when the identification mode is the second identification mode, the expert system is used to identify and process the respiratory waveform data of the target object and the patient data of the target object to determine the current ventilation status of the target object.
  • the processor 50 is further configured to: when it is determined that there is an abnormal ventilation event, output prompt information.
  • the display 70 is used to obtain and display the prompt information, which may include a text description of the event type of the abnormal ventilation event, or a simulated image prompt, etc., through the prompt information, the user can be intuitively prompted, so that The user obtains the abnormal ventilation events of the patient in a timely manner, and takes corresponding measures to deal with the abnormal ventilation events, thereby improving the breathing comfort of the patients.
  • the display 70 is used to display the prompt information in a preset display manner, and the preset display manner includes one or more of the following manners: highlighting, Additional symbol display, differentiated color, differentiated shading, flashing, or other suitable display methods for display.
  • a button is set on the display interface of the display 70 of the medical device, and the processor 50 is configured to: control the display 70 to display or hide the prompt information when acquiring a user instruction input by the user through the button, by setting the button, The user can display the prompt information on the display interface of the display 70 only when necessary, thereby avoiding too much display information on the display interface and affecting the user experience, and can also hide the prompt information through this button when not needed.
  • the medical device of the present application further includes an alarm device, and the medical device also includes an alarm device, which is used to give an alarm prompt when the number or frequency of abnormal ventilation events within a preset time meets a preset condition, for example, When the man-machine antagonism index exceeds the preset threshold within the preset time, an alarm is given, wherein the man-machine antagonism index is the ratio of the number of cycles of abnormal ventilation events to the total number of cycles within the preset time. When the preset condition is met, it indicates that the abnormal ventilation event is relatively serious, and it is likely to have a negative impact on the patient. Therefore, by means of an alarm prompt, the doctor's attention is immediately drawn to make reasonable treatment measures for the abnormal ventilation event .
  • a preset condition for example, When the man-machine antagonism index exceeds the preset threshold within the preset time, an alarm is given, wherein the man-machine antagonism index is the ratio of the number of cycles of abnormal ventilation events to the total number of cycles within the preset
  • the processor 50101 may generate alarm information (such as an alarm signal) when the number or frequency of abnormal ventilation events within a preset time meets a preset condition.
  • the alarm device is configured to obtain the alarm information and give an alarm prompt.
  • the alarm mode of the alarm device includes but is not limited to light, sound and other alarm methods.
  • the specific form of the alarm device can be a flashing LED light, a buzzer, a speaker, etc., for example, When the alarm device is a loudspeaker, it can also ventilate the prompt sound of abnormal events, such as voice broadcast, etc., which meets the requirements for the strength of the alarm signal, and is enough to attract the attention and vigilance of observers. In this way, a real-time alarm can be realized to prompt the user.
  • the processor 50 is further configured to: when the recognition result is an abnormal ventilation event, according to the recognition result, automatically adjust the setting parameters of the ventilator for ventilating the target subject, such as adjusting tidal volume (Vt), minute ventilation (MV), respiratory rate (RR), inspiratory time, PEEP, mean airway pressure, inspired oxygen concentration (FiO2), PS level (pressure support level), maximum inspiratory pressure (MIP), maximum expiratory pressure (MEP ) and other setting parameters, so that the patient can quickly recover from the abnormal ventilation event to the normal ventilation state, and improve the safety and comfort of the patient's mechanical ventilation.
  • Vt tidal volume
  • MV minute ventilation
  • RR respiratory rate
  • PEEP adjusting tidal volume
  • mean airway pressure such as adjusting tidal volume (Vt), minute ventilation (MV), respiratory rate (RR), inspiratory time, PEEP, mean airway pressure, inspired oxygen concentration (FiO2), PS level (pressure support level), maximum inspiratory pressure (MIP), maximum expiratory pressure (ME
  • the processor 50 is further configured to: when the recognition result is an abnormal ventilation event, output adjustment suggestion information of the setting parameters of the ventilator for ventilating the target subject according to the recognition result, and the display 70 is used to obtain and display the adjustment Suggestion information, in this way, can give the user adjustment suggestion information to assist the user in handling abnormal ventilation events.
  • the ventilation state identification method of the present application includes the following steps S910 to S920:
  • step S910 the current respiratory waveform data of the target object is acquired, wherein the respiratory waveform data includes basic respiratory waveform data and auxiliary waveform data.
  • the respiratory waveform data includes basic respiratory waveform data and auxiliary waveform data.
  • step S920 the respiratory waveform data of the target subject is recognized based on the trained recognition model, and the recognition result of the target subject's current ventilation state is determined, and the recognition result includes normal ventilation or abnormal ventilation events.
  • the trained identification model is obtained through machine learning training.
  • the trained recognition model includes a trained neural network model.
  • the trained recognition model is obtained by using deep learning methods based on the historical data information of the patient's respiratory waveform data.
  • the historical data information of the respiratory waveform data includes basic respiratory History data information of waveform data and history data information of auxiliary waveform data.
  • performing recognition processing on the respiratory waveform data to determine the recognition result of the current ventilation state of the target object includes: performing preprocessing on the respiratory waveform data of the target object to obtain processed respiratory waveform data, wherein the preprocessing includes Normalization processing; based on the trained identification model, the processed respiratory waveform data is identified and processed to determine the identification result of the current ventilation state of the target object.
  • the respiratory waveform data of the target object is identified and processed based on the trained identification model, and the recognition result of the current ventilation state of the target object is determined, including: judging whether the quality of the auxiliary waveform data and the basic respiratory waveform data meet the preset conditions ; When it is determined that the quality of the auxiliary waveform data and the basic respiratory waveform data meet the preset conditions, input the auxiliary waveform data and the basic respiratory waveform data that meet the preset conditions into the trained identification model for identification processing, and determine the current ventilation state of the target object recognition results; when it is determined that the quality of the auxiliary waveform data and the basic waveform does not meet the preset conditions, the auxiliary waveform data and the basic respiratory waveform data are not input into the trained recognition model for recognition processing.
  • the accuracy of the recognition result can be guaranteed through quality judgment, and the problem of outputting wrong results or not outputting results due to unqualified quality of respiratory waveform data can be avoided.
  • performing identification processing on the respiratory waveform data based on the trained identification model, and determining the identification result of the current ventilation state of the target object includes: performing basic respiratory waveform data of the target object and/or auxiliary waveform data of the target object Fusion processing to obtain fusion data; input the fusion data into the trained recognition model for recognition processing, and determine the recognition result of the current ventilation state of the target object.
  • the recognition process is performed on the respiratory waveform data of the target object based on the trained identification model, and the identification result of the current ventilation state of the target object is determined, including: identifying the respiratory waveform data of the target object based on the trained identification model Processing, determining the recognition result of the current ventilation state of the target object and the corresponding reliability of the recognition result.
  • performing recognition processing on the respiratory waveform data to determine the recognition result of the target object's current ventilation state includes: after performing recognition processing on the target object's respiratory waveform data based on the trained recognition model, using an expert system to identify the target object Recognition processing is performed on the breathing waveform data of the target object to determine the recognition result of the current ventilation state of the target object.
  • the method of the present application further includes: when it is determined that there is an abnormal ventilation event, outputting prompt information, and displaying the prompt information through a display.
  • the method of the present application further includes: giving an alarm prompt when the number or frequency of abnormal ventilation events within a preset time meets a preset condition.
  • the alarm prompt can be performed based on the aforementioned alarm device.
  • the method of the present application further includes: when the recognition result is an abnormal ventilation event, automatically adjusting the setting parameters of the ventilator for ventilating the target subject according to the recognition result.
  • the basic waveform and auxiliary waveform data of the target object can be recognized and processed based on the trained recognition model, and the recognition result of the current ventilation state of the target object can be determined. Therefore, the human-machine synchronization can be improved.
  • the accuracy of identifying abnormal ventilation events can assist doctors to make judgments on man-machine confrontation events, thereby reducing the occurrence of man-machine confrontation and improving the breathing comfort and safety of mechanically ventilated patients.
  • the medical device in another embodiment of the present application will be described below with reference to FIGS. 1 , 2 and 10 to 13 .
  • the medical equipment in this embodiment of the present application may be a ventilator, an anesthesia machine, a monitor, or a central station.
  • the medical device when it is a monitor or a central station, it includes a communication interface, which can communicate with a ventilator, so as to obtain various data output from the ventilator or anesthesia machine, such as respiratory waveform data, various Setting parameter information, etc.
  • respiratory waveform data refers to waveform data related to respiration.
  • the central station can directly obtain various data output by the ventilator or anesthesia machine through its communication interface, or the central station can also be connected to a monitor, and the monitor is connected to the respiratory system.
  • the ventilator or anesthesia machine may also include a communication interface through which a monitor or a central station is communicatively connected to obtain various monitoring parameters of the ventilated patient from the monitor or the central station, such as blood oxygen ( SpO2), heart rate (HR), blood pressure (BP), mean arterial pressure (MAP), central venous pressure (CVP), temperature (T), etc.
  • a monitor or a central station is communicatively connected to obtain various monitoring parameters of the ventilated patient from the monitor or the central station, such as blood oxygen ( SpO2), heart rate (HR), blood pressure (BP), mean arterial pressure (MAP), central venous pressure (CVP), temperature (T), etc.
  • the medical device 100 of the present application includes one or more processors 50 , a display 70 , a memory 60 , and a communication interface. These components are interconnected by a bus system and/or other form of connection mechanism (not shown). It should be noted that the components and structure of the medical device 100 shown in FIG. 1 are only exemplary rather than limiting, and the medical device 100 may also have other components and structures as required. For some details about the medical device 100, reference may also be made to the foregoing description, which will not be repeated here.
  • the medical device can be a ventilator, which is an artificial mechanical ventilation device used to assist or control the breathing movement of the patient to achieve gas exchange in the lungs and reduce the work of breathing for the patient to facilitate respiratory function. recovery.
  • the ventilator can also include a breathing interface 211 (i.e., a patient interface), an air source interface 212, a breathing circuit (i.e., a breathing circuit), a breathing assistance device (i.e., an airflow providing device), for A ventilation detection device for detecting ventilation parameters such as respiratory waveform data, a processor 50, a memory 60, a display 70, and the like.
  • a breathing interface 211 i.e., a patient interface
  • an air source interface 212 i.e., a breathing circuit
  • a breathing assistance device i.e., an airflow providing device
  • the processor 50 of the medical device is used to execute the program instructions stored in the memory, so that the processor 50 performs the following steps: acquire the current respiratory waveform data and patient data of the target object, wherein the respiratory waveform data includes basic respiratory waveform data and/or auxiliary waveform data; perform identification processing on the respiratory waveform data and the patient data based on the trained identification model, and determine the identification result of the current ventilation state of the target object, the identification result including normal ventilation or abnormal ventilation event.
  • the respiratory waveform data refers to the data related to respiration, which may be the waveform of the parameter or the data that can characterize the waveform of the parameter.
  • the basic respiratory waveform data includes one or more of the following waveform data: pressure, flow rate, or volume.
  • the auxiliary waveform data includes one or more of the following waveforms: transpulmonary pressure, esophageal pressure, intragastric pressure or diaphragm electromyography, or other waveform data that can reflect the respiratory state of the patient.
  • the above-mentioned respiratory waveform data can be obtained based on the detection of the ventilation detection device used to detect ventilation parameters such as respiratory waveform data of the ventilator. Esophageal pressure is measured by inserting a catheter for measuring intragastric pressure into the patient's esophagus, and intragastric pressure can be measured by inserting a catheter for measuring intragastric pressure into the patient's stomach.
  • one or more of the above respiratory waveform data can also be calculated based on one or more data, for example, the transpulmonary pressure is the difference between the alveolar pressure and the intrathoracic pressure, where the airway pressure can be measured Characterize the alveolar pressure, measure the esophageal pressure to estimate the intrathoracic pressure, and then calculate the transpulmonary pressure based on the airway pressure and intrathoracic pressure.
  • the volume can be obtained by integrating the flow velocity.
  • new auxiliary waveform data can also be calculated based on one or more waveform data in respiratory waveform data (such as basic respiratory waveform data and auxiliary waveform data).
  • Pulmonary pressure, esophageal pressure, intragastric pressure or other auxiliary waveform data that can reflect the respiratory state of the patient, such as combining the waveforms of pressure, flow rate, volume and esophageal pressure to derive new auxiliary waveform data.
  • the trained recognition model may be trained by machine learning methods, including but not limited to methods based on decision trees, random forests, Bayesian learning, deep learning, and the like.
  • the trained recognition model can be obtained by using deep learning to train the neural network model, that is, the trained recognition model includes a trained neural network model.
  • the neural network model includes but is not limited to CNN multi-layer Convolutional neural network, RNN cyclic neural network or CNN+RNN network, wherein the trained recognition model is obtained by training with deep learning method based on the historical data information of the patient data of the previous patients and the respiratory waveform data of the previous patients.
  • the historical data information of the respiratory waveform data includes the historical data information of the basic respiratory waveform data and the historical data information of the auxiliary waveform data.
  • the process of training the neural network model may include the following steps: First, establish a waveform database, and the samples in the database contain historical data information of previous patients' respiratory waveform data, such as including basic Historical data information of respiratory waveform data, historical data information of basic respiratory waveform data includes historical data information of pressure, historical data information of flow velocity and historical data information of volume information, and also includes historical data information of basic respiratory waveform data and esophageal pressure Different combinations of historical data information of auxiliary waveform information such as intragastric pressure or diaphragm EMG.
  • the database can contain training set and test set.
  • the database also includes a verification set (also called a tuning set), the training set adjusts the weight of each neuron in order to train the parameters of the neural network;
  • the verification set also called a tuning set
  • the hyperparameters such as the number of network layers, the number of neurons in each layer, and the learning rate, etc.
  • the test set is for the final evaluation of the performance of the model, and the generalization ability of the trained neural network model is given. Make a review.
  • the training set, validation set (tuning set) and test set together constitute the respiratory waveform, which are mutually exclusive and evenly distributed.
  • the duration of the respiratory waveform data of each sample in the database can include at least one or several complete respiratory cycles, wherein the database contains both normal waveforms and abnormal ventilation waveforms, and the abnormal ventilation waveforms are included in the total waveform data
  • the proportion is within the preset threshold range.
  • the probability of occurrence of abnormal ventilation waveforms can be simulated in order to ensure that the sample distribution conforms to the actual clinical situation, so that the trained recognition model For example, the neural network model can still give reasonable predictions and judgments when the information is incomplete.
  • each sample has its own label, whether it is a normal waveform or an abnormal ventilation waveform. These labels can be given by multiple clinical experts.
  • the neural network model is trained using deep learning methods to obtain Before the trained recognition model, the first data processing is performed on the historical data information of the respiratory waveform data to obtain abnormal waveforms with missing waveforms, so as to obtain more abnormal waveforms. Afterwards, the deep learning method is used to perform neural network model Training, to obtain the trained recognition model, put the abnormal waveforms with missing waveforms into the database, so as to expand the samples in the database, and then enhance the adaptability of the neural network model to the missing waveforms.
  • the first data processing method includes but is not limited to: data masking, adding noise, removing part of the waveform or removing any type of waveform, and the like.
  • the respiratory waveform data of the samples in the database can also be preprocessed to improve the recognition accuracy of the neural network model.
  • the preprocessing methods include But not limited to data normalization and/or filtering processing, taking pressure, flow velocity, volume and esophageal pressure as examples, data normalization processing can be performed based on the following formula:
  • P 0 (t), F 0 (t), V 0 (t), Pes 0 (t) is the pressure, flow rate, volume and esophageal pressure at time t
  • P(t), F(t) , V(t), Pes(t) are normalized pressure, flow rate, volume and esophageal pressure, It can be an average value, or a fixed value can be given based on experience
  • P std , F std , V std , and Pes std are standard deviations.
  • the normalization method of other auxiliary waveform information is basically the same as that of esophageal pressure.
  • the neural network model can be trained by a deep learning algorithm to complete the mapping from respiratory waveform data (such as basic respiratory waveform data and auxiliary waveform data) to abnormal ventilation event types.
  • respiratory waveform data such as basic respiratory waveform data and auxiliary waveform data
  • a neural network model can include multiple layers, such as an input layer, a hidden layer, and an output layer.
  • its hidden layer can include a convolutional layer, a pooling layer, and a fully connected layer.
  • the respiratory waveform data is input to the input layer of the neural network model, and the neural network output is obtained after being processed by the hidden layer of the neural network model (the neural network output can refer to the extracted output of the preset layer in the hidden layer features), and the patient data of previous patients are mapped by F transformation to obtain the parameter matrix, and the parameter matrix and the neural network output are obtained in the neural network model (for example, the parameter matrix and the neural network output are jointly input to the default layer of the hidden layer.
  • the next layer which may be a hidden layer or a fully connected layer), performs information fusion, such as inner product, to output the final recognition result.
  • the processor 50 executes recognition processing on the respiratory waveform data based on the trained recognition model, and determines the recognition result of the current ventilation state of the target subject, including: performing recognition processing on the respiratory waveform data and the The patient data is identified and processed to determine the identification result of the current ventilation state of the target object, and the identification result includes normal ventilation or abnormal ventilation events.
  • the preprocessing method includes, but is not limited to, one or more of data normalization and filtering processing, wherein when performing data normalization processing, its process is generally the same as the data normalization in the training process described above.
  • the method of normalization is the same and will not be described again here.
  • the processor 50 performs recognition processing on the respiratory waveform data of the target subject based on the trained recognition model, and determines the recognition result of the target subject's current ventilation state, including: judging the respiratory waveform data (for example, including basic respiratory waveform data and (or whether the quality of the auxiliary waveform data) satisfies the preset condition; when it is determined that the quality of the respiratory waveform data meets the preset condition, the patient data of the satisfied respiratory waveform data and the target object are input into the trained identification model for identification processing , to determine the recognition result of the current ventilation state of the target object; when it is determined that the quality of the respiratory waveform data does not meet the preset condition, the respiratory waveform data and the patient data of the target object are not input into the trained recognition model for recognition processing.
  • the accuracy of the recognition result can be guaranteed through quality judgment, and the problem of outputting wrong results or not outputting results due to unqualified quality of respiratory waveform data can be avoided.
  • the processor 50 can judge whether the quality of the auxiliary waveform data and the basic respiratory waveform data meets the preset condition by using methods such as time domain method, frequency domain method or waveform template.
  • the time domain method refers to the judgment in the time domain by the preset characteristics of the waveform, such as maximum inspiratory pressure (MIP), average airway pressure, slope or time of pressure rise, inspiratory time, maximum flow rate,
  • MIP maximum inspiratory pressure
  • Vt tidal volume
  • each preset parameter indicator corresponds to its normal range, and the waveform quality can be judged by these parameters.
  • the waveform quality is good, otherwise Identified as unqualified; for the frequency domain method: transform the waveform into the frequency domain for spectrum analysis, get the proportion of different frequencies, compare it with the spectrum of the normal waveform, and judge it by the threshold value, for example, when the proportion of the high frequency part When the preset threshold is reached, it is considered that the waveform has too much noise and the quality is unqualified, and when the proportion of the high-frequency part is lower than the preset threshold, the quality of the waveform is judged to be qualified, or, for some waveforms, when the proportion of the low-frequency part When the ratio reaches the preset threshold, it can be judged that the waveform has too much noise and the quality is unqualified.
  • a waveform template can be established according to the normal waveform (for example, a pressure template can be established according to multiple normal pressure waveforms), and the respiratory waveform data scheduled to be input into the trained recognition model is compared with its corresponding waveform template, and the whole cycle The degree of difference is calculated, and the quality is judged by the threshold. For example, when the degree of difference is greater than the preset threshold, it is judged that the quality of the respiratory waveform data is unqualified; when the degree of difference is lower than or equal to the preset threshold, the quality of the respiratory waveform data is judged Quality standards.
  • the quality of part of the basic respiratory waveform data in the multiple basic respiratory waveform data and part of the auxiliary waveform data in the multiple auxiliary waveform data can also be input into the trained recognition model for recognition.
  • the respiratory waveform data and the patient data are identified based on a trained identification model, and the identification result of the current ventilation state of the target object is determined, including:
  • the respiratory waveform data of the target object is input to the neural network model that has been trained, to obtain the network output result (the network output result can be the feature that extracts the output in the hidden layer of the neural network model); the patient of the target object
  • the data is input into the trained neural network model, and the patient data of the target object is obtained through transformation mapping to obtain a parameter matrix.
  • transformation mapping may include but not limited to fully connected layer mapping, Fourier transform (this paper Also known as F transformation) or other function transformations; in the trained neural network model, the parameter matrix and the network output result are subjected to information fusion to obtain the recognition result.
  • the information fusion includes But not limited to inner product.
  • the number of trained identification models may be one, for example, the trained identification model is a trained neural network model, and the processor 50 is also used to: use the basic respiratory waveform data and/or The auxiliary waveform data of the target object is fused to obtain fused data; the fused data and the patient data of the target object are input into the trained recognition model for recognition processing, and the recognition result of the current ventilation state of the target object is determined.
  • the trained identification model is a trained neural network model
  • the processor 50 is also used to: use the basic respiratory waveform data and/or The auxiliary waveform data of the target object is fused to obtain fused data; the fused data and the patient data of the target object are input into the trained recognition model for recognition processing, and the recognition result of the current ventilation state of the target object is determined.
  • different types and quantities of basic respiratory waveform data and/or different types of auxiliary waveform data can be combined to form multiple types of combined waveform data, for example, at least two types of basic respiratory waveform data A combination of waveform data, or a combination of at least one type of waveform data in the basic respiratory waveform data and at least one type of waveform data in the auxiliary waveform data, or a combination of at least two types of waveform data in the auxiliary waveform data.
  • Fusion processing refers to the fusion of each combination of waveform data (the combined waveform data includes at least two types of waveform data), such as size alignment, etc., and the purpose of fusion processing is to determine the final input to the trained Identify the format of the model.
  • the identification result of the current ventilation state includes but not limited to normal ventilation or abnormal ventilation events, wherein the abnormal ventilation events include one or more of the following events: invalid trigger, double inhalation, false trigger, too small flow rate , Reverse trigger, early switching, delayed switching, slow pressure rise, too fast pressure rise, endogenous positive end-expiratory pressure ventilation, short inspiratory time, long inspiratory time, water accumulation in the pipeline, excessive resistance, compliance Hypothyroidism, inverse ratio ventilation, etc., or other identifiable abnormal ventilation events can also be included.
  • the abnormal ventilation events include one or more of the following events: invalid trigger, double inhalation, false trigger, too small flow rate , Reverse trigger, early switching, delayed switching, slow pressure rise, too fast pressure rise, endogenous positive end-expiratory pressure ventilation, short inspiratory time, long inspiratory time, water accumulation in the pipeline, excessive resistance, compliance Hypothyroidism, inverse ratio ventilation, etc., or other identifiable abnormal ventilation events can also be included.
  • the processor 50 performs recognition processing on the respiratory waveform data of the target object and the patient data of the target object based on the trained recognition model, and determines the recognition result of the current ventilation state of the target object, including: based on the trained recognition model
  • the model performs recognition processing on the respiratory waveform data of the target object and the patient data of the target object, and determines the recognition result of the target object's current ventilation state and the corresponding reliability of the recognition result (also referred to as the confidence degree herein).
  • the reliability can give the user an assessment of the occurrence probability of the recognition result, thereby assisting the user in judging whether the recognition result is credible.
  • the number of trained recognition models can also be multiple.
  • the respiratory waveform data includes at least one of a variety of combined waveform data.
  • Each trained neural network model corresponds to a combined waveform data.
  • the combined waveform data includes one or more of the following combinations: at least one type of basic respiratory waveform data, at least one type of auxiliary waveform data and A combination of at least one type of basic respiratory waveform data
  • the basic respiratory waveform data includes one or more of the following waveform data: pressure, flow rate or volume
  • the auxiliary waveform data includes one or more of the following waveforms: transpulmonary Pressure, esophageal pressure, intragastric pressure or diaphragmatic electromyography
  • at least one type of basic respiratory waveform data can be pressure, flow rate or solvent, and can also include any combination of two types of pressure, flow rate and solvent, and a combination of three of them
  • at least The combination of one type of auxiliary waveform data and at least one type of basic respiratory waveform data may include the following combinations: a combination of one type of auxiliary waveform data and one type of basic respiratory waveform data (such as pressure and esophageal pressure, flow velocity and intragastric pressure, etc.), A combination of two types of auxiliary waveform
  • the processor 50 is also used to: obtain at least one combined waveform data based on the obtained basic respiratory waveform data and auxiliary waveform data, input various combined waveform data into their corresponding trained neural network models for identification, and
  • the patient data of the target object can also be input into each trained neural network model to separately identify and obtain a set of identification results, wherein the identification results can include normal ventilation or abnormal ventilation events, and correspond to the credibility of the identification results.
  • Spend. the user can obtain one or more recognition results, and for one or more recognition results that may be different or partly the same and partly different, the user can choose the credible result according to his own experience, or he can reliability to determine the patient's current ventilation status.
  • the identification result 1 includes but is not limited to invalid triggering, double inhalation, too small flow rate, endogenous PEEP, short inspiratory time, water accumulation in the circuit, excessive resistance or low compliance, etc.
  • the recognition result 2 and recognition result n include but are not limited to invalid trigger, double inhalation, false trigger, too small flow rate, reverse trigger, endogenous PEEP, short inspiratory time, water accumulation in the pipeline, excessive resistance and compliance Too small and so on.
  • the processor 50 performs recognition processing on the respiratory waveform data of the target object, and determines the recognition result of the current ventilation state of the target object, and further includes: performing recognition processing on the respiratory waveform data of the target object based on the trained recognition model. After the data and the patient data of the target object are identified and processed, the expert system is used to identify the respiratory waveform data and the patient data of the target object to determine the recognition result of the current ventilation state of the target object.
  • use The expert system performs identification processing (for example, using the knowledge base of the expert system for reasoning) according to the output value of the trained neural network model in combination with the respiratory waveform data and patient data, to determine the identification result of the patient's current ventilation state (the result can be Occurrence probability of each abnormal event), for another example, when the number of trained neural network models is multiple, use expert system to identify according to the output value of each trained neural network model combined with respiratory waveform data and patient data Processed to determine the recognition result of the patient's current ventilation status.
  • identification processing for example, using the knowledge base of the expert system for reasoning
  • the processor 50 performs identification processing on the respiratory waveform data of the target subject by using an expert system, including: acquiring patient data of the target subject (for example, acquiring patient data input through a human-computer interaction interface); Waveform data and patient data are identified and processed. For example, as shown in FIG. to the expert system, and use the expert system to identify the acquired patient data and respiratory waveform data, use the knowledge base in the expert system to reason, and combine the output results of the acquired neural network model to give the final identification result, and The final confidence level P1.
  • an expert system including: acquiring patient data of the target subject (for example, acquiring patient data input through a human-computer interaction interface); Waveform data and patient data are identified and processed. For example, as shown in FIG. to the expert system, and use the expert system to identify the acquired patient data and respiratory waveform data, use the knowledge base in the expert system to reason, and combine the output results of the acquired neural network model to give the final identification result, and The final confidence level P1.
  • the patient data may be input by the user through the human-computer interaction interface, or, the processor 50 may obtain the electronic medical record information of the target object; based on the electronic medical record information, obtain at least part of the patient data of the target object, or For some parameters such as mechanical ventilation parameters (such as parameters measured by ventilators or anesthesia machines), monitoring parameters (usually monitored by monitors), ultrasound parameters (usually measured by ultrasound equipment), etc., can be monitored from various parameters
  • the ultrasound parameters obtained by the equipment may also be that the ultrasound equipment communicates with the central station, the ventilator communicates with the central station, and the ventilator obtains the ultrasound parameters measured by the ultrasound equipment from the central station.
  • the patient data may include one or more of the following data: patient's medical record information, mechanical ventilation parameters, monitoring parameters, blood gas value parameters, ultrasound parameters, disease severity score, fluid balance information, etc.
  • the patient's Medical record information includes at least one of the following information: age, height, weight, BMI index (where BMI index is weight (kg) divided by height (m) squared), comorbidity information, genetic disease information, drug treatment information ( Such as antibiotic type), etc.
  • mechanical ventilation parameters include at least one of the following information: tidal volume (Vt), minute ventilation (MV), respiratory rate (RR), inspiratory time, PEEP, mean airway pressure, inspired oxygen Concentration (FiO2), pressure support level (PS level), maximum inspiratory pressure (MIP), maximum expiratory pressure (MEP), end-tidal carbon dioxide (ETCO2), lung compliance, airway resistance, mechanical energy, work of breathing, drive
  • the monitoring parameters include at least one of the following parameters: blood oxygen (SpO2), heart rate (HR), blood pressure (BP),
  • an expert system is a computer program designed to model the problem-solving capabilities of human experts. It is an intelligent computer program system, which contains the knowledge and experience of one or more experts in the field, and can use the knowledge and problem-solving methods of human experts to solve problems in this field.
  • the expert system is pre-stored in the memory, and the processor 50 to be used when needed.
  • the expert system in this article combines the professional knowledge, experience and problem-solving methods of many ICU clinicians to identify, judge and process the respiratory waveform, and further confirm the identification results of the neural network model, so that the obtained identification results can be obtained. It is more accepted and recognized by users, and makes the accuracy of recognition results higher.
  • the neural network model can be used alone, the expert system can be used alone, or both models can be used at the same time.
  • the processor 50 is used to: select the corresponding recognition mode according to the instruction input by the user; When the identification mode is the first identification mode, the respiratory waveform data of the target object and the patient data of the target object are identified based on the trained identification model, and the identification result of the current ventilation state of the target object is determined; when the identification mode is the second identification mode, use the expert system to identify and process the respiratory waveform data of the target object and the patient data of the target object, and determine the recognition result of the current ventilation state of the target object; when the recognition mode is the third recognition mode, use the trained recognition model and The expert system identifies and processes the respiratory waveform data of the target object and the patient data of the target object, and determines the recognition result of the current ventilation state of the target object. In this way, more choices can be provided to the user, so that the user can choose a suitable recognition mode according to his needs.
  • the processor 50 is further configured to: when it is determined that there is an abnormal ventilation event, output prompt information.
  • the display 70 is used to obtain and display the prompt information, which may include a text description of the event type of the abnormal ventilation event, or a simulated image prompt, etc., through the prompt information, the user can be intuitively prompted, so that The user obtains the abnormal ventilation events of the patient in a timely manner, and takes corresponding measures to deal with the abnormal ventilation events, thereby improving the breathing comfort of the patients.
  • the display 70 is used to display the prompt information in a preset display manner, and the preset display manner includes one or more of the following manners: highlighting, Additional symbol display, differentiated color, differentiated shading, flashing, or other suitable display methods for display.
  • a button is provided on the display interface of the display 70 of the medical device, and the processor 50 is used to control the display 70 to display or hide the prompt information when the user instruction input by the user through the button is obtained.
  • the processor 50 is used to control the display 70 to display or hide the prompt information when the user instruction input by the user through the button is obtained.
  • the medical device of the present application further includes an alarm device, and the medical device also includes an alarm device, which is used to give an alarm prompt when the number or frequency of abnormal ventilation events within a preset time meets a preset condition, for example, When the man-machine antagonism index exceeds the preset threshold within the preset time, an alarm is given, wherein the man-machine antagonism index is the ratio of the number of cycles of abnormal ventilation events to the total number of cycles within the preset time. When the preset condition is met, it indicates that the abnormal ventilation event is relatively serious, and it is likely to have a negative impact on the patient. Therefore, by means of an alarm prompt, the doctor's attention is immediately drawn to make reasonable treatment measures for the abnormal ventilation event .
  • a preset condition for example, When the man-machine antagonism index exceeds the preset threshold within the preset time, an alarm is given, wherein the man-machine antagonism index is the ratio of the number of cycles of abnormal ventilation events to the total number of cycles within the preset
  • the preset time can be reasonably set according to actual needs, for example, it can be 30s, 60s, 90s or any other suitable duration, or it can also be set based on the patient’s breathing cycle, for example, it can be is the time length corresponding to 10 breathing cycles of the patient, or the time length corresponding to any other integer multiple of breathing cycles.
  • the processor 50 may generate alarm information (such as an alarm signal) when the number or frequency of abnormal ventilation events within a preset time meets a preset condition.
  • the alarm device is configured to obtain the alarm information and give an alarm prompt.
  • the alarm mode of the alarm device includes but is not limited to light, sound and other alarm methods.
  • the specific form of the alarm device can be a flashing LED light, a buzzer, a speaker, etc., for example, When the alarm device is a loudspeaker, it can also ventilate the prompt sound of abnormal events, such as voice broadcast, etc., which meets the requirements for the strength of the alarm signal, and is enough to attract the attention and vigilance of observers. In this way, a real-time alarm can be realized to prompt the user.
  • the processor 50 is further configured to: when the recognition result is an abnormal ventilation event, according to the recognition result, automatically adjust the setting parameters of the ventilator for ventilating the target subject, such as trigger sensitivity, adjust tidal volume (Vt), Pressure rise time, minute ventilation (MV), respiratory rate (RR), inspiratory time, PEEP, average airway pressure, inspired oxygen concentration (FiO2), PS level (pressure support level), maximum inspiratory pressure (MIP) , maximum expiratory pressure (MEP) and other setting parameters, so that patients can quickly recover from abnormal ventilation events to normal ventilation, achieve better man-machine synchronization, and improve the safety and comfort of mechanical ventilation for patients.
  • the setting parameters of the ventilator for ventilating the target subject such as trigger sensitivity, adjust tidal volume (Vt), Pressure rise time, minute ventilation (MV), respiratory rate (RR), inspiratory time, PEEP, average airway pressure, inspired oxygen concentration (FiO2), PS level (pressure support level), maximum inspiratory pressure (MIP) , maximum expiratory pressure (MEP) and other
  • the processor 50 is further configured to: when the recognition result is an abnormal ventilation event, output adjustment suggestion information of the setting parameters of the ventilator for ventilating the target subject according to the recognition result, and the display 70 is used to obtain and display the adjustment Suggestion information, in this way, can give the user adjustment suggestion information to assist the user in handling abnormal ventilation events.
  • the processor 50 is further configured to: when the recognition result is an abnormal ventilation event, evaluate the degree of lung injury of the target subject according to the recognition result and the patient data of the target subject; and the display is used to The evaluation result of the lung injury degree is displayed, so that the user can judge the lung injury degree of the patient according to the evaluation result.
  • the lung injury degree of the target subject can be evaluated based on the recognition result and the tidal volume (for example, minute tidal volume) in the patient data of the target subject. Specifically, for example, when the recognition result is a double inspiratory event, the Minute tidal volume to predict the degree of lung injury.
  • the ventilation state identification method 700 of the present application includes the following steps S710 to S720:
  • step S710 current respiratory waveform data and patient data of the target subject are acquired, wherein the respiratory waveform data includes basic respiratory waveform data and/or auxiliary waveform data.
  • the respiratory waveform data includes basic respiratory waveform data and/or auxiliary waveform data.
  • step S720 the respiratory waveform data and the patient data are identified based on the trained identification model, and the identification result of the current ventilation state of the target object is determined, and the identification result includes normal ventilation or abnormal ventilation event.
  • the trained identification model is obtained through machine learning training.
  • the trained recognition model includes a trained neural network model, and the trained recognition model is obtained by using deep learning method training based on historical data information of previous patients' patient data and respiratory waveform data of the previous patients Yes, the historical data information of the respiratory waveform data of the previous patient includes the historical data information of the basic respiratory waveform data and/or the historical data information of the auxiliary waveform data.
  • performing recognition processing on the respiratory waveform data to determine the recognition result of the current ventilation state of the target object includes: performing preprocessing on the respiratory waveform data of the target object to obtain processed respiratory waveform data, wherein the preprocessing includes Normalization processing and/or filtering processing; performing recognition processing on the processed respiratory waveform data and patient data of the target object based on the trained recognition model, and determining the recognition result of the current ventilation state of the target object.
  • the recognition process is performed on the respiratory waveform data of the target subject based on the trained recognition model, and the recognition result of the current ventilation state of the target subject is determined, including: judging whether the quality of the respiratory waveform data meets a preset condition; When the quality of the respiratory waveform data of the target object satisfies a preset condition, input the respiratory waveform data of the target object and the patient data of the target object into the trained identification model for identification processing, and determine the target object The recognition result of the current ventilation state of the subject; when it is determined that the respiratory waveform data of the target subject does not meet the preset condition, the respiratory waveform data and the patient data are not input into the trained recognition model for recognition processing, A recognition result of the current ventilation state of the target object is determined.
  • the accuracy of the recognition result can be guaranteed through quality judgment, and the problem of outputting wrong results or not outputting results due to unqualified quality of respiratory waveform data can be avoided.
  • performing identification processing on the respiratory waveform data based on the trained identification model, and determining the identification result of the current ventilation state of the target object includes: performing basic respiratory waveform data of the target object and/or auxiliary waveform data of the target object Fusion processing to obtain fusion data; input the fusion data and patient data into the trained recognition model for recognition processing, and determine the recognition result of the current ventilation state of the target object.
  • the recognition process is performed on the respiratory waveform data and patient data of the target object based on the trained recognition model, and the recognition result of the current ventilation state of the target object is determined, including: analyzing the respiratory waveform data of the target object based on the trained recognition model The data and patient data are identified and processed to determine the identification result of the current ventilation state of the target object and the corresponding reliability of the identification result.
  • performing recognition processing on the respiratory waveform data to determine the recognition result of the target subject's current ventilation state includes: after performing recognition processing on the target subject's respiratory waveform data and patient data based on a trained recognition model, using an expert system Perform recognition processing on the respiratory waveform data and patient data of the target object, and determine the recognition result of the current ventilation state of the target object.
  • the method of the present application further includes: when it is determined that there is an abnormal ventilation event, outputting prompt information, and displaying the prompt information through a display.
  • the method of the present application further includes: giving an alarm prompt when the number or frequency of abnormal ventilation events within a preset time meets a preset condition.
  • the alarm prompt can be performed based on the aforementioned alarm device.
  • the method of the present application further includes: when the recognition result is an abnormal ventilation event, automatically adjusting the setting parameters of the ventilator for ventilating the target subject according to the recognition result.
  • the method of the present application further includes: when the recognition result is an abnormal ventilation event, evaluating the lung injury degree of the target subject according to the recognition result and the patient data of the target subject; The evaluation results of the degree of lung injury described above.
  • the degree of lung injury of the target subject is evaluated based on the identification result (eg double inspiratory event) and the tidal volume in the patient data of the target subject.
  • the basic waveform and auxiliary waveform data of the target object can be recognized and processed based on the trained recognition model, and the recognition result of the current ventilation state of the target object can be determined. Therefore, the human-machine synchronization can be improved.
  • the accuracy of identifying abnormal ventilation events can assist doctors to make judgments on man-machine confrontation events, thereby reducing the occurrence of man-machine confrontation and improving the breathing comfort and safety of mechanically ventilated patients.
  • an embodiment of the present invention also provides a computer storage medium on which a computer program is stored.
  • One or more computer program instructions can be stored on the computer-readable storage medium, and the processor can execute the program instructions stored in the storage device to realize the functions (implemented by the processor) and/or other desired functions in the embodiments of the present invention herein. function, for example, to execute the corresponding steps of the ventilation state identification method according to the embodiment of the present invention, various application programs and various data may also be stored in the computer-readable storage medium, such as various application programs used and/or generated data etc.
  • computer storage media may include, for example, a memory card of a smartphone, a memory component of a tablet computer, a hard disk of a personal computer, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disk read-only Memory (CD-ROM), USB memory, or any combination of the above storage media.
  • ROM read-only memory
  • EPROM erasable programmable read-only memory
  • CD-ROM portable compact disk read-only Memory
  • USB memory or any combination of the above storage media.
  • the disclosed devices and methods may be implemented in other ways.
  • the device embodiments described above are only illustrative.
  • the division of the units is only a logical function division. In actual implementation, there may be other division methods.
  • multiple units or components can be combined or May be integrated into another device, or some features may be omitted, or not implemented.
  • the various component embodiments of the present invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof.
  • a microprocessor or a digital signal processor (DSP) may be used in practice to implement some or all functions of some modules according to the embodiments of the present invention.
  • DSP digital signal processor
  • the present invention can also be implemented as an apparatus program (for example, a computer program and a computer program product) for performing a part or all of the methods described herein.
  • Such a program for realizing the present invention may be stored on a computer-readable medium, or may be in the form of one or more signals.
  • Such a signal may be downloaded from an Internet site, or provided on a carrier signal, or provided in any other form.

Landscapes

  • Health & Medical Sciences (AREA)
  • Emergency Medicine (AREA)
  • Pulmonology (AREA)
  • Engineering & Computer Science (AREA)
  • Anesthesiology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Hematology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

Dispositif médical (100) et procédé de reconnaissance d'état de ventilation. Le dispositif médical (100) comprend : une mémoire (60), conçue pour stocker des instructions de programme exécutables ; et un processeur (50), conçu pour exécuter les instructions de programme stockées dans la mémoire (60), de telle sorte que le processeur (50) exécute les étapes suivantes consistant : à obtenir des données de forme d'onde respiratoire actuelle d'un objet cible (S910), les données de forme d'onde respiratoire comprenant des données de forme d'onde respiratoire de base et des données de forme d'onde auxiliaire ; et à effectuer un traitement de reconnaissance sur les données de forme d'onde respiratoire de l'objet cible sur la base d'un modèle de reconnaissance entraîné, et à déterminer un résultat de reconnaissance pour un état de ventilation actuel de l'objet cible (S920), le résultat de la reconnaissance comprenant un événement de ventilation normal ou un événement de ventilation anormal. Par conséquent, la précision de reconnaissance de l'événement de ventilation anormal vis-à-vis d'une synchronisation de ventilateur de patient peut être améliorée, des médecins sont assistés lors de la détermination d'un événement d'asynchronie patient-ventilateur, l'apparition d'une asynchronie patient-ventilateur est ainsi réduite, et le confort respiratoire et la sécurité d'un patient pendant une ventilation mécanique sont améliorés.
PCT/CN2021/136756 2021-12-09 2021-12-09 Dispositif médical et procédé de reconnaissance d'état de ventilation WO2023102820A1 (fr)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN202180103623.XA CN118159322A (zh) 2021-12-09 2021-12-09 医疗设备及通气状态识别方法
PCT/CN2021/136756 WO2023102820A1 (fr) 2021-12-09 2021-12-09 Dispositif médical et procédé de reconnaissance d'état de ventilation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2021/136756 WO2023102820A1 (fr) 2021-12-09 2021-12-09 Dispositif médical et procédé de reconnaissance d'état de ventilation

Publications (1)

Publication Number Publication Date
WO2023102820A1 true WO2023102820A1 (fr) 2023-06-15

Family

ID=86729397

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/136756 WO2023102820A1 (fr) 2021-12-09 2021-12-09 Dispositif médical et procédé de reconnaissance d'état de ventilation

Country Status (2)

Country Link
CN (1) CN118159322A (fr)
WO (1) WO2023102820A1 (fr)

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140276173A1 (en) * 2013-03-14 2014-09-18 Convergent Engineering, Inc. Methods and systems for monitoring resistance and work of breathing for ventilator-dependent patients
US20190015614A1 (en) * 2017-07-13 2019-01-17 Royal Commission Yanbu Colleges & Institutes Systems, devices, and methodologies to provide protective and personalized ventilation
CN109893732A (zh) * 2019-02-28 2019-06-18 杭州智瑞思科技有限公司 一种基于循环神经网络的机械通气人机不同步检测方法
CN110189822A (zh) * 2019-05-23 2019-08-30 广州博而济信息科技有限公司 一种基于机械通气参数及波形进行智能分析的方法
CN111563451A (zh) * 2020-05-06 2020-08-21 浙江工业大学 基于多尺度小波特征的机械通气无效吸气努力识别方法
CN111738305A (zh) * 2020-05-29 2020-10-02 浙江大学 一种基于dba-dtw-knn的机械通气人机不同步快速识别方法
CN112449607A (zh) * 2018-08-21 2021-03-05 深圳迈瑞生物医疗电子股份有限公司 通气触发检测方法、装置、通气设备及存储介质
CN113332548A (zh) * 2021-04-16 2021-09-03 深圳先进技术研究院 呼吸机通气模式的调节方法、装置及计算机可读存储介质
CN113642512A (zh) * 2021-08-30 2021-11-12 深圳先进技术研究院 呼吸机人机异步检测方法、装置、设备及存储介质

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140276173A1 (en) * 2013-03-14 2014-09-18 Convergent Engineering, Inc. Methods and systems for monitoring resistance and work of breathing for ventilator-dependent patients
US20190015614A1 (en) * 2017-07-13 2019-01-17 Royal Commission Yanbu Colleges & Institutes Systems, devices, and methodologies to provide protective and personalized ventilation
CN112449607A (zh) * 2018-08-21 2021-03-05 深圳迈瑞生物医疗电子股份有限公司 通气触发检测方法、装置、通气设备及存储介质
CN109893732A (zh) * 2019-02-28 2019-06-18 杭州智瑞思科技有限公司 一种基于循环神经网络的机械通气人机不同步检测方法
CN110189822A (zh) * 2019-05-23 2019-08-30 广州博而济信息科技有限公司 一种基于机械通气参数及波形进行智能分析的方法
CN111563451A (zh) * 2020-05-06 2020-08-21 浙江工业大学 基于多尺度小波特征的机械通气无效吸气努力识别方法
CN111738305A (zh) * 2020-05-29 2020-10-02 浙江大学 一种基于dba-dtw-knn的机械通气人机不同步快速识别方法
CN113332548A (zh) * 2021-04-16 2021-09-03 深圳先进技术研究院 呼吸机通气模式的调节方法、装置及计算机可读存储介质
CN113642512A (zh) * 2021-08-30 2021-11-12 深圳先进技术研究院 呼吸机人机异步检测方法、装置、设备及存储介质

Also Published As

Publication number Publication date
CN118159322A (zh) 2024-06-07

Similar Documents

Publication Publication Date Title
US20190015614A1 (en) Systems, devices, and methodologies to provide protective and personalized ventilation
US11247009B2 (en) Anomaly detection device and method for respiratory mechanics parameter estimation
CN102770070B (zh) 用于实时评估肺力学的系统和方法
CN101874734B (zh) 检测和治疗呼吸功能不全的装置
US8579829B2 (en) System and method for monitoring breathing
US20120096381A1 (en) Ventilator-Initiated Prompt In Response To Proposed Setting Adjustment
WO2013016608A1 (fr) Procédés et systèmes pour une ventilation assistée proportionnelle transformée basée sur un modèle
CN110753564B (zh) 呼吸机和呼吸机的通气控制方法
EP3148426B1 (fr) Procédé, système et logiciel pour l'évaluation d'un échec d'extubation
US9295803B2 (en) Ventilator sychronization indicator
CN105031784B (zh) 呼吸器的警报方法及呼吸器警报系统
US20140150795A1 (en) System and method for detecting double triggering with remote monitoring
WO2023102820A1 (fr) Dispositif médical et procédé de reconnaissance d'état de ventilation
US20220198219A1 (en) Machine-learning image recognition for classifying conditions based on ventilatory data
US20210393902A1 (en) One-touch ventilation mode
WO2022141125A1 (fr) Dispositif d'assistance respiratoire et son procédé de commande, et support de stockage
CA2651287C (fr) Systeme d'aide au sevrage et a la decision pour ventilation mecanique
WO2023230957A1 (fr) Procédé d'analyse de ventilation, dispositif médical, station centrale et système médical
US20230094608A1 (en) Pendelluft detection by acoustic interferometry through an endotracheal tube
EP4268869A1 (fr) Dispositif de ventilation et son procédé d'affichage
WO2022133942A1 (fr) Dispositif de ventilation medicale et procédé de commande de ventilation
CN111093745A (zh) 一种通气过程中的状态显示方法及装置
WO2023077448A1 (fr) Procédé de surveillance de la fréquence respiratoire et appareil de ventilation médicale
WO2023115531A1 (fr) Procédé de surveillance de respiration et appareil de surveillance de respiration
EP4264624A1 (fr) Reconnaissance d'image par apprentissage machine pour la classification d'états sur la base de données de ventilation

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21966743

Country of ref document: EP

Kind code of ref document: A1