Search

CN-122004896-A - Unintended tube drawing early warning method and related equipment for tracheal intubation patient

CN122004896ACN 122004896 ACN122004896 ACN 122004896ACN-122004896-A

Abstract

The invention discloses an unscheduled tube drawing early warning method and related equipment for a patient with trachea cannula, the method discloses an unscheduled tube drawing early warning method for a patient with trachea cannula, electroencephalogram data of the patient are acquired through an electroencephalogram signal acquisition module, and denoising, filtering, artifact removal and sliding window segmentation processing are carried out on the original electroencephalogram signals, time domain, frequency domain and time domain electroencephalogram characteristics are extracted, and a pre-trained tube drawing intention recognition model is utilized to calculate a tube drawing intention probability value. On the basis, the multi-parameter fusion decision is carried out by combining clinical parameters such as vital signs, sedation depth and pipeline fixation state of a patient, the risk level of unplanned tube drawing is determined, and a grading early warning and recording mechanism is triggered. The invention advances the early warning time from the behavior generation stage to the intention formation stage, realizes the early recognition of the tube drawing risk, has the advantages of high early warning timeliness, strong anti-interference capability and good clinical suitability, and can effectively reduce the occurrence rate of unplanned tube drawing events.

Inventors

  • Yang Xueke
  • LIU SHUO

Assignees

  • 华中科技大学同济医学院附属同济医院

Dates

Publication Date
20260512
Application Date
20260317

Claims (10)

  1. 1. An unintended extubation early warning method for a patient with trachea cannula is characterized by comprising the following steps: collecting an electroencephalogram signal of a patient with tracheal intubation through an electroencephalogram signal collecting module, and outputting original electroencephalogram signal data; Performing signal preprocessing operation on the original electroencephalogram signal data to obtain preprocessed electroencephalogram signals, wherein the preprocessing operation comprises denoising, filtering, artifact removal and sliding window segmentation processing; extracting a tube drawing intention related electroencephalogram characteristic from the preprocessed electroencephalogram signal, and inputting the electroencephalogram characteristic into a pre-trained tube drawing intention recognition model for classification recognition so as to obtain a tube drawing intention probability value; Acquiring clinical parameters of a patient, and carrying out multi-parameter fusion calculation on the probability value of the tube drawing intention and the clinical parameters of the patient so as to determine an unplanned tube drawing risk level; Triggering corresponding early warning actions according to the unplanned tube drawing risk level, and recording early warning information corresponding to the early warning actions.
  2. 2. The method for unintended extubation pre-warning of tracheal cannula patient according to claim 1, wherein the performing signal pre-processing operation on the raw electroencephalogram signal data to obtain a pre-processed electroencephalogram signal comprises: Performing 50Hz notch filtering on the original electroencephalogram signal data to remove power frequency interference; Carrying out 0.5 Hz-30 Hz band-pass filtering on the original electroencephalogram signal data to extract an electroencephalogram effective frequency band; removing eye electric artifacts, myoelectric artifacts and electrocardio artifacts in the electroencephalogram signals by using an independent component analysis method; And carrying out segmentation processing on the filtered electroencephalogram signals by adopting a sliding window method, wherein the sliding window has a length of 2-5 seconds and an overlapping rate of 50-70%.
  3. 3. The method for unintended extubation pre-warning of tracheal cannula patient according to claim 1, wherein extracting the extubation intention related electroencephalogram features from the preprocessed electroencephalogram signals and inputting the electroencephalogram features into a pre-trained extubation intention recognition model for classification recognition to obtain extubation intention probability values comprises: Extracting time domain features of the electroencephalogram signals, wherein the time domain features comprise peaks, average values, variances, kurtosis and skewness; Extracting frequency domain features of the electroencephalogram signals, wherein the frequency domain features comprise mu wave power, beta wave power, mu wave inhibition rate and beta/mu power ratio; extracting time-frequency domain characteristics of the electroencephalogram signals, and obtaining energy of each frequency band through a wavelet packet decomposition technology; Inputting the time domain features, the frequency domain features and the time-frequency domain features into a pre-trained tube drawing intention recognition model, recognizing the tube drawing intention, and outputting a tube drawing intention probability value.
  4. 4. The method for pre-planning a tube drawing for an endotracheal intubation patient according to claim 3, wherein the specific structure of the pre-trained tube drawing intention recognition model comprises: an input layer, configured to receive the time domain feature, the frequency domain feature, and the time-frequency domain feature, where a total dimension of the time domain feature, the frequency domain feature, and the time-frequency domain feature is 32 dimensions; The convolution layer comprises two convolution layers, the convolution kernel of the convolution layer is 3 multiplied by 3, and the activation function is ReLU; the pooling layer adopts the maximum pooling operation, and the pooling core size is 2 multiplied by 2; the LSTM layer comprises a hidden layer, the node number is 64, and a tanh activation function is adopted; the full-connection layer comprises 32 nodes, an activation function is a ReLU, and the full-connection layer is used for mapping the characteristics output by the LSTM layer to a final classification space; An output layer comprising 2 nodes, respectively representing a tube drawing intention and a tube drawing no intention, wherein an activation function is Softmax; in the training process of the pre-trained tube drawing intention recognition model, a cross entropy loss function is used, an Adam optimizer is adopted for training, the learning rate is set to be 0.001, and the batch processing size is set to be 32.
  5. 5. The method of claim 1, wherein the acquiring patient clinical parameters and performing a multi-parameter fusion calculation of the probability value of tube drawing intention and the patient clinical parameters to determine an unplanned tube drawing risk level comprises: acquiring vital sign parameters of the patient, wherein the vital sign parameters comprise heart rate, blood pressure, respiratory rate and blood oxygen saturation; obtaining a sedation depth parameter of the patient, wherein the sedation depth parameter comprises a RASS score and a BIS score; Acquiring a pipeline fixing state parameter of the patient, wherein the pipeline fixing state parameter comprises the depth of an endotracheal intubation and the tension of a fixing belt; Carrying out weighted fusion calculation on the extubation intention probability value, the vital sign parameter, the sedation depth parameter and the pipeline fixed state parameter by a weighted voting method to obtain a fusion decision value; And grading the risk of the unplanned tube drawing according to the fusion decision value and a preset threshold value, and determining the risk grade of the unplanned tube drawing.
  6. 6. The method for pre-warning an airway tube according to claim 1, wherein triggering a corresponding pre-warning action according to the risk level of the airway tube and recording pre-warning information corresponding to the pre-warning action comprises: under the condition that the risk level of the unplanned tube drawing is low, only recording the early warning information in a data storage module, and not triggering audible and visual alarm; Under the condition that the risk level of the unplanned tube drawing is low and medium, the nursing working platform pops up prompt information, and the green indicator lamp is normally on; Under the condition that the risk level of the unplanned tube drawing is medium and high, the nursing working platform pops up warning information and triggers buzzing alarm, and a yellow indicator lamp flashes; Under the condition that the risk level of the unplanned tube drawing is high, the nursing working platform pops up emergency warning information, triggers high-frequency buzzing alarm, flashes a red indicator lamp and pushes early warning information to a mobile phone APP of a responsible nurse; Recording the early warning information and the intervention record of medical staff, storing the early warning information and the intervention record of the medical staff in a data storage module, and supporting the data export through a USB interface or wireless communication.
  7. 7. The method of unscheduled extubation pre-warning for an endotracheal intubation patient according to claim 1, wherein the pre-trained extubation intent recognition model further comprises: Generating a model feedback label based on the medical staff intervention record after the early warning action is triggered and the result of whether the unplanned tube drawing actually occurs; Correlating the model feedback label with the preprocessed electroencephalogram signals and the electroencephalogram characteristics in the corresponding time period to construct an incremental learning sample; performing periodic parameter updating or threshold self-adaptive adjustment on the pre-trained tube drawing intention recognition model according to the increment learning sample so as to realize dynamic optimization aiming at different sedation levels, different recognition states and different individual differences; And after the parameter updating or the threshold self-adaptive adjustment is completed, reusing the updated model for calculating the probability value of the tube drawing intention so as to improve the individuation accuracy and the long-term stability of the unplanned tube drawing early warning.
  8. 8. An unscheduled tube drawing early warning system for a patient with an endotracheal tube, comprising: The acquisition unit is used for acquiring the brain electrical signals of the tracheal cannula patient through the brain electrical signal acquisition module and outputting original brain electrical signal data; the preprocessing unit is used for carrying out signal preprocessing operation on the original electroencephalogram signal data to obtain preprocessed electroencephalogram signals, wherein the preprocessing operation comprises denoising, filtering, artifact removal and sliding window segmentation processing; The extraction unit is used for extracting the brain electrical characteristics related to the tube drawing intention from the preprocessed brain electrical signals, inputting the brain electrical characteristics into a pre-trained tube drawing intention recognition model for classification recognition so as to obtain a tube drawing intention probability value; the determining unit is used for acquiring clinical parameters of a patient, carrying out multi-parameter fusion calculation on the probability value of the tube drawing intention and the clinical parameters of the patient, and determining the risk level of unplanned tube drawing; And the trigger recording unit is used for triggering corresponding early warning actions according to the unplanned tube drawing risk level and recording early warning information corresponding to the early warning actions.
  9. 9. An electronic device comprising a memory and a processor, wherein the processor is adapted to perform the steps of the method for unintended extubation of an endotracheal intubation patient according to any of claims 1-7 when executing a computer program stored in the memory.
  10. 10. A computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor performs the steps of the method for unintended extubation of an endotracheal tube patient according to any of claims 1 to 7.

Description

Unintended tube drawing early warning method and related equipment for tracheal intubation patient Technical Field The specification relates to the field of intelligent medical treatment, in particular to an unscheduled tube drawing early warning method and related equipment for a patient with an endotracheal tube. Background Tracheal intubation is an important therapeutic measure for critical patients to maintain respiratory function, but as a clinically common adverse event, unplanned extubation is easy to cause serious consequences such as acute hypoxia, respiratory failure, airway injury and the like, and the re-intubation rate and medical risk are obviously increased. At present, early warning, prevention and control are mainly carried out in clinic through physical constraint, pipeline sensor monitoring, nursing grading, vital sign monitoring and other modes, but the problems of early warning lag, poor anti-interference capability, dependence on manual experience or insufficient specificity and the like of the method generally exist, and accurate identification of early extubation intention of a patient is difficult to realize. The brain-computer interface technology can acquire the neural activity information in the action preparation stage by analyzing the brain-computer signal, and is applied to the fields of rehabilitation and neural regulation, and researches show that the human body can generate characteristic brain-computer signal change before performing actions such as grasping and the like. Therefore, the brain-computer interface technology is introduced into the unplanned tube drawing early warning of the tracheal intubation patient, and the transition from behavior monitoring to intention level early warning is hopefully realized by identifying the tube drawing related electroencephalogram intention signal, so that the early warning advancement and accuracy are improved. However, there is no systematic solution in the prior art that combines brain-computer interfaces with an unscheduled extubation risk pre-warning of the endotracheal tube. Therefore, there is a need to provide an unintended extubation early warning method and related equipment for tracheal intubation patients. Disclosure of Invention In the summary, a series of concepts in a simplified form are introduced, which will be further described in detail in the detailed description. The summary of the application is not intended to define the key features and essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter. In a first aspect, the present application provides an unintended extubation early warning method for a patient with tracheal intubation, comprising: collecting an electroencephalogram signal of a patient with tracheal intubation through an electroencephalogram signal collecting module, and outputting original electroencephalogram signal data; performing signal preprocessing operation on the original electroencephalogram signal data to obtain preprocessed electroencephalogram signals, wherein the preprocessing operation comprises denoising, filtering, artifact removal and sliding window segmentation processing; Extracting a tube drawing intention related electroencephalogram characteristic from the preprocessed electroencephalogram signal, and inputting the electroencephalogram characteristic into a pre-trained tube drawing intention recognition model for classification recognition so as to obtain a tube drawing intention probability value; acquiring clinical parameters of a patient, and carrying out multi-parameter fusion calculation on the probability value of the tube drawing intention and the clinical parameters of the patient so as to determine an unplanned tube drawing risk level; Triggering corresponding early warning actions according to the unplanned tube drawing risk level, and recording early warning information corresponding to the early warning actions. In a possible implementation manner, the performing a signal preprocessing operation on the raw electroencephalogram signal data to obtain a preprocessed electroencephalogram signal includes: Performing 50Hz notch filtering on the original EEG signal data to remove power frequency interference; carrying out 0.5 Hz-30 Hz band-pass filtering on the original electroencephalogram signal data to extract an electroencephalogram effective frequency band; Removing the electro-oculogram artifact, myoelectric artifact and electrocardio artifact in the electroencephalogram signal by using an independent component analysis method; And carrying out sectional processing on the filtered electroencephalogram signals by adopting a sliding window method, wherein the sliding window has a length of 2-5 seconds and an overlapping rate of 50-70%. In a possible implementation manner, the extracting the electroencephalogram feature related to the tube drawing intention from the preprocessed el