WO2023123437A1 - 麻醉控制系统，麻醉系统及非暂态计算机可读介质 - Google Patents

麻醉控制系统，麻醉系统及非暂态计算机可读介质 Download PDF

Info

Publication number
WO2023123437A1
WO2023123437A1 PCT/CN2021/143896 CN2021143896W WO2023123437A1 WO 2023123437 A1 WO2023123437 A1 WO 2023123437A1 CN 2021143896 W CN2021143896 W CN 2021143896W WO 2023123437 A1 WO2023123437 A1 WO 2023123437A1
Authority
WO
WIPO (PCT)
Prior art keywords
anesthesia
index
anesthetic
depth
parameters
Prior art date
Application number
PCT/CN2021/143896
Other languages
English (en)
French (fr)
Inventor
宋敏
吴凯凯
Original Assignee
通用电气精准医疗有限责任公司
宋敏
吴凯凯
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 通用电气精准医疗有限责任公司, 宋敏, 吴凯凯 filed Critical 通用电气精准医疗有限责任公司
Priority to PCT/CN2021/143896 priority Critical patent/WO2023123437A1/zh
Priority to CN202180104853.8A priority patent/CN118354807A/zh
Publication of WO2023123437A1 publication Critical patent/WO2023123437A1/zh

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • 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
    • A61M16/01—Devices for influencing the respiratory system of patients by gas treatment, e.g. mouth-to-mouth respiration; Tracheal tubes specially adapted for anaesthetising
    • 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
    • A61M5/00—Devices for bringing media into the body in a subcutaneous, intra-vascular or intramuscular way; Accessories therefor, e.g. filling or cleaning devices, arm-rests
    • A61M5/14—Infusion devices, e.g. infusing by gravity; Blood infusion; Accessories therefor
    • A61M5/168—Means for controlling media flow to the body or for metering media to the body, e.g. drip meters, counters ; Monitoring media flow to the body
    • A61M5/172—Means for controlling media flow to the body or for metering media to the body, e.g. drip meters, counters ; Monitoring media flow to the body electrical or electronic

Definitions

  • the present application relates to anesthesia control, and more particularly, to an anesthesia control system, an anesthesia system and a non-transitory computer readable medium.
  • Anesthesia is a widely used medical procedure. For example, during surgery, doctors suppress the patient's central nervous system by anesthesizing the patient, thereby creating surgical conditions and eliminating the patient's pain and conscious discomfort.
  • Anesthesia can include inhalation anesthesia and injection anesthesia.
  • the former produces anesthesia by inhaling vaporized anesthetics into the patient.
  • the latter delivers anesthetics into the patient's body through veins or blood injections to produce anesthesia.
  • Different anesthesia uses different anesthetics, and the effects of anesthesia vary.
  • different anesthesia techniques can be used at the same time to perform mixed anesthesia on patients, so as to achieve good anesthesia effect.
  • anesthetics are used in mixed anesthesia, and different anesthetics may have the same anesthetic effect.
  • the anesthetic effects of different anesthetics will affect each other.
  • when adjusting the dose of an anesthetic it is necessary to pay attention to all the physiological indicators of the patient at the same time.
  • anesthesia control system configured to: acquire multiple physiological parameters of the patient; acquire anesthetic parameters of multiple anesthesia devices acting on the patient; acquire the Anesthesia depth index of the patient; and automatically generating control instructions based on the multiple physiological parameters, the anesthetic agent parameters, the anesthesia depth index and the target anesthesia depth index, the control instructions are used to control the multiple anesthesia devices The anesthetic parameters of each anesthesia device, so that the patient's anesthesia depth index is adjusted to the target anesthesia depth index.
  • An anesthesia system including: a physiological parameter collection device for collecting a plurality of physiological parameters of a patient; a plurality of anesthesia devices for applying anesthetics to the patient; anesthesia depth index collection A device for collecting the anesthesia depth index of the patient; and an anesthesia control system, the anesthesia control system is connected to the physiological parameter collection device, the plurality of anesthesia equipment and the anesthesia depth index collection device.
  • the anesthesia control system is configured to: acquire a plurality of physiological parameters of the patient; acquire anesthetic parameters of a plurality of anesthesia devices acting on the patient; acquire an anesthesia depth index of the patient; based on the plurality of physiological parameters, the The anesthesia parameters, the anesthesia depth index and the target anesthesia depth index automatically generate control instructions, and the control instructions are used to control the anesthesia parameters of each anesthesia device in the plurality of anesthesia devices, so that the patient's anesthesia depth index is towards Target anesthesia depth index adjustment.
  • Some embodiments of the present application provide a non-transitory computer-readable medium, the non-transitory computer-readable medium stores a computer program, the computer program has at least one code segment, and the at least one code segment can Executed by a machine to: acquire a plurality of physiological parameters of a patient; acquire anesthetic parameters of a plurality of anesthesia devices acting on the patient; acquire an anesthesia depth index of the patient; and based on the plurality of physiological parameters, the anesthetic parameter , the anesthesia depth index and the target anesthesia depth index automatically generate control instructions, the control instructions are used to control the anesthetic parameters of each anesthesia device in the plurality of anesthesia devices, so that the patient's anesthesia depth index moves toward the target anesthesia depth Index adjustment.
  • Figure 1 is a schematic diagram of an anesthesia system according to some embodiments of the present application.
  • Fig. 2 is a flowchart of a control method of an anesthesia control system according to some embodiments of the present application
  • Figure 3 is a diagram of a learning network according to some embodiments of the application.
  • Fig. 4 is a flowchart of a control method of an anesthesia control system according to some other embodiments of the present application.
  • FIG. 1 a schematic diagram of an anesthesia system 100 according to some embodiments of the present application is shown.
  • the anesthesia system 100 includes: a physiological parameter acquisition device 101, configured to acquire the multiple physiological parameters of the patient 110; a plurality of anesthesia equipment including the first anesthesia equipment 102 and the second anesthesia equipment 103, used to apply anesthetics to the The patient 110; the depth of anesthesia index collection device 104, used to collect the depth of anesthesia index of the patient; and the anesthesia control system 105, the anesthesia control system 105 and the physiological parameter collection device 101, the multiple anesthesia equipment And the anesthesia depth index collection device 104 is connected.
  • a physiological parameter acquisition device 101 configured to acquire the multiple physiological parameters of the patient 110
  • a plurality of anesthesia equipment including the first anesthesia equipment 102 and the second anesthesia equipment 103 used to apply anesthetics to the The patient 110
  • the depth of anesthesia index collection device 104 used to collect the depth of anesthesia index of the patient
  • the types of the physiological parameter collection device 101 may be various. For example, it may be either a monitor, an electrocardiograph, or both.
  • the physiological parameter collection device 101 can be used for measuring routine physiological parameters of the patient 110 .
  • the physiological parameter collection device 101 includes a variety of different measurement modules, such as a heart rate measurement module, a blood oxygen saturation measurement module, a blood pressure measurement module, a blood glucose measurement module, an end-tidal carbon dioxide concentration measurement module, an end-tidal oxygen multiple of the concentration measurement modules. These different modules can be integrated in the same instrument, or they can be scattered in different medical devices, and doctors can choose freely according to their needs.
  • the plurality of anesthesia devices may include a first anesthesia device 102 and a second anesthesia device 103 .
  • the first anesthesia device 102 and the second anesthesia device 103 are different.
  • the first anesthesia device 102 may be an inhalation anesthesia device
  • the second anesthesia device 103 is an injection anesthesia device.
  • the specific structures of the inhalation anesthesia equipment and the injection anesthesia equipment can be arbitrary in the field, and this application is only described as an example.
  • Inhalation anesthesia equipment may include an anesthesia machine or anesthesia agent and its associated air supply system.
  • the gas supply system may provide gas to the anesthetic agent through tubing.
  • Types of gases include, but are not limited to, air, oxygen, and nitrous oxide.
  • the anesthesia machine contains a vaporizer for vaporizing the anesthetic. After the above gas of the gas supply system enters the anesthesia machine, it can be mixed with vaporized anesthetic and provided to the patient.
  • the parameters of the anesthetic agent of the inhalation anesthesia device can be adjusted. For example, during use of the anesthetic agent vaporizer, the dose of anesthetic agent supplied to the patient may be adjusted by adjusting the gas flow rate of the gas supply system to the vaporizer.
  • the specific adjustment method can be realized by several flow adjustment devices (eg, valves).
  • the amount of vaporized anesthetic can also be controlled by controlling the vaporizer of the anesthesia machine.
  • the adjustment of the parameters of the anesthetic agent supplied to the patient can be realized.
  • the inhalation anesthesia device may also include modules such as sensors for monitoring the state of the anesthesia machine or the state of the patient, which will not be repeated here.
  • An injectable anesthesia device may include an anesthetic injection module and a control module.
  • the anesthetic agent injection module may include a plurality, and each of the plurality of anesthetic agent injection modules may be used to inject anesthetic agents of different effects.
  • the anesthetic injection module may include a syringe pump. The specific structure of the syringe pump can be arbitrary in the art, and will not be repeated here.
  • the control module can obtain the current injection rate of the syringe pump and/or the dose of anesthetic agent by means of calculation, etc., and then control the injection rate of the syringe pump.
  • the anesthesia depth index collecting device 104 is used for collecting the anesthesia depth index of the patient.
  • the index of depth of anesthesia may use common indexes in this field, for example, the index of depth of anesthesia may include at least one of sedation index, analgesia index and muscle relaxation index.
  • the depth of anesthesia index includes three indexes of sedation, analgesia and muscle relaxation.
  • EEG electroencephalogram
  • RE reaction entropy
  • SE state entropy
  • the reaction entropy can be obtained by analyzing the frontal electromyography and electroencephalogram (EEG); the state entropy can be obtained by the EEG.
  • EEG electroencephalogram
  • the two can reflect the degree of excitation of the frontal skeletal muscle and the degree of inhibition of the cerebral cortex during the recovery phase.
  • Both RE and SE are maintained at a high level, indicating that the subject is awake; both RE and SE are maintained at a low level, and the hemodynamic parameters are stable, indicating that the subject is in an appropriate level of anesthesia; an increase in RE, If SE remains unchanged at a relatively low level, the patient may have physical activity or the patient may feel pain; if RE increases, SE remains unchanged at a relatively high level, indicating that the patient may be waking up.
  • the RE and SE can be used to judge analgesia and sedation index.
  • the aforementioned anesthesia depth index may also include a surgical plethysmographic index (SPI, also known as a surgical stress index). It can be used to monitor the number of patients' hemodynamic responses to surgical stimuli and analgesic drug therapy during general anesthesia, and can reflect the increase of sympathetic nerve activity in response to painful or nociceptive stimuli.
  • the EEG bispectral index (BIS) can be selected as the above-mentioned anesthesia depth index. It can be understood that the depth of anesthesia index can also be arbitrary in the art, and will not be exemplified.
  • the combination of multiple anesthesia equipment can provide better anesthesia effect, illustrated by the combination of inhalation anesthesia and injection anesthesia.
  • Inhalational anesthesia has a faster onset and failure rate during the induction and recovery periods of anesthesia, but some patients may not be able to withstand the physiological reactions caused by inhalational anesthesia, such as vasodilation, for a short time.
  • Injection anesthesia is more likely to cause respiratory depression than inhalation anesthesia.
  • the combination of the two can exert their respective advantages to achieve better anesthesia effect. This places high demands on the dosing of anesthetics for inhalation and injection anesthesia.
  • the present application provides the aforementioned anesthesia control system 105 .
  • the aforementioned connection of the anesthesia control system 105 enables communication between various settings.
  • the physiological parameters of the patient can be obtained from the physiological parameter acquisition device 101;
  • the current anesthetic dosage can be obtained from the anesthesia equipment, such as the information of anesthetic concentration, dosage, and injection rate; it can be obtained from the anesthesia depth index acquisition device 104
  • the patient's current depth of anesthesia index can be obtained from the physiological parameter acquisition device 101.
  • the anesthesia control system 105 is also set with a target anesthesia depth index.
  • the target anesthesia depth index may be artificially set by a user such as an anesthesiologist based on experience.
  • the target anesthesia depth index can also be automatically set by the anesthesia control system 105, for example, automatically assigned according to factors such as the patient's age, gender, and medical history.
  • the anesthesia control system 105 can be of various types, for example, it can be a controller circuit including a processor.
  • the controller circuit can be configured in an independent control system.
  • the control system can communicate with various other components in the anesthesia control system 105 in a wired or wireless manner.
  • the anesthesia control system 105 may also be integrated in a certain device in the above-mentioned anesthesia system 100 .
  • it can be integrated in the above-mentioned anesthesia equipment to improve the degree of integration of the instrument.
  • the anesthesia control system can be located at a remote terminal.
  • it can be installed on a public server of a hospital, so that anesthesia operations in multiple operating rooms can be controlled simultaneously.
  • the anesthesia control system can be set on the cloud server by means of remote communication, such as 5G communication technology.
  • the anesthesia control system 105 can comprehensively judge the current physiological condition of the patient, the equipment parameters of multiple anesthesia equipment such as inhalation anesthesia equipment and injection anesthesia equipment, the patient's anesthesia depth index, etc., so as to provide comprehensive control Program, and adjust the anesthetic parameters of various anesthesia equipment at the same time.
  • Such a control scheme can simultaneously consider the interaction between different anesthesia devices. For example, common sedative and/or analgesic effects between different drugs used in inhalation anesthesia devices and injectable anesthesia devices. Antagonism of anesthetic agents between different anesthetic devices can also be considered.
  • anesthesia control system 105 is also more secure than human judgment and adjustment.
  • Some embodiments of the present application also provide a control method of an anesthesia control system.
  • the anesthesia system includes a controller circuit for performing an anesthesia control method.
  • FIG. 2 it shows a flowchart of a control method of an anesthesia control system in some embodiments of the present application.
  • step 201 a plurality of physiological parameters of a patient are acquired.
  • This step can be obtained from the physiological parameter collection device 101 by the anesthesia control system 105 shown in FIG. 1 .
  • the physiological parameter acquisition device 101 can be connected to and acquire the physiological parameters of the patient 110 in real time, and further transmit the physiological parameters to the anesthesia control system 105 .
  • the plurality of physiological parameters may include a plurality of heart rate, blood oxygen saturation, blood pressure, blood glucose, end-tidal carbon dioxide concentration, and end-tidal oxygen concentration.
  • step 202 anesthesia parameters of a plurality of anesthesia devices acting on the patient are acquired.
  • the aforementioned multiple anesthesia devices may include a first anesthesia device 102 and a second anesthesia device 103 .
  • the first anesthesia device 102 may be an inhalation anesthesia device.
  • its anesthetics may include sevoflurane, desflurane, isoflurane, etc.
  • the anesthetic parameter may be the dosage of one or more of the above-mentioned anesthetics (for example, concentration, inhalation rate, etc.).
  • the second anesthesia device 103 may be an injectable anesthesia device.
  • the anesthetics thereof may include hexbital sodium, methohexital sodium and the like.
  • the anesthetic parameter may be the injection volume of the above-mentioned one or more anesthetics (for example, injection rate, anesthetic concentration, etc.).
  • the above-mentioned anesthesia equipment may have its own monitoring module for the amount of anesthetic agent, and the amount of anesthetic agent may be transmitted to the anesthesia control system through this module.
  • the anesthesia depth index of the patient is acquired.
  • the depth of anesthesia index can be acquired by the depth of anesthesia index acquisition device 104 .
  • the anesthesia depth index collection device 104 may include an EEG device, so as to collect the EEG signal of the patient in real time, and the EEG signal collected in real time can be analyzed and processed to obtain the desired anesthesia depth index.
  • the depth of anesthesia index can refer to the description above in this application, and be evaluated by indexes in the field such as RE, SE, BIS, etc., and will not be repeated here. Further, the above anesthesia depth index is transmitted to the anesthesia control system 105 for comprehensive processing.
  • control instructions are automatically generated based on the plurality of physiological parameters, the anesthetic agent parameters, the anesthesia depth index and the target anesthesia depth index, and the control instructions are used to control each of the plurality of anesthesia devices The anesthetic agent parameters of the anesthesia equipment, so that the patient's anesthesia depth index is adjusted to the target anesthesia depth index.
  • the anesthesia control system 105 can make a comprehensive judgment by inputting the physiological parameters of the patient, the anesthetic agent parameters of the anesthesia equipment, the patient's anesthesia depth index and the target anesthesia depth index. Anesthesia parameter adjustments for each of the plurality of anesthesia devices are simultaneously considered. Potential adverse effects of isolated adjustments are thereby avoided.
  • control instructions can be generated based on a trained learning network.
  • An exemplary description of the training and use of this learned network is given below.
  • the learning network is completed by preparing training data, selecting and constructing a learning network model, training, testing and optimizing the learning network.
  • the learning network is obtained by training a data set based on multiple parameters of samples (known input) and corresponding anesthetic parameters (expected output), specifically, the training of the learning network includes the following Step one to step three.
  • Step 1 Obtain different patient physiological parameters, anesthesia equipment anesthetic parameters, and patient anesthesia depth indices of multiple successful anesthesia cases as a sample parameter set. Further, the parameter set is input into the learning network, and then the normalization process is performed based on the normalization layer in the learning network.
  • Step 2 Obtain the anesthetic parameter set for each of the above successful cases.
  • the anesthesia parameters include anesthesia parameters of each of the plurality of anesthesia devices.
  • Step 3 Using the sample parameter sets of different patient physiological parameters, anesthesia equipment parameters, and patient anesthesia depth index as input, and the anesthetic parameter set as output, train the learning network to obtain the trained learning network.
  • the trained learning network may be updated based on a new sample set and its corresponding anesthetic parameter set.
  • the learning network is trained based on ResNet (Residual Network) or VGGNet (Visual Geometry Group Network) or other known models. Since the number of processing layers in ResNet can be set a lot (up to more than 1000 layers), correspondingly, the effect of classification based on the network structure (for example, the judgment of artifact type) is better. In addition, ResNet is also more effective. It is easy to optimize the learning network based on more training data.
  • ResNet Residual Network
  • VGGNet Visual Geometry Group Network
  • Figure 4 illustrates a learning network 150 according to some embodiments of the invention.
  • the learning network 150 includes an input layer 151 , a processing layer (or called a hidden layer) 152 and an output layer 153 .
  • the processing layer 152 includes a first convolutional layer 155 , a first pooling layer 156 , and a fully connected layer 157 .
  • the first convolutional layer 155 is used to convolve each input parameter to obtain a feature map of the first convolutional layer.
  • the first pooling layer 156 pools (downsamples) the feature map of the first convolutional layer to compress the feature map of the first convolutional layer and extract the main features to obtain the feature map of the first pooling layer .
  • the fully connected layer 157 can output a judgment result based on the feature map of the first pooling layer.
  • Figure 4 only shows an example of one convolutional layer, in other examples, the number of convolutional layers can be arbitrary, and the number of convolutional layers can be adaptively adjusted according to the size of the input data in the learning network, etc.
  • a second convolutional layer and a second pooling layer are also included between the first pooling layer 156 and the fully connected layer 157, or, between the first pooling layer 156 and the fully connected
  • the layers 157 also include a second convolutional layer and a second pooling layer, a third convolutional layer and a third pooling layer (not shown in the figure), and the like.
  • Figure 4 only shows that the convolutional layer is connected to the input layer, the pooling layer is connected to the convolutional layer, and the fully connected layer is connected to the pooling layer, in other examples, any Any number of processing layers of any type, for example, a normalization layer is set between the convolutional layer and the input layer to normalize the input parameters, or an activation layer is set between the fully connected layer and the pooling layer to The feature map of the pooling layer is non-linearly mapped using the Rectified linear unit (ReLU) activation function.
  • ReLU Rectified linear unit
  • each layer includes several neurons 160, and the number of neurons in each layer may be the same, or may be set differently as required.
  • the learning process usually utilizes (partial) input Data, and create a network output for the input data, and then compare the created network output based on the known input with the expected output of the data set, and the difference is the loss function (loss function), which can be iteratively Update the parameters (weights and/or biases) of the network to continuously reduce the loss function, thereby training a neural network model with higher accuracy.
  • loss function loss function
  • the loss function there are many functions that can be used as the loss function, including but not limited to mean square error (mean suqared), cross entropy error (cross entropy error) and so on.
  • mean square error mean suqared
  • cross entropy error cross entropy error
  • the learning network 150 while the configuration of the learning network 150 will be guided by prior knowledge of the estimation problem, the dimensions of the inputs, outputs, etc., the learning itself is considered a "black box" and relies primarily on or exclusively from the input data to achieve the desired The best approximation to the output data.
  • certain aspects and/or features of the data, imaging geometry, reconstruction algorithms, etc. may be utilized to assign explicit meaning to certain data representations in the learning network 150 . This can help speed up training. Because this creates an opportunity to individually train (or pre-train) or define certain layers in the learning network 150 .
  • Deep learning methods are characterized by the use of one or more network architectures to extract or model data of interest.
  • Deep learning methods can use one or more processing layers (for example, input layer, output layer, convolutional layer, normalization layer, sampling layer, etc., according to different deep learning network models can have different numbers and functions of processing layers ), where the configuration and number of layers allow deep learning networks to handle complex information extraction and modeling tasks.
  • Specific parameters of the network also called “weights” or “biases” are usually estimated through a so-called learning process (or training process).
  • the parameters being learned or trained typically result in (or output) a network corresponding to different levels of layers, so extracting or simulating different aspects of the initial data or the output of a previous layer can often represent a hierarchy or cascade of layers.
  • Processing can be done hierarchically, i.e., earlier or higher-level layers may correspond to extracting "simple" features from the input data, followed by layers that combine these simple features into features exhibiting higher complexity.
  • each layer (or more specifically, each "neuron” in each layer) can employ one or more linear and/or nonlinear transformations (so-called activation functions) to process the input data into an output data representation .
  • the number of "neurons" may be constant across layers, or may vary from layer to layer.
  • a training dataset includes known input values as well as desired (target) output values (eg, judgment results) for the final output of the deep learning process.
  • target desired output values
  • a deep learning algorithm can process this training data set (in a supervised or guided manner or in an unsupervised or unsupervised manner) until it identifies a mathematical relationship between known inputs and desired outputs and/or identifies and represent the mathematical relationship between the input and output of each layer.
  • the learning process typically takes (part of) input data and creates a network output for that input data, then compares the created network output to the desired output for that dataset, and then uses the difference between the created and desired output to iteratively Update the parameters (weights and/or biases) of the network.
  • a stochastic gradient descent (SGD) method can be used to update network parameters, but those skilled in the art should understand that other methods known in the art can also be used to update network parameters.
  • a separate verification data set can be used to verify the trained learning network, where the known input and expected output are known, and the network output can be obtained by providing the known input to the trained learning network, This network output is then compared to the (known) expected output to validate previous training and/or prevent overtraining.
  • FIG. 4 a flowchart 400 of a control method of an anesthesia control system according to some other embodiments of the present application is shown.
  • the control method of this example can be performed on the basis of the method flowchart 200 such as step 204 .
  • a detailed description is given below.
  • step 401 updates of the multiple physiological parameters, the anesthesia depth index, and the anesthetic agent parameters of the multiple anesthesia devices are obtained; based on the updates, updated control instructions are automatically generated.
  • the above example of the present application discloses that the dynamic update of the control instruction is realized according to the dynamic change of the above parameters, so that anesthesia can be performed more accurately.
  • the above process can be realized by the anesthesia control system 105 .
  • the dynamically updated control instructions can more accurately meet the needs of suitable anesthesia, and better adapt to changes in the situation caused by individual accidental factors of patients.
  • step 402 based on the plurality of physiological parameters, the depth of anesthesia index, the anesthetic agent parameters and the target depth of anesthesia index, generate an update prediction on the depth of anesthesia index; in real time Comparing the update of the anesthesia depth index with the update prediction of the anesthesia depth index, if the two are greater than a certain threshold, a notification is sent.
  • the anesthesia control system 105 can be used to provide accurate prediction of anesthesia abnormalities. Compared with the traditional prediction method, which only uses intuitive physiological indicators to judge whether the patient is in a safe state, it cannot effectively predict the possibility of anesthesia risk.
  • the anesthesia control system 105 of the present application using artificial intelligence means such as learning networks, can comprehensively predict the future trend of changes in the patient's future anesthesia depth index based on various current physiological parameters, the anesthesia depth index, and the anesthetic agent parameters. This provides conditions for the comparison of the actual anesthesia depth index and the predicted anesthesia depth index. Once the comparison results show that there is a large deviation between the two, it means that the risk of anesthesia is high.
  • the doctor can be reminded by means of an alarm, which greatly improves the safety of anesthesia.
  • the aforementioned thresholds can be set manually, or can be independently determined by the anesthesia control system 105 , and will not be repeated here.
  • the anesthetic parameter threshold is determined, the anesthetic parameter threshold is compared with the anesthetic parameter, and a notification is issued if the anesthetic parameter exceeds the threshold.
  • the control command of the anesthesia control system 105 not only focuses on whether the anesthesia depth index reaches the standard, but also ensures that the anesthetic agent dosage will not be excessive on the basis of the anesthesia depth index reaching the standard. In case of overdose, the operator will be notified. Similar to step 402, the above-mentioned threshold in step 403 can be set manually, or can be judged by the anesthesia control system 105 independently.
  • the anesthesia control system 105 can be any system for executing the method described in any embodiment of the present application.
  • the anesthesia control system 105 can be of various types, for example, it can be a controller circuit including a processor. The controller circuit may be configured within a stand-alone control system.
  • the anesthesia control system 105 may also be integrated in a certain device in the above-mentioned anesthesia system 100 .
  • the anesthesia control system can be located at a remote terminal.
  • it can be installed on a public server in a hospital, so that anesthesia operations in multiple operating rooms can be controlled simultaneously.
  • the anesthesia control system can be set on the cloud server by means of remote communication, such as 5G communication technology.
  • Some embodiments of the present application also provide a non-transitory computer readable medium.
  • the non-transitory computer readable medium stores a computer program having at least one code segment executable by a machine to: obtain a plurality of physiological parameters of a patient; anesthesia parameters of a plurality of anesthesia equipment; obtain the anesthesia depth index of the patient; and automatically generate control instructions based on the multiple physiological parameters, the anesthesia parameters, the anesthesia depth index and the target anesthesia depth index, the control The instructions are used to control the anesthetic agent parameters of each of the multiple anesthesia devices, so that the patient's anesthesia depth index is adjusted to a target anesthesia depth index.
  • control instruction is automatically generated based on a trained learning network.
  • the anesthetic parameters of the plurality of anesthesia devices include anesthetic parameters of inhalation anesthesia devices and anesthetic parameters of injection anesthesia devices.
  • the anesthesia depth index includes at least one of sedation index, analgesia index and muscle relaxation index.
  • the multiple physiological parameters include heart rate, blood oxygen saturation, blood pressure, blood sugar, end-tidal carbon dioxide concentration, and end-tidal oxygen concentration.
  • the non-transitory computer readable medium is further configured to: obtain updates of the plurality of physiological parameters, the depth of anesthesia index, and the anesthetic parameters of the plurality of anesthesia devices; and based on the updates, Automatically generate updated control instructions.
  • the non-transitory computer readable medium is further configured to: generate an updated prediction for the depth of anesthesia index based on the plurality of physiological parameters, the depth of anesthesia index and the anesthetic agent parameter and a target depth of anesthesia index ; Real-time comparison of the update of the depth of anesthesia index and the update forecast of the depth of anesthesia index, and if the two are greater than a certain threshold, a notification is issued.
  • the non-transitory computer readable medium is further configured to: determine an anesthetic parameter threshold, compare the anesthetic parameter threshold with the anesthetic parameter, and issue a notification if the anesthetic parameter exceeds the threshold.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Anesthesiology (AREA)
  • Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • General Health & Medical Sciences (AREA)
  • Hematology (AREA)
  • Surgery (AREA)
  • Physics & Mathematics (AREA)
  • Molecular Biology (AREA)
  • Medical Informatics (AREA)
  • Vascular Medicine (AREA)
  • Pathology (AREA)
  • Biophysics (AREA)
  • Emergency Medicine (AREA)
  • Pulmonology (AREA)
  • Apparatus For Radiation Diagnosis (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)

Abstract

一种麻醉控制系统（105），该麻醉控制系统（105）被配置为：获取患者（110）的多个生理参数；获取作用于患者（110）的多个麻醉设备（102、103）的麻醉剂参数；获取患者（110）的麻醉深度指数；基于多个生理参数、麻醉剂参数、麻醉深度指数以及目标麻醉深度指数自动生成控制指令，控制指令用于控制多个麻醉设备（102、103）中的每一个麻醉设备（102、103）的麻醉剂参数，从而使患者（110）的麻醉深度指数向目标麻醉深度指数调节。还提供了一种麻醉系统（100）和一种非暂态计算机可读介质。

Description

麻醉控制系统，麻醉系统及非暂态计算机可读介质 技术领域
本申请涉及麻醉控制，更具体地，涉及一种麻醉控制系统、麻醉系统及非暂态计算机可读介质。
背景技术
麻醉术是一种广泛使用的医疗手段。例如，在外科手术时，医生通过对患者进行麻醉来抑制患者的中枢神经系统，从而创造手术条件并能够消除患者的疼痛感和意识上的不适。
麻醉术可以包括吸入式麻醉和注射式麻醉。前者通过将气化的麻醉剂使患者吸入而产生麻醉效果。后者通过静脉或者血液注射的方式将麻醉剂送入患者体内而产生麻醉效果。不同的麻醉术采用不同的麻醉剂，麻醉效果各有不同。临床上可以同时使用不同麻醉术来对患者进行混合麻醉，从而达到良好的麻醉效果。
混合麻醉中采用多种不同的麻醉剂，不同的麻醉剂可能具有相同的麻醉作用。另外，不同的麻醉剂之间的麻醉作用会相互影响。此外，不同的麻醉剂之间还可能具有拮抗的作用。因此，混合麻醉中麻醉剂剂量的调节变得十分复杂。一方面，调节某一麻醉剂的剂量时需要同时关注患者全部的生理指标。另一方面，还需要考虑是否调整其他麻醉剂的用量。这对麻醉医生的专业技能和工作量都是极大的考验。
发明内容
上述的缺陷、缺点和问题在本文中得到解决，通过阅读和理解以下的说明会理解这些问题和方案。
本申请的一些实施例中提供了一种麻醉控制系统，所述麻醉控制系统被配置为：获取患者的多个生理参数；获取作用于所述患者的多个麻醉设备的麻醉剂参数；获取所述患者的麻醉深度指数；以及基于所述多个生理参数、所述麻醉剂参数、所述麻醉深度指数以及目标麻醉深度指数自动生成控制指令，所述控制指令用于控制所述多个麻醉设备中的每一个麻醉设备的麻醉剂参数，从而使患者的麻醉深度指数向目标麻醉深度指数调节。
本申请的一些实施例中提供了一种麻醉系统，包括：生理参数采集装置，用于采集患者的多个生理参数；多个麻醉设备，用于将麻醉剂作用于所述患者；麻醉深度指数采集装置，用于采集所述患者的麻醉深度指数；以及麻醉控制系统，所述麻醉控制系统和所述生理参数采集装置、所述多个麻醉设备以及所述麻醉深度指数采集装置连接。所述麻醉控制系统被配置为：获取患者的多个生理参数；获取作用于所述患者的多个麻醉设备的麻醉剂参数；获取所述患者的麻醉深度指数；基于所述多个生理参数、所述麻醉剂参数、所述麻醉深度指数以及目标麻醉深度指数自动生成控制指令，所述控制指令用于控制所述多个麻醉设备中的每一个麻醉设备的麻醉剂参数，从而使患者的麻醉深度指数向目标麻醉深度指数调节。
本申请的一些实施例中提供了一种非暂态计算机可读介质，所述非暂态计算机可读介质存储有计算机程序，所述计算机程序具有至少一个代码段，所述至少一个代码段能够由机器执行以：获取患者的多个生理参数；获取作用于所述患者的多个麻醉设备的麻醉剂参数；获取所述患者的麻醉深度指数；以及基于所述多个生理参数、所述麻醉剂参数、所述麻醉深度指数以及目标麻醉深度指数自动生成控制指令，所述控制指令用于控制所述多个麻醉设备中的每一个麻醉设备的麻醉剂参数，从而使患者的麻醉深度指数向目标麻醉深度指数调节。
应理解，提供上文的简要描述是为了以简化的形式介绍在具体实施方式中进一步描述的一些概念。这并不意味着识别所要求保护的主题的关键或必要特征，其范围由详细描述之后的权利要求唯一地限定。此外，所要求保护的主题不限于解决在上文中或在本公开的任一区段中所提及的任何缺点的实现。
附图说明
参考所附附图，通过阅读下列非限制性实施例的描述，本发明将被更好的理解，其中：
图1是根据本申请一些实施例的麻醉系统的示意图；
图2是根据本申请的一些实施例的麻醉控制系统的控制方法流程图；
图3是根据本申请的一些实施例的学习网络的图；
图4是根据本申请的另外一些实施例的麻醉控制系统的控制方法流程图。
具体实施方式
以下将描述本公开的具体实施方式，需要指出的是，在这些实施方式的具体描述过程中，为了进行简明扼要的描述，本公开不可能对实际的实施方式的所有特征均作详尽的描述。应当可以理解的是，在任意一种实施方式的实际实施过程中，正如在任意一个工程项目或者设计项目的过程中，为了实现开发者的具体目标，为了满足系统相关的或者商业相关的限制，常常会做出各种各样的具体决策，而这也会从一种实施方式到另一种实施方式之间发生改变。此外，还可以理解的是，虽然这种开发过程中所作出的努力可能是复杂并且冗长的，然而对于与本公开公开的内容相关的本领域的普通技术人员而言，在本公开揭露的技术内容的基础上进行的一些设计，制造或者生产等变更只是常规的技术手段，不应当理解为本公开的内容不充分。
除非另作定义，权利要求书和说明书中使用的技术术语或者科学术语应当为所属技术领域内具有一般技能的人士所理解的通常意义。本公开以及权利要求书中使用的“第一”、“第二”以及类似的词语并不表示任何顺序、数量或者重要性，而只是用来区分不同的组成部分。“一个”或者“一”等类似词语并不表示数量限制，而是表示存在至少一个。“包括”或者“包含”等类似的词语意指出现在“包括”或者“包含”前面的元件或者物件涵盖出现在“包括”或者“包含”后面列举的元件或者物件及其等同元件，并不排除其他元件或者物件。“连接”或者“相连”等类似的词语并非限定于物理的或者机械的连接，也不限于是直接的还是间接的连接。
参考图1，示出了根据本申请一些实施例的麻醉系统100的示意图。
该麻醉系统100包括：生理参数采集装置101，用于采集患者110的所述多个生理参数；第一麻醉设备102和第二麻醉设备103在内的多个麻醉设备，用于将麻醉剂作用于所述患者110；麻醉深度指数采集装置104，用于采集所述患者的麻醉深度指数；以及麻醉控制系统105，所述麻醉控制系统105和所述生理参数采集装置101、所述多个麻醉设备以及所述麻醉深度指数采集装置104连接。下面对麻醉系统100的各个组成部分进行示例性说明。
生理参数采集装置101的类型可以是多样的。例如，其可以是监护仪，心电图仪中的任意一个或者两者。生理参数采集装置101可以用于患者110常规的生理参数的测量。在一个示例中，生理参数采集装置101包括多种不同的测量模块，例如心率测量模块，血氧饱和度测量模块，血压测量模块，血糖测量模块，呼气末二氧化碳浓度 测量模块，呼气末氧浓度测量模块中的多个。这些不同的模块可以集成在同一仪器内，也可以是分散在不同的医学设备中的模块，医生可以根据需要进行自由选择。
多个麻醉设备可以包括第一麻醉设备102和第二麻醉设备103。在一个示例中，第一麻醉设备102和第二麻醉设备103是不同的。例如，第一麻醉设备102可以是吸入式麻醉设备，第二麻醉设备103是注射式麻醉设备。吸入式麻醉设备和注射式麻醉设备的具体构造，可以是本领域任意的，本申请仅作示例性描述。
吸入式麻醉设备可以包括麻醉机或者麻醉剂及其配套的供气系统。在一个示例中，供气系统可以通过管道将气体提供给麻醉剂。气体的种类包括但不限于空气、氧气和一氧化二氮。麻醉机中包含有气化器，用于将麻醉剂气化。供气系统的上述气体在进入麻醉机后，能够和气化麻醉剂混合而提供给患者。该吸入式麻醉设备的麻醉剂参数是可以调节的。例如，在麻醉剂气化器的使用期间，可以通过调节供气系统到气化器的气体流速来调节供应给患者的麻醉剂的剂量。而具体的调节方式，可以是通过若干流量调节装置(例如，阀门)来实现。此外，还可以通过控制麻醉机气化器来对气化麻醉剂的量进行控制。这样的配置方式，即可实现供应给患者的麻醉剂参数的调节。除此以外，吸入式麻醉设备上还可以包括传感器等模块，用于监测麻醉机的状态或者患者的状态，不再赘述。
注射式麻醉设备可以包括麻醉剂注射模块以及控制模块。在一个示例中，麻醉剂注射模块可以包括多个，多个麻醉剂注射模块中的每一个可以用于注射不同效果的麻醉剂。麻醉剂注射模块可以包括注射泵。注射泵具体构造可以是本领域任意的，不再赘述。而控制模块可以通过计算等方式得出当前的注射泵注射速率和/或麻醉剂剂量，进而对注射泵的注射速度进行控制。
可以理解，上文对于吸入式麻醉设备和注射式麻醉设备仅是示例性的说明，并不做唯一限定。具体的设备构造可以是现有技术中所任意的，不再一一列举。
麻醉深度指数采集装置104用于患者的麻醉深度指数的采集。麻醉深度指数可以采用本领域通用的指标，例如，麻醉深度指数可以包括镇静指数，镇痛指数和肌肉松弛指数中的至少一个。在一些示例中，麻醉深度指数包括镇静指数，镇痛指数和肌肉松弛指数三者。上述三个指数的采集、判断方式可以是多样的。例如，麻醉深度指数采集装置104可以包括脑电图(EEG)仪，脑电图仪用于采集反应熵(RE)以及状态熵(SE)。其中，反应熵可以通过前额肌电图与脑电图(EEG)分析而得；状态熵 可以通过脑电图获取。二者能够反映复苏阶段前额骨骼肌兴奋程度及大脑皮层的受抑制程度。RE、SE两者均维持在高水平值表示被监护者已清醒；RE、SE两者均维持低水平值，且血流动力学参数稳定表示被监护者处于合适的麻醉水平；RE升高，SE维持不变在相对低水平值表示病人可能有肌体活动或病人可能感觉有疼痛；RE升高，SE维持不变在相对高水平值表示病人可能在苏醒。RE和SE可以用于判断镇痛、镇静指数。此外，上述麻醉深度指数还可以包括手术体积描计指数(SPI，亦称为手术应激指数)。其可以用来监测全身麻醉期间患者对于手术刺激和镇痛药物治疗的血流动力学反应的数字，并且能够反映患者对疼痛或伤害性感受刺激产生反应而增加交感神经活性。在可替代的实施例中，可以选择脑电双频指数(BIS)作为上述麻醉深度指数。可以理解，麻醉深度指数还可以本领域中任意的，不再例举。
多个麻醉设备的联用能够提供更好的麻醉效果，以吸入式麻醉和注射式麻醉的组合加以说明。吸入式麻醉在麻醉诱导期和恢复期起效和失效速度较快，但部分患者可能短时间无法承受吸入麻醉引起的生理反应例如血管舒张。而注射式麻醉相比于吸入式麻醉容易导致呼吸抑制。二者的结合可以发挥各自优势达到更好的麻醉效果。这对于吸入式麻醉和注射式麻醉的麻醉剂用量分配提出了很高要求。
鉴于此，本申请提供了上述麻醉控制系统105。如图1所示，所述麻醉控制系统105和所述生理参数采集装置101、所述多个麻醉设备(如，第一麻醉设备102和第二麻醉设备103)以及所述麻醉深度指数采集装置104连接。麻醉控制系统105的上述连接能够和各个设置之间进行通讯。具体而言，可以从生理参数采集装置101处获取患者的生理参数；可以从麻醉设备处获取当前的麻醉剂用量，例如麻醉剂浓度、剂量、注射速率的信息；可以从麻醉深度指数采集装置104处获取患者当前的麻醉深度指数。同时，麻醉控制系统105还设置有目标麻醉深度指数。目标麻醉深度指数可以是使用者例如麻醉医生根据经验所人为设定的。在另一个实施例中，目标麻醉深度指数还可以是麻醉控制系统105自动设定的，例如，根据患者的年龄、性别、病史等因素自动分配的。
麻醉控制系统105的种类可以是多样的，例如，其可以是包括处理器在内的控制器电路。该控制器电路可以被配置在独立存在的控制系统内，相应地，该控制系统可以通过有线或者无线的方式来和麻醉控制系统105中的各个其他部件之间进行通讯。此外，麻醉控制系统105也可以是集成在上述麻醉系统100中的某一装置内的。例如， 其可以集成在上述麻醉设备中，以提高仪器的整合程度。在一个示例中，麻醉控制系统可以设置于远程终端。例如，可以设置在医院公共的服务器上，从而能够同时控制多个手术室内的麻醉手术。在另外一个示例中，麻醉控制系统可以以远程通讯的方式，例如5G通讯技术等，设置于云端服务器。
本申请上述配置方式，麻醉控制系统105能够综合判断当前患者的生理状况、多个麻醉设备例如吸入式麻醉设备和注射式麻醉设备的设备参数、患者的麻醉深度指标等，从而给出综合的控制方案，同时对多种麻醉设备的麻醉剂参数进行调节。这样的控制方案，能够同时考虑不同麻醉设备之间的相互影响。例如，吸入式麻醉设备和注射式麻醉设备所采用的不同药物之间共同的镇静/或者镇痛效果。还可以考虑不同麻醉设备之间的麻醉剂的拮抗作用。例如，镇痛类麻醉剂对于镇静效果的影响或者镇静类麻醉剂对于镇痛效果的影响。这样，既不会因为拮抗作用而导致麻醉剂的剂量不足，也不会因为麻醉剂总量使用过量而对患者产生危害。此外，麻醉控制系统105相比于人为的判断和调节也更加具有安全性。
本申请的一些实施例中还提供了一种麻醉控制系统的控制方法。该麻醉系统包括控制器电路用以执行麻醉控制方法。参考图2，示出了本申请一些实施例中麻醉控制系统的控制方法流程图。
在步骤201中，获取患者的多个生理参数。该步骤可以由图1所示的麻醉控制系统105从生理参数采集装置101处获得。具体而言，生理参数采集装置101可以连接并实时获取患者110的生理参数，进一步将该生理参数传送至麻醉控制系统105。
多个生理参数可以包括心率，血氧饱和度，血压，血糖，呼气末二氧化碳浓度，呼气末氧浓度中的多个。发明人发现，患者的生理参数将会对适宜的麻醉剂用量产生影响，并且不同的人群之间也存在着差异。因此，本申请中通过多个生理参数的采集，能够为麻醉控制系统105的麻醉控制提供更为全面、可靠的分析数据。
在步骤202中，获取作用于所述患者的多个麻醉设备的麻醉剂参数。综合参考图1，上述多个麻醉设备可以包括第一麻醉设备102和第二麻醉设备103。
第一麻醉设备102可以是吸入式麻醉设备。相应地，其麻醉剂可以包括七氟醚，地氟醚，异氟醚等。进一步，麻醉剂参数可以是上述一种或者多种麻醉剂的用量(例如，浓度、吸入速率等)。
第二麻醉设备103可以是注射式麻醉设备。相应地，其麻醉剂可以包括海索比妥 钠，美索比妥钠等。进一步，麻醉剂参数可以是上述一种或者多种麻醉剂的注射量(例如，注射速率、麻醉剂浓度等)。
上述麻醉设备中可以自带麻醉剂用量监测模块，并通过该模块将麻醉剂用量传送至麻醉控制系统中。
在步骤203中，获取所述患者的麻醉深度指数。综合参考图1，麻醉深度指数可以通过麻醉深度指数采集装置104处获取。具体而言，麻醉深度指数采集装置104可以包括EEG设备，从而对患者的EEG信号进行实时采集，实时采集得到的EEG信号经过分析、处理即可得到所需麻醉深度指数。麻醉深度指数可以参考本申请上文所描述，通过RE、SE、BIS等本领域的指数进行评价，在此不再赘述。进一步地，上述麻醉深度指数被传送至麻醉控制系统105进行综合处理。
至此，麻醉控制系统105获取到了可以进行控制指令生成的必要参数。在步骤204中，基于所述多个生理参数、所述麻醉剂参数、所述麻醉深度指数以及目标麻醉深度指数自动生成控制指令，所述控制指令用于控制所述多个麻醉设备中的每一个麻醉设备的麻醉剂参数，从而使患者的麻醉深度指数向目标麻醉深度指数调节。
麻醉控制系统105通过对患者生理参数、麻醉设备麻醉剂参数、患者麻醉深度指数以及目标麻醉深度指数的输入能够进行综合的判断。同时考虑多个麻醉设备中的每一个的麻醉剂参数调节。从而避免了孤立的调节可能造成的不利影响。
应当理解，上文示意性地对本申请实施例进行了说明，但本申请不限于此。例如可以适当地调整各个操作之间的执行顺序，此外还可以增加其他的一些操作或者减少其中的某些操作。本领域的技术人员可以根据上述内容进行适当地变型，而不仅限于上述记载。
上述控制指令可以基于经过训练的学习网络而生成。下文将对该学习网络的训练和使用进行示例性描述。
所述学习网络是通过训练数据的准备，学习网络模型的选择和构建，学习网络的训练、测试以及优化完成的。
在一些实施例中，所述学习网络是基于样本的多个参数(已知输入)及其对应的麻醉剂参数(期望输出)的数据集训练得到的，具体地，所述学习网络的训练包括以下步骤一到步骤三。
步骤一：获取多个成功的麻醉案例的不同患者生理参数、麻醉设备麻醉剂参数、 患者麻醉深度指数作为样本参数集。进一步，将该参数集输入到学习网络中，然后基于学习网络中的归一化层进行归一化处理。
步骤二：获取上述成功案例中的每一个的麻醉剂参数集。其中，麻醉剂参数包括多个麻醉设备的每一个的麻醉参数。
步骤三：将所述不同患者生理参数、麻醉设备麻醉剂参数、患者麻醉深度指数的样本参数集作为输入，所述麻醉剂参数集作为输出，训练学习网络，以得到所述训练的学习网络。在一些实施例中，可以基于新的样本集及其对应的麻醉剂参数集对所述训练好的学习网络进行更新。
在一些实施例中，所述学习网络是基于ResNet(Residual Network)或VGGNet(Visual Geometry Group Network)或其他公知的模型进行训练的。由于ResNet中的处理层的数量可以设置的很多(可以达到1000层以上)，相应地，基于该网络结构的分类(例如，伪影类型的判断)的效果也就更好，此外，Resnet也更容易基于更多的训练数据进行学习网络的优化。
图4示出了根据本发明一些实施例的学习网络150。如图4所示，学习网络150包括包括输入层151、处理层(或称为隐藏层)152以及输出层153。在一些实施例中，如图4所示，处理层152包括第一卷积层155，第一池化层156，以及全连接层157。第一卷积层155用于对输入的各个参数进行卷积，得到第一卷积层的特征图。第一池化层156对第一卷积层的特征图进行池化(下采样)，以对第一卷积层的特征图进行压缩以及提取主要特征，以得到第一池化层的特征图。全连接层157可以基于第一池化层的特征图输出判断结果。
虽然图4仅示出了一个卷积层的示例，在其它示例中，卷积层的数量可以为任意个，所述卷积层的数量可以根据学习网络中输入数据的大小等进行适应性调整，例如，在第一池化层156以及全连接层157之间还包括第二卷积层以及第二池化层(图中未示出)，或者，在第一池化层156以及全连接层157之间还包括第二卷积层以及第二池化层，以及第三卷积层以及第三池化层(图中未示出)等等。
虽然图4仅示出了卷积层与输入层连接，池化层与卷积层连接，全连接层与池化层连接，在其它示例中，在上述任一两个层之间可以设置任意数量的任意类型的处理层，例如，在卷积层与输入层之间设置归一化层，以对输入参数进行归一化处理，或者，在全连接层与池化层设置激活层，以利用修正线性单元(Rectified linear unit, ReLU)激活函数对池化层的特征图进行非线性映射。
在一些实施例中，每个层都包括若干个神经元160，且每个层中的神经元的数量可以相同，也可以根据需要进行不同的设置。通过将样本数据集(已知输入)和麻醉剂参数(期望输出)输入到学习网络中，通过设置学习网络中处理层的数量和每个处理层中神经元的数量，并估计(或调整或校准)学习网络的权重和/或偏差，以识别出已知输入和期望输出之间的数学关系和/或识别和表征每层的输入和输出之间的数学关系,学习过程通常利用(部分)输入数据，并为该输入数据创建网络输出，然后基于已知输入对应的创建的网络输出与该数据集的期望输出进行比较，其差值即为损失函数(loss function)，通过损失函数可以迭代地更新网络的参数(权重和/或偏差)，使损失函数不断减小，从而训练出更高的准确率的神经网络模型。在一些实施例中，可以用作损失函数的函数有很多，包括但不限于均方误差(mean suqared)，交叉熵误差(cross entropy error)等。当学习网络创建或训练好时，只要将待生成控制信号的上述患者生理参数、麻醉设备麻醉剂参数、患者麻醉深度指数输入到学习网络中，就可以获取到所需麻醉剂参数。
在一个实施例中，虽然学习网络150的配置将由估计问题的先验知识、输入、输出等的维度引导，但学习本身被视为“黑盒子”，并且主要依赖于或者专门根据输入数据实现所需输出数据的最佳近似。在各种替代实施方式中，可以利用数据，成像几何，重建算法等的某些方面和/或特征来为学习网络150中的某些数据表示赋予明确的含义。这可以有助于加速训练。因为这创建了在学习网络150中单独训练(或预训练)或定义某些层的机会。
可以理解，深度学习方法的特征在于使用一个或多个网络架构来提取或模拟感兴趣数据。深度学习方法可以使用一个或多个处理层(例如，输入层、输出层、卷积层、归一化层、采样层等等，按照不同的深度学习网络模型可以具有不同数量和功能的处理层)来完成，其中层的配置和数量允许深度学习网络处理复杂的信息提取和建模任务。通常通过所谓的学习过程(或训练过程)来估计网络的特定参数(又可以称为“权重”或“偏差”)。被学习或训练的参数通常会导致(或输出)一个对应于不同级别层的网络，因此提取或模拟初始数据的不同方面或前一层的输出通常可以表示层的层次结构或级联。处理可以分层进行，即，较早或较高级别的层可以对应于从输入数据中提取“简单”特征，接着是将这些简单特征组合成表现出更高复杂度的特征的层。实 际上，每层(或者更具体地，每层中的每个“神经元”)可以采用一个或多个线性和/或非线性变换(所谓的激活函数)来将输入数据处理为输出数据表示。多个“神经元”的数量可以是在多个层之间恒定的，或者可以从层到层变化。
如本文所讨论的，作为解决特定问题的深度学习过程的初始训练的一部分，训练数据集包括已知输入值以及深度学习过程的最终输出的期望(目标)输出值(例如，判断结果)。以这种方式，深度学习算法可以(以监督或引导的方式或以无监督或非指导的方式)处理该训练数据集，直到识别出已知输入和期望输出之间的数学关系和/或识别和表征每层的输入和输出之间的数学关系。学习过程通常利用(部分)输入数据，并为该输入数据创建网络输出，然后将创建的网络输出与该数据集的期望输出进行比较，然后使用创建的和期望的输出之间的差异来迭代地更新网络的参数(权重和/或偏差)。通常可以使用随机梯度下降(Stochastic gradient descent,SGD)方法来更新网络的参数，然而本领域技术人员应当理解也可以使用本领域中已知的其他方法来更新网络参数。类似地，可以采用单独的验证数据集以对所训练的学习网络进行验证，其中已知输入和期望输出都是已知的，通过将已知输入提供给所训练的学习网络可以得到网络输出，然后将该网络输出与(已知的)期望输出进行比较以验证先前的培训和/或防止过度培训。
麻醉过程中麻醉控制精度是十分重要的。在本申请的一些实施例中，提供了一些提高控制精度、确保麻醉安全的示例。下面将对这些措施进行示例性描述。
参考图4，示出了根据本申请的另外一些实施例的麻醉控制系统的控制方法流程图400。该示例的控制方法可以在方法流程图200例如步骤204的基础上进行。下文进行详细描述。
在步骤401中，获取所述多个生理参数、所述麻醉深度指数、所述多个麻醉设备的所述麻醉剂参数的更新；基于所述更新来自动生成更新的控制指令。
可以理解，麻醉的过程中，生理参数、麻醉深度指术、麻醉剂参数是不断动态变化的过程。因此，本申请的上述示例公开了根据上述参数的动态变化来实现控制指令的动态更新，从而能够更准确的进行麻醉。上述过程可以通过麻醉控制系统105来实现。动态更新的控制指令能够更加准确的满足适宜麻醉的需求，更好的适应患者个体偶然因素导致的情况变化。
为了进一步的提高麻醉的安全性，在步骤402中，基于所述多个生理参数、所述 麻醉深度指数以及所述麻醉剂参数以及目标麻醉深度指数，生成关于所述麻醉深度指数的更新预测；实时对比所述麻醉深度指数的更新和所述麻醉深度指数的更新预测，若二者大于一定阈值则发出通知。
通过上述配置，麻醉控制系统105可以被用来提供精确的麻醉异常的预测。相比于传统的预测方式中，仅通过直观的生理指标来判断患者是否处于安全的状态，这样不能够有效预知即将发生麻醉风险的可能性。而本申请的麻醉控制系统105，采用例如学习网络的人工智能手段，可以根据当前的各个生理参数、所述麻醉深度指数以及所述麻醉剂参数等因素综合预测出患者未来的麻醉深度指数变化趋势。这样就为实际麻醉深度指数和预测麻醉深度指数的对比提供了条件。一旦对比结果显示二者偏差较大，则说明麻醉风险较高，此时可以以警报等方式提醒医生的注意，极大提高了麻醉的安全性。可以理解，上述阈值可以通过人工设定，也可以是麻醉控制系统105自主判断，不再赘述。
进一步，在步骤403中，确定麻醉剂参数阈值，比较所述麻醉剂参数阈值和所述麻醉剂参数，若麻醉剂参数超过所述阈值则发出通知。
上述配置能够进一步提高安全性。尤其是在全自动的麻醉过程中，麻醉控制系统105的控制指令不仅仅关注于麻醉深度指数是否达标，还能够确保在麻醉深度指数达标的基础上各麻醉剂用量不会过量。一旦过量，将向操作者发出通知。类似于步骤402，步骤403中的上述阈值可以通过人工设定，也可以是麻醉控制系统105自主判断。
应当理解，上文示意性地对本申请实施例进行了说明，但本申请不限于此。例如可以适当地调整各个操作之间的执行顺序，此外还可以增加其他的一些操作或者减少其中的某些操作。本领域的技术人员可以根据上述内容进行适当地变型，而不仅限于上述记载。
本申请一些实施例提供了上述麻醉控制系统105。可以理解，麻醉控制系统105可以是任意用于执行本申请任意实施例中所记载的方法的系统。如上文所记载，麻醉控制系统105的种类可以是多样的，例如，其可以是包括处理器在内的控制器电路。该控制器电路可以被配置在独立存在的控制系统内。此外，麻醉控制系统105也可以是集成在上述麻醉系统100中的某一装置内的。例如，其可以集成在上述麻醉设备中。在一个示例中，麻醉控制系统可以设置于远程终端。例如，可以设置在医院公共的服 务器上，从而能够同时控制多个手术室内的麻醉手术。在另外一个示例中，麻醉控制系统可以以远程通讯的方式，例如5G通讯技术等，设置于云端服务器。
本申请一些实施例还提供了一种非暂态计算机可读介质。所述非暂态计算机可读介质存储有计算机程序，所述计算机程序具有至少一个代码段，所述至少一个代码段能够由机器执行以：获取患者的多个生理参数；获取作用于所述患者的多个麻醉设备的麻醉剂参数；获取所述患者的麻醉深度指数；以及基于所述多个生理参数、所述麻醉剂参数、所述麻醉深度指数以及目标麻醉深度指数自动生成控制指令，所述控制指令用于控制所述多个麻醉设备中的每一个麻醉设备的麻醉剂参数，从而使患者的麻醉深度指数向目标麻醉深度指数调节。
可选地，所述控制指令基于训练的学习网络而自动生成。
可选地，所述多个麻醉设备的麻醉剂参数包括吸入式麻醉设备的麻醉剂参数和注射式麻醉设备的麻醉剂参数。
可选地，所述麻醉深度指数包括镇静指数，镇痛指数和肌肉松弛指数中的至少一个。
可选地，所述多个生理参数包括心率，血氧饱和度，血压，血糖，呼气末二氧化碳浓度，呼气末氧浓度中的多个。
可选地，非暂态计算机可读介质进一步被配置为：获取所述多个生理参数、所述麻醉深度指数、所述多个麻醉设备的所述麻醉剂参数的更新；以及基于所述更新来自动生成更新的控制指令。
可选地，非暂态计算机可读介质进一步被配置为：基于所述多个生理参数、所述麻醉深度指数以及所述麻醉剂参数以及目标麻醉深度指数，生成关于所述麻醉深度指数的更新预测；实时对比所述麻醉深度指数的更新和所述麻醉深度指数的更新预测，若二者大于一定阈值则发出通知。
可选地，非暂态计算机可读介质进一步被配置为：确定麻醉剂参数阈值，比较所述麻醉剂参数阈值和所述麻醉剂参数，若麻醉剂参数超过所述阈值则发出通知。
提供以上具体的实施例的目的是为了使得对本申请的公开内容的理解更加透彻全面，但本申请并不限于这些具体的实施例。本领域技术人员应理解，还可以对本申请做各种修改、等同替换和变化等等，只要这些变换未违背本申请的精神，都应在本申请的保护范围之内。

Claims (17)

  1. 一种麻醉控制系统，所述麻醉控制系统被配置为：
    获取患者的多个生理参数；
    获取作用于所述患者的多个麻醉设备的麻醉剂参数；
    获取所述患者的麻醉深度指数；以及
    基于所述多个生理参数、所述麻醉剂参数、所述麻醉深度指数以及目标麻醉深度指数自动生成控制指令，所述控制指令用于控制所述多个麻醉设备中的每一个麻醉设备的麻醉剂参数，从而使患者的麻醉深度指数向目标麻醉深度指数调节。
  2. 根据权利要求1所述的麻醉控制系统，其中：
    所述控制指令基于训练的学习网络而自动生成。
  3. 根据权利要求1所述的麻醉控制系统，其中：
    所述多个麻醉设备的麻醉剂参数包括吸入式麻醉设备的麻醉剂参数和注射式麻醉设备的麻醉剂参数。
  4. 根据权利要求1所述的麻醉控制系统，其中：
    所述麻醉深度指数包括镇静指数，镇痛指数和肌肉松弛指数中的至少一个。
  5. 根据权利要求1所述的麻醉控制系统，其中：
    所述多个生理参数包括心率，血氧饱和度，血压，血糖，呼气末二氧化碳浓度，呼气末氧浓度中的多个。
  6. 根据权利要求1所述的麻醉控制系统，其中，所述麻醉控制系统进一步被配置为：
    获取所述多个生理参数、所述麻醉深度指数、所述多个麻醉设备的所述麻醉剂参数的更新；以及
    基于所述更新来自动生成更新的控制指令。
  7. 根据权利要求6所述的麻醉控制系统，其中，所述麻醉控制系统进一步被配置为：
    基于所述多个生理参数、所述麻醉深度指数以及所述麻醉剂参数以及目标麻醉深度指数，生成关于所述麻醉深度指数的更新预测；
    实时对比所述麻醉深度指数的更新和所述麻醉深度指数的更新预测，若二者大于 一定阈值则发出通知。
  8. 根据权利要求1所述的麻醉控制系统，其中，所述麻醉控制系统进一步被配置为：
    确定麻醉剂参数阈值，比较所述麻醉剂参数阈值和所述麻醉剂参数，若麻醉剂参数超过所述阈值则发出通知。
  9. 一种麻醉系统，包括：
    生理参数采集装置，用于采集患者的多个生理参数；
    多个麻醉设备，用于将麻醉剂作用于所述患者；
    麻醉深度指数采集装置，用于采集所述患者的麻醉深度指数；以及
    如权利要求1-8任一项所述的麻醉控制系统，所述麻醉控制系统和所述生理参数采集装置、所述多个麻醉设备以及所述麻醉深度指数采集装置连接。
  10. 一种非暂态计算机可读介质，所述非暂态计算机可读介质存储有计算机程序，所述计算机程序具有至少一个代码段，所述至少一个代码段能够由机器执行以：
    获取患者的多个生理参数；
    获取作用于所述患者的多个麻醉设备的麻醉剂参数；
    获取所述患者的麻醉深度指数；以及
    基于所述多个生理参数、所述麻醉剂参数、所述麻醉深度指数以及目标麻醉深度指数自动生成控制指令，所述控制指令用于控制所述多个麻醉设备中的每一个麻醉设备的麻醉剂参数，从而使患者的麻醉深度指数向目标麻
    醉深度指数调节。
  11. 根据权利要求10所述的非暂态计算机可读介质，其中：
    所述控制指令基于训练的学习网络而自动生成。
  12. 根据权利要求10所述的非暂态计算机可读介质，其中：
    所述多个麻醉设备的麻醉剂参数包括吸入式麻醉设备的麻醉剂参数和注射式麻醉设备的麻醉剂参数。
  13. 根据权利要求10所述的非暂态计算机可读介质，其中：
    所述麻醉深度指数包括镇静指数，镇痛指数和肌肉松弛指数中的至少一个。
  14. 根据权利要求10所述的非暂态计算机可读介质，其中：
    所述多个生理参数包括心率，血氧饱和度，血压，血糖，呼气末二氧化碳浓度， 呼气末氧浓度中的多个。
  15. 根据权利要求10所述的非暂态计算机可读介质，进一步被配置为：
    获取所述多个生理参数、所述麻醉深度指数、所述多个麻醉设备的所述麻醉剂参数的更新；以及
    基于所述更新来自动生成更新的控制指令。
  16. 根据权利要求15所述的非暂态计算机可读介质，进一步被配置为：
    基于所述多个生理参数、所述麻醉深度指数以及所述麻醉剂参数以及目标麻醉深度指数，生成关于所述麻醉深度指数的更新预测；
    实时对比所述麻醉深度指数的更新和所述麻醉深度指数的更新预测，若二者大于一定阈值则发出通知。
  17. 根据权利要求10所述的非暂态计算机可读介质，进一步被配置为：
    确定麻醉剂参数阈值，比较所述麻醉剂参数阈值和所述麻醉剂参数，若麻醉剂参数超过所述阈值则发出通知。
PCT/CN2021/143896 2021-12-31 2021-12-31 麻醉控制系统，麻醉系统及非暂态计算机可读介质 WO2023123437A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2021/143896 WO2023123437A1 (zh) 2021-12-31 2021-12-31 麻醉控制系统，麻醉系统及非暂态计算机可读介质
CN202180104853.8A CN118354807A (zh) 2021-12-31 2021-12-31 麻醉控制系统，麻醉系统及非暂态计算机可读介质

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2021/143896 WO2023123437A1 (zh) 2021-12-31 2021-12-31 麻醉控制系统，麻醉系统及非暂态计算机可读介质

Publications (1)

Publication Number Publication Date
WO2023123437A1 true WO2023123437A1 (zh) 2023-07-06

Family

ID=86997151

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/143896 WO2023123437A1 (zh) 2021-12-31 2021-12-31 麻醉控制系统，麻醉系统及非暂态计算机可读介质

Country Status (2)

Country Link
CN (1) CN118354807A (zh)
WO (1) WO2023123437A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117017234A (zh) * 2023-10-09 2023-11-10 深圳市格阳医疗科技有限公司 多参数集成麻醉监测与分析系统
CN117814760A (zh) * 2024-03-04 2024-04-05 江西杰联医疗设备有限公司 基于多指标的麻醉深度检测装置及电子设备
CN117838063A (zh) * 2024-03-04 2024-04-09 江西杰联医疗设备有限公司 麻醉场景下的生理信息预警处理系统及电子设备

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020117176A1 (en) * 1996-09-11 2002-08-29 Haralambos Mantzaridis Anaesthesia control system
US20040079372A1 (en) * 2002-10-23 2004-04-29 John Erwin R. System and method for guidance of anesthesia, analgesia and amnesia
CN104869897A (zh) * 2012-10-12 2015-08-26 通用医疗公司 用于在施用麻醉化合物期间和之后监测和控制患者的状态的系统和方法
CN104887225A (zh) * 2015-06-04 2015-09-09 卞汉道 麻醉精度监护仪器及方法
CN110349642A (zh) * 2019-07-09 2019-10-18 泰康保险集团股份有限公司 智能麻醉实施方法、装置、设备及存储介质
CN111991664A (zh) * 2020-09-04 2020-11-27 深圳市威浩康医疗器械有限公司 麻醉智能辅助控制系统运行方法、电子装置及存储介质
CN112057701A (zh) * 2020-09-04 2020-12-11 深圳市威浩康医疗器械有限公司 舒适麻醉辅助控制系统运行方法、电子装置及存储介质
CN113133744A (zh) * 2021-04-30 2021-07-20 鹤壁市人民医院 一种多功能麻醉科用麻醉深度监测装置

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020117176A1 (en) * 1996-09-11 2002-08-29 Haralambos Mantzaridis Anaesthesia control system
US20040079372A1 (en) * 2002-10-23 2004-04-29 John Erwin R. System and method for guidance of anesthesia, analgesia and amnesia
CN104869897A (zh) * 2012-10-12 2015-08-26 通用医疗公司 用于在施用麻醉化合物期间和之后监测和控制患者的状态的系统和方法
CN104887225A (zh) * 2015-06-04 2015-09-09 卞汉道 麻醉精度监护仪器及方法
CN110349642A (zh) * 2019-07-09 2019-10-18 泰康保险集团股份有限公司 智能麻醉实施方法、装置、设备及存储介质
CN111991664A (zh) * 2020-09-04 2020-11-27 深圳市威浩康医疗器械有限公司 麻醉智能辅助控制系统运行方法、电子装置及存储介质
CN112057701A (zh) * 2020-09-04 2020-12-11 深圳市威浩康医疗器械有限公司 舒适麻醉辅助控制系统运行方法、电子装置及存储介质
CN113133744A (zh) * 2021-04-30 2021-07-20 鹤壁市人民医院 一种多功能麻醉科用麻醉深度监测装置

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117017234A (zh) * 2023-10-09 2023-11-10 深圳市格阳医疗科技有限公司 多参数集成麻醉监测与分析系统
CN117017234B (zh) * 2023-10-09 2024-01-16 深圳市格阳医疗科技有限公司 多参数集成麻醉监测与分析系统
CN117814760A (zh) * 2024-03-04 2024-04-05 江西杰联医疗设备有限公司 基于多指标的麻醉深度检测装置及电子设备
CN117838063A (zh) * 2024-03-04 2024-04-09 江西杰联医疗设备有限公司 麻醉场景下的生理信息预警处理系统及电子设备
CN117814760B (zh) * 2024-03-04 2024-05-17 江西杰联医疗设备有限公司 基于多指标的麻醉深度检测装置及电子设备
CN117838063B (zh) * 2024-03-04 2024-05-24 江西杰联医疗设备有限公司 麻醉场景下的生理信息预警处理系统及电子设备

Also Published As

Publication number Publication date
CN118354807A (zh) 2024-07-16

Similar Documents

Publication Publication Date Title
WO2023123437A1 (zh) 麻醉控制系统，麻醉系统及非暂态计算机可读介质
Ramírez-Carrasco et al. Effectiveness of hypnosis in combination with conventional techniques of behavior management in anxiety/pain reduction during dental anesthetic infiltration
US9474452B2 (en) Method and system for quantifying anaesthesia or a state of vigilance
Harrison et al. The Filter Detection Task for measurement of breathing-related interoception and metacognition
Wysocki et al. Chaotic dynamics of resting ventilatory flow in humans assessed through noise titration
US20150019237A1 (en) Holistic patient advisory system, method, and software
KR20210066271A (ko) 마취 분야에서의 의료 딥러닝을 활용한 처방 시스템
CN114177417B (zh) 一种基于脑电参数反馈的麻醉靶控智能化输注泵控制器
Guy et al. Quantifying ventilator unloading in CPAP ventilation
Liu et al. HRV-derived data similarity and distribution index based on ensemble neural network for measuring depth of anaesthesia
Yu et al. An adaptive monitoring scheme for automatic control of anaesthesia in dynamic surgical environments based on bispectral index and blood pressure
CN116548920A (zh) 一种基于神经网络的麻醉状态监测系统及方法
Kanbar et al. Cardiorespiratory behavior of preterm infants receiving continuous positive airway pressure and high flow nasal cannula post extubation: randomized crossover study
CN115299881A (zh) 一种麻醉维持期的监测与调控系统及方法
CN115551579B (zh) 用于评估通风患者状况的系统和方法
CN117219227B (zh) 一种基于模糊神经网络的麻醉给药控制系统
CN115211816A (zh) 监护方法及监护系统
CN114730622A (zh) 用于训练可用于计算伤害感受指数的模型的方法
Rapoport On beyond zebra (and the apnea–hypopnea index) in obstructive sleep apnea
Tehrany et al. Breathing pattern recordings using respiratory inductive plethysmography, before and after a physiotherapy breathing retraining program for asthma: A case report
US20230201504A1 (en) System and method for generating patient-specific ventilation settings based on lung modeling
Beck et al. Modeling and control of anesthetic pharmacodynamics
Komorowski et al. AIM in Anesthesiology
Vajapey Predicting optimal sedation control with reinforcement learning
TOMA A Comprehensive Review on Automated Control of Anesthesia: Recent Methods, Challenges and Future Trends

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: 21969782

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2021969782

Country of ref document: EP

Effective date: 20240731