Automatic control methods, devices, computer equipment, readable storage media, and program products for ventilators

Abstract

This application relates to an automatic control method, apparatus, computer device, computer-readable storage medium, and computer program product for a ventilator. The method includes: acquiring ambient sound monitoring data during ventilator operation; the ambient sound monitoring data includes snoring monitoring data and non-snoring monitoring data; detecting snoring events in the ambient sound monitoring data; when a target snoring event indicating an impending respiratory abnormality is detected, determining a target pressure adjustment scheme matching the target snoring event, and controlling the ventilator to increase pressure according to the target pressure adjustment scheme. This method enables timely and accurate control of the ventilator to increase pressure according to the corresponding target pressure adjustment scheme when an impending respiratory abnormality is detected, rather than adjusting after the respiratory abnormality occurs. Therefore, it can adjust the ventilator pressure in a timely and accurate manner, thereby ensuring the ventilator's operational effectiveness and preventing respiratory abnormalities in the ventilator user.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to an automatic control method, apparatus, computer equipment, computer-readable storage medium, and computer program product for a ventilator. Background Technology

[0002] A ventilator is an important medical device used to assist users in maintaining normal respiratory function. Traditional technology allows for automatic pressure adjustment of the ventilator by setting a target tidal volume. However, this control method can lead to excessive pressure during automatic adjustment, causing user discomfort. Traditional technology can also detect abnormal breathing and adjust the ventilator pressure accordingly, but this method results in a delay in pressure adjustment.

[0003] Therefore, traditional automatic control methods for ventilators cannot adjust the ventilator pressure in a timely and accurate manner, which makes it impossible to ensure the effectiveness of ventilator operation. Summary of the Invention

[0004] Therefore, it is necessary to provide an automatic control method, device, computer equipment, computer-readable storage medium, and computer program product for ventilators that can adjust ventilator pressure in a timely and accurate manner to ensure the effectiveness of ventilator operation, in response to the above-mentioned technical problems.

[0005] In a first aspect, this application provides an automatic control method for a ventilator, comprising:

[0006] Acquire ambient sound monitoring data during ventilator operation; the ambient sound monitoring data includes snoring monitoring data and non-snoring monitoring data;

[0007] Detect snoring events using environmental sound monitoring data;

[0008] When a target snoring event that is a precursor to respiratory abnormality is detected, a target pressure adjustment scheme that matches the target snoring event is determined, and the ventilator is controlled to increase pressure according to the target pressure adjustment scheme.

[0009] In one embodiment, snoring event detection is performed on ambient sound monitoring data, including:

[0010] Obtain a snoring event detection model;

[0011] Based on the snoring event detection model, snoring events are detected using environmental sound monitoring data.

[0012] In one embodiment, the snoring event detection model is generated in the following manner:

[0013] Acquire multiple historical ambient sound monitoring data; the historical ambient sound monitoring data carries annotations of historical target snoring events;

[0014] The dataset is divided into multiple historical environmental sound monitoring data according to the first ratio, resulting in multiple subsets of historical environmental sound monitoring data.

[0015] Based on multiple subsets of historical ambient sound monitoring data, the initial snoring event detection model is iteratively optimized until the first iteration optimization stopping condition is met, thus obtaining the snoring event detection model.

[0016] In one embodiment, determining a target pressure adjustment scheme that matches the target snoring event includes:

[0017] Obtain the ventilator pressure adjustment model;

[0018] Extract the target snoring monitoring data corresponding to the target snoring event from the ambient sound monitoring data;

[0019] The target snoring monitoring data is input into the ventilator pressure adjustment model. The ventilator pressure adjustment model is used to analyze the pressure adjustment scheme of the target snoring monitoring data to determine the target pressure adjustment scheme that matches the target snoring event.

[0020] In one embodiment, the ventilator pressure adjustment model is generated in the following manner:

[0021] Acquire ventilator pressure adjustment data for each historical target snoring event in multiple historical environmental sound monitoring data sets;

[0022] For each historical target snoring event, establish the correlation between the historical target snoring event and the ventilator pressure adjustment data of the historical target snoring event, and generate a pressure adjustment scheme group;

[0023] The dataset is divided into multiple stress adjustment scheme groups according to the second ratio to obtain multiple subsets of stress adjustment scheme groups;

[0024] Based on multiple subsets of pressure adjustment schemes, the initial ventilator pressure adjustment model is iteratively optimized until the second iteration stop condition is met, thus obtaining the ventilator pressure adjustment model.

[0025] In one embodiment, the ventilator automatic control method further includes:

[0026] When the ventilator mask is detected to be removed, or when the duration of the event-free segment in the ambient sound monitoring data has reached the preset duration, the ventilator pressure is controlled to decrease.

[0027] If a target snoring event is detected from the latest ambient sound monitoring data after the ventilator has been depressurized, the ventilator will automatically increase its pressure by executing the above-mentioned automatic control method.

[0028] Secondly, this application also provides an automatic control device for a ventilator, comprising:

[0029] The monitoring data acquisition module is used to acquire ambient sound monitoring data during the operation of the ventilator; the ambient sound monitoring data includes snoring monitoring data and non-snoring monitoring data;

[0030] The snoring event detection module is used to detect snoring events from environmental sound monitoring data.

[0031] The automatic control module of the ventilator is used to determine the target pressure adjustment scheme that matches the target snoring event when a target snoring event that is a precursor to respiratory abnormality is detected, and to control the ventilator to increase the pressure according to the target pressure adjustment scheme.

[0032] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0033] Acquire ambient sound monitoring data during ventilator operation; the ambient sound monitoring data includes snoring monitoring data and non-snoring monitoring data;

[0034] Detect snoring events using environmental sound monitoring data;

[0035] When a target snoring event that is a precursor to respiratory abnormality is detected, a target pressure adjustment scheme that matches the target snoring event is determined, and the ventilator is controlled to increase pressure according to the target pressure adjustment scheme.

[0036] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0037] Acquire ambient sound monitoring data during ventilator operation; the ambient sound monitoring data includes snoring monitoring data and non-snoring monitoring data;

[0038] Detect snoring events using environmental sound monitoring data;

[0039] When a target snoring event that is a precursor to respiratory abnormality is detected, a target pressure adjustment scheme that matches the target snoring event is determined, and the ventilator is controlled to increase pressure according to the target pressure adjustment scheme.

[0040] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0041] Acquire ambient sound monitoring data during ventilator operation; the ambient sound monitoring data includes snoring monitoring data and non-snoring monitoring data;

[0042] Detect snoring events using environmental sound monitoring data;

[0043] When a target snoring event that is a precursor to respiratory abnormality is detected, a target pressure adjustment scheme that matches the target snoring event is determined, and the ventilator is controlled to increase pressure according to the target pressure adjustment scheme.

[0044] The aforementioned automatic control method, device, computer equipment, computer-readable storage medium, and computer program product for ventilators first acquires environmental sound monitoring data during ventilator operation. This environmental sound monitoring data includes snoring monitoring data and non-snoring monitoring data. Further, snoring event detection is performed on the environmental sound monitoring data. When a target snoring event indicative of an impending respiratory abnormality is detected, a target pressure adjustment scheme matching the target snoring event is determined, and the ventilator is controlled to increase pressure according to the target pressure adjustment scheme. Throughout the control process, by detecting environmental sound monitoring data, it is possible to determine whether a target snoring event indicative of an impending respiratory abnormality has occurred, and to specifically determine a ventilator pressure adjustment scheme matching the impending respiratory abnormality. This allows for timely and accurate pressure adjustment of the ventilator according to the corresponding ventilator pressure adjustment scheme when an impending respiratory abnormality is detected, rather than adjusting after a respiratory abnormality occurs, and avoiding discomfort to the ventilator user due to a mismatch in pressure adjustment schemes. Therefore, timely and accurate adjustment of the ventilator pressure ensures the effective operation of the ventilator. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is an application environment diagram of the automatic control method for a ventilator in one embodiment;

[0047] Figure 2 This is a flowchart illustrating an automatic control method for a ventilator in one embodiment;

[0048] Figure 3 This is a schematic diagram of the process for implementing automatic control of a ventilator based on a snoring event detection model in one embodiment;

[0049] Figure 4This is a flowchart illustrating the process of generating a snoring event detection model in one embodiment;

[0050] Figure 5 This is a flowchart illustrating the process of generating a snoring event detection model in another embodiment;

[0051] Figure 6 This is a schematic diagram of the process for implementing automatic ventilator pressure control based on a ventilator pressure control model in one embodiment;

[0052] Figure 7 This is a schematic diagram of the process for generating a ventilator pressure adjustment model in one embodiment;

[0053] Figure 8 This is a schematic diagram of the process for generating a ventilator pressure adjustment model in another embodiment;

[0054] Figure 9 This is a schematic diagram of a process in one embodiment where the ventilator automatically depressurizes and then automatically repressurizes based on the latest ambient sound monitoring data.

[0055] Figure 10 This is a comparison chart of the ventilator control effects in one embodiment;

[0056] Figure 11 This is a comparison diagram of the ventilator control effect in another embodiment;

[0057] Figure 12 This is a structural block diagram of an automatic control device for a ventilator in one embodiment;

[0058] Figure 13 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0060] The automatic control method for ventilators provided in this application can be applied to, for example... Figure 1The application environment is shown. The ventilator 100 is equipped with a controller 102 and an ambient sound acquisition device 104 connected to the controller 102. The controller 102 of the ventilator 100 can be used to receive, process, and analyze data in real time, and control the operation of the ventilator 100 based on the data analysis results. The controller 102 can integrate multiple control algorithms / data processing models, enabling autonomous and intelligent operation. The ambient sound acquisition device 104 is a device or apparatus used to capture and record snoring sounds produced by the human body during sleep, as well as other sounds around the ventilator 100. It has high sensitivity and can accurately detect and record ambient sounds, providing support for subsequent data processing and analysis by the controller 102. The ambient sound acquisition device 104 can be equipped with multiple microphones / sound sensors for collecting ambient sounds.

[0061] Specifically, in Figure 1 In the illustrated application environment, the ambient sound acquisition device 104 can transmit real-time acquired ambient sound monitoring data to the controller 102. This ambient sound monitoring data includes snoring monitoring data and non-snoring monitoring data. Furthermore, the controller 102 can perform snoring event detection on the ambient sound monitoring data. When a target snoring event, representing a precursor to respiratory abnormality, is detected, the controller 102 can determine a target pressure adjustment scheme matching the target snoring event and control the ventilator 100 to increase pressure according to the target pressure adjustment scheme, so that the user of the ventilator 100 will not experience respiratory abnormalities.

[0062] In one exemplary embodiment, such as Figure 2 As shown, an automatic control method for a ventilator is provided, which can be applied to... Figure 1 Taking controller 102 as an example, the explanation includes the following steps 202 to 206. Wherein:

[0063] Step 202: Obtain ambient sound monitoring data during ventilator operation; the ambient sound monitoring data includes snoring monitoring data and non-snoring monitoring data.

[0064] Among them, snoring monitoring data refers to monitoring data that includes snoring sounds, while non-snoring monitoring data includes, but is not limited to, monitoring data of various external interference sounds such as the motor sound of the ventilator and the sound of people talking.

[0065] Optionally, during the operation of the ventilator, the ambient sound acquisition device in the ventilator can collect ambient sounds around the ventilator in real time, such as the snoring of the ventilator user and other external sounds around the ventilator, thereby generating ambient sound monitoring data and transmitting the ambient sound monitoring data to the controller.

[0066] Step 204: Detect snoring events from the ambient sound monitoring data.

[0067] Optionally, the controller can use a pre-deployed snoring event detection model to perform real-time snoring event detection on the ambient sound monitoring data, determining whether a target snoring event exists in the data. The snoring event detection model can be built on a neural network model, capable of learning how to detect target snoring events from complex and interfering ambient sound monitoring data. A neural network model is a computational model that simulates the structure and function of a biological nervous system. It consists of a large number of interconnected nodes (neurons). Each node receives input signals from other nodes, processes these signals through an activation function, and then outputs the signals to the next layer of nodes. The neural network model can learn the mapping relationship between input and output by adjusting the connection weights between nodes.

[0068] Step 206: When a target snoring event that is a precursor to respiratory abnormality is detected, a target pressure adjustment scheme that matches the target snoring event is determined, and the ventilator is controlled to increase pressure according to the target pressure adjustment scheme.

[0069] Specifically, abnormal breathing refers to hypoventilation or apnea caused by airway collapse. Snoring is produced by the vibration of the airway during breathing and is closely related to human respiration. It often precedes abnormal breathing. Therefore, detecting snoring can help identify early signs of abnormal breathing, allowing for timely adjustment of the ventilator pressure before these abnormalities occur. This helps the ventilator user maintain normal breathing and ensures the ventilator's effectiveness. Pressure adjustment methods include, but are not limited to, the rate of pressure increase and the percentage / volume of pressure increase required.

[0070] Optionally, when a target snoring event, representing a precursor to respiratory abnormality, is detected from snoring monitoring data carried by environmental sound monitoring data, the controller can analyze the target snoring event through a pre-deployed ventilator pressure adjustment model to determine a target pressure adjustment scheme matching the target snoring event. This allows for timely and accurate control of the ventilator to increase pressure according to the target pressure adjustment scheme. The ventilator pressure adjustment model can be built based on a neural network model, capable of learning the mapping relationship between the precursor to respiratory abnormality (target snoring event) and the corresponding ventilator pressure adjustment scheme.

[0071] In the aforementioned automatic control method for ventilators, environmental sound monitoring data during ventilator operation is first collected. This data includes snoring and non-snoring monitoring data. Further, snoring event detection is performed on the environmental sound monitoring data. When a target snoring event indicating an impending respiratory abnormality is detected, a target pressure adjustment scheme matching the target snoring event is determined, and the ventilator is controlled to increase pressure according to the target pressure adjustment scheme. Throughout the control process, by detecting environmental sound monitoring data, it is possible to determine whether a target snoring event indicating an impending respiratory abnormality has occurred, and to specifically determine a ventilator pressure adjustment scheme matching the impending respiratory abnormality. This allows for timely and accurate pressure adjustment of the ventilator according to the corresponding ventilator pressure adjustment scheme when an impending respiratory abnormality is detected, rather than adjusting after the respiratory abnormality occurs. It also prevents discomfort to the ventilator user due to mismatched pressure adjustment schemes. Therefore, timely and accurate adjustment of ventilator pressure ensures the effective operation of the ventilator.

[0072] In one exemplary embodiment, in Figure 2 On the basis of, such as Figure 3 As shown, a flowchart illustrating automatic ventilator control based on a snoring event detection model is provided, mainly including the following steps:

[0073] Step 302: Obtain the snoring event detection model.

[0074] Optionally, the controller can first acquire a snoring event detection model pre-deployed on the controller. This snoring event detection model can be built based on a neural network model, capable of learning how to detect target snoring events representing precursors to breathing abnormalities from complex and interfering environmental sound monitoring data by using historical environmental sound monitoring data pre-labeled with snoring events representing precursors to breathing abnormalities. Neural networks that can be used in this embodiment include, but are not limited to: Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN) and their variants (such as LSTM, GRU), Deep Neural Networks (DNN), etc.

[0075] Step 304: Based on the snoring event detection model, perform snoring event detection on the ambient sound monitoring data.

[0076] Optionally, the controller can use a snoring event detection model to detect snoring events in the real-time collected ambient sound monitoring data, so as to identify whether the snoring monitoring data carried by the ambient sound monitoring data has a target snoring event that represents a precursor to abnormal breathing.

[0077] In this embodiment, a pre-trained snoring event detection model can be used to identify in real time and accurately whether the current user of the ventilator is experiencing any signs of respiratory abnormality. This allows the ventilator pressure to be adjusted promptly when signs of respiratory abnormality occur, thus helping the user improve their breathing.

[0078] In one embodiment, such as Figure 4 As shown, the snoring event detection model can be generated through the following steps:

[0079] Step 402: Obtain multiple historical ambient sound monitoring data; the historical ambient sound monitoring data carries annotations of historical target snoring events.

[0080] Multiple historical ambient sound monitoring data points can originate from multiple users, specifically from ambient sound monitoring data generated by each user at different times in the past. Historical target snoring events specifically refer to snoring sounds carried in historical ambient sound monitoring data that indicate precursors to abnormal breathing.

[0081] Optionally, during the collection of historical ambient sound monitoring data for each user, the user of the ventilator also needs to wear a polysomnography device. This allows professionals with relevant knowledge to annotate the historical snoring monitoring data carried by the historical ambient sound monitoring data with historical target snoring events that indicate impending respiratory abnormalities, based on the physiological data monitored by the polysomnography device. The annotated historical ambient sound monitoring data is then transmitted to the controller. Alternatively, the controller can automatically annotate the historical snoring monitoring data carried by the historical ambient sound monitoring data with historical target snoring events according to a preset program. Based on this, the controller can obtain multiple historical ambient sound monitoring data points with annotations of historical target snoring events.

[0082] Polysomnography (PSG) is a widely used professional device in the field of sleep medicine. It records various physiological indicators of the monitored subject during sleep by connecting multiple electrodes. For example, by connecting electrodes to the scalp, face, and other parts of the body, it monitors and records physiological signals such as electroencephalogram (EEG), electrooculogram (EOG), and electromyography (EMG). Simultaneously, it monitors airflow through the mouth and nose, blood oxygen saturation, and chest and abdominal respiratory movements to comprehensively assess the subject's sleep status. These signals and data also help identify the presence of sleep disorders such as insomnia, hypersomnia, sleep apnea, and hypopnea. In this embodiment, the physiological data monitored by the PSG helps the operator / controller identify abnormal breathing precursors (i.e., snoring that precedes apnea / hypopnea) in the snoring produced by the ventilator user during sleep, thereby assisting the operator / controller in labeling historical target snoring events in the collected historical environmental sound monitoring data.

[0083] Step 404: Divide the dataset of multiple historical ambient sound monitoring data according to the first ratio to obtain multiple subsets of historical ambient sound monitoring data.

[0084] The first ratio refers to the proportion of dividing multiple historical environmental sound monitoring data into training set, validation set, and test set. In this embodiment, the proportion of dividing the dataset can be flexibly configured according to the actual application scenario, including but not limited to: training set: validation set: test set = 7:1:2.

[0085] Optionally, after preprocessing multiple historical ambient sound monitoring data sets, the controller can divide the dataset according to a first ratio to obtain multiple subsets of historical ambient sound monitoring data. These subsets include a training set (as one subset), a validation set (as another subset), and a test set (as yet another subset). Data preprocessing includes, but is not limited to, data cleaning (noise removal, missing value imputation, etc.) to improve model performance and training efficiency.

[0086] Step 406: Based on multiple subsets of historical ambient sound monitoring data, iteratively optimize the initial snoring event detection model until the first iteration optimization stopping condition is met, thus obtaining the snoring event detection model.

[0087] Optionally, after dividing the dataset of multiple historical ambient sound monitoring data points according to a first ratio, the controller can input the training set into the initial snoring event detection model, calculate the output of each layer in the model, and use a pre-set loss function (such as mean squared error, cross-entropy, etc.) to calculate the error between the model output and the actual result. Based on the loss function, the gradient of the parameters of each layer in the model is calculated, and the gradient is backpropagated to the previous layer of the network. Then, gradient descent or other optimization algorithms are used to update the weights and biases of each layer of the neural network in the model. When the first iteration stopping condition is met, the optimization stops. The controller can repeat the above process multiple times, and in each repetition, different neural network models can be selected for training, resulting in multiple trained models. A validation set is then used to evaluate the multiple trained models to select the best model. Finally, the controller can also perform model performance testing on the selected model based on a test set, and the model whose test results meet expectations is taken as the final snoring event detection model.

[0088] Based on this, such as Figure 5 As shown, a flowchart illustrating another method for generating a snoring event detection model is provided, which mainly includes the following steps:

[0089] Step 502: When the user wears both a polysomnography device and a ventilator at the same time, acquire the physiological data monitored by the polysomnography device and the historical environmental sound monitoring data collected by the ventilator.

[0090] Step 504: Based on the physiological data monitored by the polysomnography device, label the historical environmental sound monitoring data collected by the ventilator with historical target snoring events that characterize the precursors of respiratory abnormalities, and obtain multiple historical environmental sound monitoring data with labels;

[0091] Step 506: Divide the labeled historical environmental sound monitoring data into a dataset to obtain a training set, a validation set, and a test set;

[0092] Step 508: Train the initial snoring event detection model based on the training set;

[0093] Step 510: Select the best model from the trained models based on the validation set, and use the test set to test the model performance of the selected best model to determine the final snoring event detection model.

[0094] For example, after training a snoring event detection model using the controllers of one or more ventilators, the snoring event detection model can be integrated into the controller of each ventilator, so that subsequent ventilators can process real-time collected environmental sound monitoring data based on the snoring event detection model.

[0095] In this embodiment, the optimal snoring event detection model can be obtained through iterative training, selection, and performance testing of the model. This ensures that all ventilators can process real-time ambient sound monitoring data based on the snoring event detection model integrated in the controller, thereby accurately determining whether the current user of the ventilator is experiencing any signs of respiratory abnormality. This facilitates timely and accurate adjustment of the ventilator pressure when signs of respiratory abnormality occur, ensuring the ventilator's operational effectiveness.

[0096] In one possible implementation, Figure 3 On the basis of, such as Figure 6 As shown, a flowchart illustrating automatic ventilator pressure control based on a ventilator pressure control model is provided, mainly including the following steps:

[0097] Step 602: When a target snoring event characterizing a precursor to respiratory abnormality is detected, the ventilator pressure adjustment model is obtained.

[0098] Optionally, when a target snoring event representing a precursor to respiratory abnormality is detected, the controller can first acquire (call) a ventilator pressure adjustment model pre-deployed on the controller. This ventilator pressure adjustment model is built on a neural network and can learn the mapping relationship between the precursor to respiratory abnormality and the corresponding ventilator pressure adjustment scheme. This allows it to be used to determine the target pressure adjustment scheme corresponding to the real-time detected target snoring event. Neural networks that can be used in this embodiment include, but are not limited to: Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN) and their variants (such as LSTM, GRU), Deep Neural Networks (DNN), etc.

[0099] Step 604: Extract the target snoring monitoring data corresponding to the target snoring event from the ambient sound monitoring data.

[0100] Optionally, when a target snoring event is detected from the real-time collected ambient sound monitoring data (carrying snoring monitoring data and non-snoring monitoring data) through the snoring event detection model, the controller can further extract the target snoring monitoring data corresponding to the target snoring event from the ambient sound monitoring data.

[0101] Step 606: Input the target snoring monitoring data into the ventilator pressure adjustment model, analyze the pressure adjustment scheme of the target snoring monitoring data through the ventilator pressure adjustment model, determine the target pressure adjustment scheme that matches the target snoring event, and control the ventilator to increase pressure according to the target pressure adjustment scheme.

[0102] Optionally, the controller can input the target snoring monitoring data carrying the target snoring event into the ventilator pressure adjustment model. The ventilator pressure adjustment model analyzes the pressure adjustment scheme of the target snoring monitoring data and determines the ventilator pressure adjustment scheme (target pressure adjustment scheme) that matches the respiratory abnormality precursor (target snoring event) carried by the target snoring monitoring data. Thus, before the respiratory abnormality occurs, the controller can control the ventilator to increase the pressure in a timely and accurate manner according to the target pressure adjustment scheme.

[0103] It should be noted that the target pressure adjustment scheme output by the ventilator pressure adjustment model also includes a pressure ramp rate that matches the target snoring event, so that the controller will not cause discomfort to the user during the pressure ramping process.

[0104] In this embodiment, a ventilator pressure adjustment scheme (target pressure adjustment scheme) that matches the respiratory abnormality precursors of the current ventilator user can be determined in real time and accurately using a trained ventilator pressure adjustment model. This allows the ventilator pressure to be adjusted promptly and accurately when respiratory abnormality precursors appear, thus helping the user improve breathing before respiratory abnormalities occur.

[0105] In one embodiment, such as Figure 7 As shown, the ventilator pressure adjustment model can be generated in the following ways:

[0106] Step 702: Obtain ventilator pressure adjustment data for each historical target snoring event in multiple historical ambient sound monitoring data sets.

[0107] Optionally, during the process of collecting each historical ambient sound monitoring data for each user, the controller will also synchronously store the ventilator pressure adjustment data corresponding to each historical target snoring event in each historical ambient sound monitoring data, such as the pressure increase rate and pressure increase ratio.

[0108] Step 704: For each historical target snoring event, establish the correlation between the historical target snoring event and the ventilator pressure adjustment data of the historical target snoring event, and generate a pressure adjustment scheme group.

[0109] Optionally, for each historical target snoring event, the controller will establish a correlation between the historical target snoring event and the ventilator pressure adjustment data of the historical target snoring event, and generate a pressure adjustment scheme group. This will facilitate the neural network model to learn the mapping relationship between the precursors of respiratory abnormalities and the corresponding ventilator pressure adjustment schemes when training the model later.

[0110] Step 706: Divide the dataset into multiple pressure adjustment scheme groups according to the second ratio to obtain multiple pressure adjustment scheme group subsets.

[0111] The second ratio refers to the proportion of the multiple pressure adjustment scheme groups to be divided into training set, validation set, and test set. In this embodiment, the proportion of the dataset can be flexibly configured according to the actual application scenario, including but not limited to training set: validation set: test set = 7:1:2.

[0112] Optionally, after preprocessing the data for multiple stress adjustment scheme groups, the controller can divide the dataset according to a second ratio to obtain multiple subsets of stress adjustment scheme groups, namely, a training set (as a subset of stress adjustment scheme groups), a validation set (as a subset of stress adjustment scheme groups), and a test set (as a subset of stress adjustment scheme groups). Data preprocessing includes, but is not limited to, data cleaning (noise removal, missing value imputation, etc.) to improve model performance and training efficiency.

[0113] Step 708: Based on multiple pressure adjustment scheme subsets, perform iterative optimization of the initial ventilator pressure adjustment model until the second iteration optimization stopping condition is met, and obtain the ventilator pressure adjustment model.

[0114] The stopping condition for the second iteration can be flexibly configured according to the actual application scenario. For example, the iteration optimization can be stopped when the value of the loss function is less than a certain threshold.

[0115] Optionally, after dividing the dataset into multiple pressure adjustment scheme groups according to a second ratio, the controller can input the training set into the initial ventilator pressure adjustment model, calculate the output of each layer in the model, and use a pre-set loss function (such as mean squared error, cross-entropy, etc.) to calculate the error between the model output and the actual result. Based on the loss function, the gradient of the parameters of each layer in the model is calculated, and the gradient is backpropagated to the previous layer of the network. Then, gradient descent or other optimization algorithms are used to update the weights and biases of each layer of the neural network in the model. When the second iteration stopping condition is reached, the optimization stops. The controller can repeat the above process multiple times, and in each repetition, different neural network models can be selected for training, resulting in multiple trained models. The validation set is then used to evaluate the multiple trained models to select the best model. Finally, the controller can also perform model performance testing on the selected best model based on the test set, and the model whose test results meet expectations is taken as the final ventilator pressure adjustment model.

[0116] Based on this, such as Figure 8 As shown, another flowchart for generating a ventilator pressure adjustment model is provided, which mainly includes the following steps:

[0117] Step 802: Collect historical target snoring events that characterize precursors to respiratory abnormalities, as well as corresponding ventilator pressure adjustment data for the historical target snoring events;

[0118] Step 804: Divide the historical target snoring events and the corresponding ventilator pressure adjustment data into a dataset to obtain a training set, a validation set, and a test set.

[0119] Step 806: Train the initial ventilator pressure adjustment model based on the training set;

[0120] Step 808: Select the best model from the trained models based on the validation set, and use the test set to test the model performance of the selected best model to determine the final ventilator pressure adjustment model.

[0121] For example, after training a ventilator pressure adjustment model using the controllers of one or more ventilators, the ventilator pressure adjustment model can be integrated into the controller of each ventilator, so that subsequent ventilators can determine the target pressure adjustment scheme that matches the target snoring event detected in real time based on the ventilator pressure adjustment model.

[0122] In this embodiment, the optimal ventilator pressure adjustment model can be obtained through iterative training, selection, and performance testing of the model. This ensures that the ventilator can accurately determine the pressure adjustment scheme that matches the target snoring event detected in real time through the ventilator pressure adjustment model integrated in the controller. This allows the ventilator pressure to be adjusted in a timely and accurate manner when respiratory abnormalities occur, thus preventing respiratory abnormalities in the current user of the ventilator.

[0123] In one embodiment, such as Figure 9 As shown, the automatic control method for ventilators also includes:

[0124] Step 902: When it is detected that the ventilator mask has been removed, or when it is detected that the duration of the event-free segment in the ambient sound monitoring data has reached the preset duration, control the ventilator to reduce pressure.

[0125] Specifically, the event-free segment in the environmental sound monitoring data refers to a segment of monitoring data where there is no snoring and no abnormal breathing. If the duration of the event-free segment in the environmental sound monitoring data reaches the preset duration, it indicates that the user of the ventilator may be awake, hence the prolonged absence of snoring. In this case, timely pressure reduction can prevent discomfort caused by excessive pressure to the awakened user or those who have not yet shown signs of abnormal breathing.

[0126] Optionally, during the operation of the ventilator, when the controller detects that the ventilator mask has been removed, or detects that the duration of the event-free segment in the real-time collected ambient sound monitoring data has reached the preset duration, the controller can first control the ventilator to reduce pressure (the pressure reduction ratio can be flexibly configured according to the actual application scenario), and can control the ventilator to maintain the pressure after pressure reduction for a period of time.

[0127] Furthermore, after controlling the ventilator to depressurize, the server can continuously monitor the latest ambient sound monitoring data collected in real time. If a target snoring event is detected from the latest ambient sound monitoring data, step 904 is executed, and the ventilator is automatically depressurized by executing the above-mentioned automatic ventilator control method.

[0128] Optionally, during the stable operation of the ventilator after depressurization, the controller will still collect ambient sounds in real time through the ambient sound acquisition device in the ventilator. The controller can continue to monitor the real-time ambient sound monitoring data through the snoring event detection model. Once a target snoring event is detected from the latest ambient sound monitoring data, the controller can determine the pressure adjustment scheme in a timely and accurate manner based on the ventilator pressure adjustment model, and control the ventilator to increase the pressure according to the determined pressure adjustment scheme to avoid breathing abnormalities in the ventilator user.

[0129] For example, after the ventilator mask is removed, the controller can first control the ventilator to depressurize by a certain percentage. Before the user puts the mask back on and enters a sleep state, the controller can continue to control the ventilator to maintain stable operation at the depressurized pressure. After the user enters a sleep state, if the controller detects a target snoring event from the real-time collected ambient sound monitoring data, the controller can further control the ventilator to increase pressure according to the target pressure adjustment scheme that matches the target snoring event.

[0130] For example, taking a ventilator user waking up at night as an example, the controller can control the ventilator to automatically reduce pressure based on snoring that has not occurred for a long time. After the user returns to sleep, if the controller detects a target snoring event from the real-time collected ambient sound monitoring data, the controller can further control the ventilator according to the target pressure adjustment scheme matched with the target snoring event. That is, the ventilator pressure can be automatically adjusted in a timely and accurate manner before breathing abnormalities occur.

[0131] In this embodiment, the pressure can be automatically reduced when the ventilator mask is removed or when the ventilator user may be awake, to ensure the comfort of the ventilator user. The pressure of the ventilator can also be automatically adjusted according to the subsequent real-time detection of target snoring events, so that when abnormal breathing signs are detected, the pressure can be increased in a timely and accurate manner to avoid hypoventilation or apnea caused by pressure reduction.

[0132] In a specific application scenario, such as Figure 10 As shown, a comparison chart of the ventilator control effects of the present invention's automatic ventilator control method and the traditional automatic ventilator control method is provided through respiratory event sequences without intervention, respiratory event sequences under the traditional automatic ventilator control method, and respiratory event sequences under the automatic ventilator control method of the present invention.

[0133] Taking the traditional automatic control method of ventilators, which "detects breathing and controls the ventilator to perform pressure compensation upon detecting abnormal breathing," as an example, while the traditional automatic control method can improve breathing abnormalities compared to no intervention, it suffers from a pressure rise delay, requiring pressure increase only after detecting a breathing abnormality. The automatic control method of this invention, on the one hand, can promptly reduce pressure after detecting a prolonged period without events, preventing discomfort to the conscious user due to excessive pressure. On the other hand, by real-time monitoring of collected ambient sound data, it can promptly and accurately detect whether the current user is experiencing any signs of impending breathing abnormalities. Therefore, when these signs occur, it can promptly and accurately control the ventilator to rise according to the corresponding target pressure adjustment scheme, avoiding pressure adjustment delays and effectively improving the user's breathing condition.

[0134] In a specific application scenario, such as Figure 11 As shown, a comparison diagram of the ventilator control effects of the automatic ventilator control method of the present invention and the traditional automatic ventilator control method is further provided. Compared with no intervention, while the traditional automatic ventilator control method can improve respiratory abnormalities, it suffers from a pressure ramp delay. In contrast, the automatic ventilator control method of the present invention can detect the presence of early signs of respiratory abnormalities in the current user by real-time monitoring of collected ambient sound data. Therefore, when these early signs occur, the method can promptly and accurately control the ventilator pressure ramp according to the corresponding target pressure adjustment scheme, avoiding pressure adjustment delay and effectively improving the user's breathing condition and preventing respiratory abnormalities.

[0135] It should be noted that, in all the embodiments described above, the principle of avoiding / minimizing human-machine asynchrony is followed when depressuring or increasing pressure, and the pressure is decreased or increased at a rate acceptable to the user to ensure the comfort of the ventilator user. The specific depressurization or pressure increase rate can be determined based on past ventilator test data.

[0136] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0137] Based on the same inventive concept, this application also provides an automatic ventilator control device for implementing the aforementioned automatic ventilator control method. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the automatic ventilator control device provided below can be found in the limitations of the automatic ventilator control method described above, and will not be repeated here.

[0138] In one exemplary embodiment, such as Figure 12 As shown, an automatic control device for a ventilator is provided, comprising: a monitoring data acquisition module 1202, a snoring event detection module 1204, and an automatic control module for the ventilator 1206, wherein:

[0139] The monitoring data acquisition module is used to acquire ambient sound monitoring data during the operation of the ventilator; the ambient sound monitoring data includes snoring monitoring data and non-snoring monitoring data;

[0140] The snoring event detection module is used to detect snoring events from environmental sound monitoring data.

[0141] The automatic control module of the ventilator is used to determine the target pressure adjustment scheme that matches the target snoring event when a target snoring event that is a precursor to respiratory abnormality is detected, and to control the ventilator to increase the pressure according to the target pressure adjustment scheme.

[0142] The aforementioned automatic control device for the ventilator first acquires ambient sound monitoring data during ventilator operation. This data includes snoring and non-snoring monitoring data. Further, it performs snoring event detection on the ambient sound monitoring data. When a target snoring event, indicative of an impending respiratory abnormality, is detected, a target pressure adjustment scheme matching the target snoring event is determined, and the ventilator is controlled to increase pressure according to the target pressure adjustment scheme. Throughout the control process, by detecting ambient sound monitoring data, it can determine whether a target snoring event, indicative of an impending respiratory abnormality, has occurred, and can specifically determine a ventilator pressure adjustment scheme matching the impending respiratory abnormality. This allows for timely and accurate pressure adjustment of the ventilator according to the corresponding scheme when an impending respiratory abnormality is detected, rather than adjusting after the respiratory abnormality occurs, and avoiding discomfort to the ventilator user due to a mismatch in pressure adjustment schemes. Therefore, it can adjust the ventilator pressure in a timely and accurate manner, thereby ensuring the ventilator's operational effectiveness.

[0143] In one embodiment, the snoring event detection module includes: a snoring event detection model acquisition unit for acquiring a snoring event detection model; and a snoring event detection unit for detecting snoring events based on the snoring event detection model and ambient sound monitoring data.

[0144] In one embodiment, the ventilator automatic control device further includes a snoring event detection model generation module, which is used to acquire multiple historical ambient sound monitoring data; the historical ambient sound monitoring data carries annotations of historical target snoring events; the multiple historical ambient sound monitoring data are divided into multiple historical ambient sound monitoring data subsets according to a first ratio; based on the multiple historical ambient sound monitoring data subsets, the initial snoring event detection model is iteratively optimized until the first iteration optimization stopping condition is reached, thereby obtaining the snoring event detection model.

[0145] In one embodiment, the ventilator automatic control module includes: a ventilator pressure adjustment model acquisition unit for acquiring a ventilator pressure adjustment model; a target snoring monitoring data extraction unit for extracting target snoring monitoring data corresponding to the target snoring event from environmental sound monitoring data; and a target pressure adjustment scheme determination unit for inputting the target snoring monitoring data into the ventilator pressure adjustment model, performing pressure adjustment scheme analysis on the target snoring monitoring data through the ventilator pressure adjustment model, and determining a target pressure adjustment scheme matching the target snoring event.

[0146] In one embodiment, the automatic control device for a ventilator further includes a ventilator pressure adjustment model generation module. This module acquires ventilator pressure adjustment data for each historical target snoring event from multiple historical ambient sound monitoring data sets. For each historical target snoring event, it establishes a correlation between the historical target snoring event and the ventilator pressure adjustment data for that event, generating a pressure adjustment scheme group. It then divides the dataset into multiple pressure adjustment scheme groups according to a second ratio, obtaining multiple pressure adjustment scheme group subsets. Based on these subsets, iteratively optimizes the initial ventilator pressure adjustment model until a second iteration optimization stopping condition is met, thus obtaining the ventilator pressure adjustment model.

[0147] In one embodiment, the automatic control device for the ventilator further includes: a pressure reduction module, used to control the ventilator to reduce pressure when the ventilator mask is detected to be removed or when the duration of the event-free segment in the ambient sound monitoring data has reached a preset duration; after controlling the ventilator to reduce pressure, if a target snoring event is detected from the latest obtained ambient sound monitoring data, the ventilator is controlled to automatically increase pressure by executing the above-described automatic control method for the ventilator.

[0148] The modules in the aforementioned automatic control device for ventilators can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0149] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 13 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores automatic control data for the ventilator. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements an automatic control method for the ventilator.

[0150] Those skilled in the art will understand that Figure 13 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0151] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0152] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0153] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0154] It should be noted that the user information (including but not limited to user personal information and information about the ventilator used by the user) and data (including but not limited to data used for analysis, stored data, and displayed data) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0155] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0156] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0157] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. An automatic control device for a ventilator, characterized in that, The device includes: The monitoring data acquisition module is used to acquire ambient sound monitoring data during the operation of the ventilator; the ambient sound monitoring data includes snoring monitoring data and non-snoring monitoring data. A snoring event detection module is used to detect snoring events in the ambient sound monitoring data; The automatic control module for the ventilator is configured to, upon detecting a target snoring event indicative of an impending respiratory abnormality, determine a target pressure adjustment scheme matching the target snoring event, and control the ventilator to increase pressure according to the target pressure adjustment scheme; the automatic control module for the ventilator also includes a depressurization module, configured to: When the mask of the ventilator is detected to be removed, or when the duration of the event-free segment in the ambient sound monitoring data has reached the preset duration, the ventilator is controlled to depressurize. The event-free segment is a monitoring data segment without snoring or abnormal breathing. After the ventilator depressurizes, if a target snoring event is detected in the latest ambient sound monitoring data, the ventilator is controlled to automatically depressurize.

2. The apparatus according to claim 1, characterized in that, The snoring event detection module includes: The snoring event detection model acquisition unit is used to acquire the snoring event detection model. The snoring event detection unit is used to detect snoring events in the ambient sound monitoring data based on the snoring event detection model.

3. The apparatus according to claim 2, characterized in that, The automatic control device for the ventilator further includes a snoring event detection model generation module, which is used for: Acquire multiple historical ambient sound monitoring data; the historical ambient sound monitoring data carries annotations of historical target snoring events; The dataset is divided according to a first ratio to obtain multiple subsets of historical ambient sound monitoring data. Based on the multiple subsets of historical ambient sound monitoring data, the initial snoring event detection model is iteratively optimized until the first iteration optimization stopping condition is met, thus obtaining the snoring event detection model.

4. The apparatus according to claim 1, characterized in that, The ventilator automatic control module also includes: The ventilator pressure adjustment model acquisition unit is used to acquire the ventilator pressure adjustment model. The target snoring monitoring data extraction unit is used to extract the target snoring monitoring data corresponding to the target snoring event from the ambient sound monitoring data. The target pressure adjustment scheme determination unit is used to input the target snoring monitoring data into the ventilator pressure adjustment model, analyze the target snoring monitoring data through the ventilator pressure adjustment model, and determine the target pressure adjustment scheme that matches the target snoring event.

5. The apparatus according to claim 4, characterized in that, The automatic control device for the ventilator further includes a ventilator pressure adjustment model generation module, which is used for: Acquire ventilator pressure adjustment data for each historical target snoring event in multiple historical environmental sound monitoring data sets; For each historical target snoring event, establish the correlation between the historical target snoring event and the ventilator pressure adjustment data of the historical target snoring event, and generate a pressure adjustment scheme group; The dataset is divided into multiple pressure adjustment scheme groups according to a second ratio to obtain multiple pressure adjustment scheme group subsets. Based on the subset of multiple pressure adjustment schemes, the initial ventilator pressure adjustment model is iteratively optimized until the second iteration optimization stopping condition is met, thus obtaining the ventilator pressure adjustment model.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the apparatus according to any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the apparatus according to any one of claims 1 to 5.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the apparatus according to any one of claims 1 to 5.

Images