Trachea cannula air bag pressure intelligent nursing system based on edge calculation

Through edge computing and intelligent algorithms, personalized, real-time monitoring and automatic adjustment of airbag pressure intratracheal intubation is achieved, solving the problem of insufficient real-time and delayed response of manual operations in traditional care, reducing the risk of VAP and increasing the airbag pressure Real-time and accuracy of management.

CN118987431APending Publication Date: 2024-11-22FUWAI CENTRAL CHINA CARDIOVASCULAR HOSPITAL
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411110999.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-14
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The monitoring and adjustment of traditional tracheal intubation airbag pressures has problems with insufficient real-time and delayed responses, which cannot promptly reflect changes in airbag pressure, resulting in airbag pressure being out of the safe range, increasing the risk of ventilator-associated pneumonia (VAP).

Method used

Using the intelligent care system for tracheal intubation airbag pressure based on edge computing and intelligent algorithms, the patient's physiological status data is obtained in real time through the data acquisition module, and the edge computing equipment is cleaned, denoised and normalized, using Bayesian formulas and Marco The Fulink Monte Carlo method predicts the time and range of airbag pressure adjustment to achieve personalized, real-time monitoring and automatic adjustment of airbag pressure.

Benefits of technology

It effectively avoids tracheal damage and air leakage caused by excessive or low airbag pressure, reduces the risk of VAP, improves the real-time and accuracy of airbag pressure management, and significantly improves the comfort and safety of patients.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118987431A_ABST
    Figure CN118987431A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of breathing machine related pneumonia nursing, and particularly discloses a trachea cannula air bag pressure intelligent nursing system based on edge calculation, which is used for solving the problems of insufficient real-time performance and delayed response caused by dependence on manual operation on air bag pressure monitoring and adjustment in a traditional nursing environment. The problems that data islands among devices cause unsmooth information, airbag pressure is difficult to keep in a safe range, and individual differences of patients are ignored based on setting of general standards are solved, and the device comprises a data acquisition module, an edge computing device, a communication module and an automatic airbag pressure regulating valve; personalized, real-time monitoring and automatic adjustment of trachea cannula air bag pressure are achieved through edge calculation and an intelligent algorithm, the problems of insufficient real-time performance and delayed response of manual operation are solved, the VAP risk is reduced, dynamic optimization is conducted through physiological data of a patient, damage caused by the air bag pressure is avoided, and the safety of the patient is improved. And the problems of data transmission delay and information isolated island are solved through distributed processing.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of ventilator-associated pneumonia nursing, and more specifically, to an edge computing-based endotracheal tube airbag pressure intelligent nursing system. Background Art

[0002] Ventilator-associated pneumonia (VAP) is a common and serious complication in patients receiving mechanical ventilation. It occurs 48 hours after mechanical ventilation or within 48 hours of extubation. VAP significantly increases the patient's mortality rate, length of hospital stay and medical expenses. The endotracheal tube balloon is a device used to ensure the airtightness and stability of the endotracheal tube during endotracheal intubation. It is a part of the endotracheal tube used to seal the trachea. The balloon is connected to the outside of the endotracheal tube through an inflation tube. The nursing staff injects or discharges air through this tube to adjust the balloon pressure. The balloon fills the space between the tracheal wall and the tube by inflation to prevent gas leakage, help maintain effective respiratory support, and help prevent gastric contents or saliva from being sucked into the respiratory tract, thereby reducing the risk of infection in VAP. It can also stabilize the position of the endotracheal tube. , reduce the displacement of the intubation caused by movement or coughing of the patient, the inflation pressure of the airbag needs to be maintained within an appropriate range, too high will cause tracheal compression and damage, too low will cause leakage and aspiration, and the pressure of the airbag is generally adjusted and monitored through the inflation pipe outside the airbag. In the traditional nursing environment, the pressure monitoring and adjustment of the airbag usually rely on manual operation, which has the problems of insufficient real-time performance and delayed response. The existence of data islands between devices will lead to poor information flow between devices and the inability to reflect the changes in the airbag pressure in time, which may easily lead to the airbag pressure being out of the safe range. When uploading data from the edge device to the automatic control device, there is a delay in data transmission and it is impossible to respond quickly. The traditional airbag pressure setting is based on universal standards, ignoring the individual differences in the patient's physiological state. Summary of the invention

[0003] In order to overcome the above-mentioned defects of the prior art, the present invention provides an intelligent nursing system for endotracheal intubation cuff pressure based on edge computing, which realizes personalized, real-time monitoring and automatic adjustment of endotracheal intubation cuff pressure through edge computing and intelligent algorithms, solves the problems of insufficient real-time performance and delayed response of manual operation, reduces the risk of VAP, and uses the patient's physiological data for dynamic optimization to avoid damage caused by excessive or low cuff pressure. It also solves the problems of data transmission delay and information islands through distributed processing, thereby improving the real-time and accuracy of management.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] An edge computing-based endotracheal intubation airbag pressure intelligent nursing system comprises a data acquisition module, an edge computing device, a communication module and an airbag pressure automatic regulating valve. The data acquisition module is connected to the edge computing device through the communication module, and the edge computing device is connected to the airbag pressure automatic regulating valve through the communication module. The edge computing device cleans, denoises and normalizes the acquired patient physiological state data. Based on the processed patient physiological state data, a likelihood function is obtained through a relationship function between a predefined physiological parameter and an airbag pressure adjustment time. Then, the posterior distribution of the model parameters is updated according to the Bayesian formula. The posterior distribution is sampled to obtain new model parameters. The sampled model parameters are used to obtain the expected value and credible interval of the predicted airbag pressure time adjustment range. Then, according to the predicted time range and current physiological parameters, based on a preset objective function, an airbag pressure adjustment prediction value is obtained. The airbag pressure automatic regulating valve adjusts the airbag pressure of the endotracheal intubation connected to the ventilator in real time according to the airbag pressure adjustment time prediction value and the airbag pressure adjustment amount prediction value, and completes the adjustment of the airbag pressure within the predicted airbag pressure adjustment time prediction value.

[0006] As a further solution of the present invention, the data acquisition module includes a physiological parameter monitoring device and an airbag pressure sensor. The physiological parameter monitoring data obtains the patient's heart rate, blood oxygen saturation, respiratory rate and blood pressure in real time. The airbag pressure sensor monitors the endotracheal tube airbag pressure connected to the ventilator in real time.

[0007] As a further solution of the present invention, the edge computing device includes a data preprocessing module, an airbag pressure adjustment parameter prediction module and a real-time decision module. The data preprocessing module is connected to the airbag pressure adjustment parameter prediction module, and the airbag pressure adjustment parameter prediction module is connected to the real-time decision module.

[0008] As a further solution of the present invention, the specific process of the airbag pressure adjustment parameter prediction module obtaining the airbag pressure adjustment time prediction value and the airbag pressure adjustment amount prediction value includes:

[0009] Step 1, data input: input the pre-processed heart rate, blood oxygen saturation, respiratory rate, blood pressure and the airbag pressure at the previous moment;

[0010] Step 2, calculating the likelihood function: obtaining the likelihood function of the pressure adjustment prediction value according to the preset physiological parameters and the time prediction function of the airbag adjustment time;

[0011] Step 3, verify the distribution after updating: According to the Bayesian formula, verify the probability distribution of the likelihood function of the pressure adjustment prediction value after updating the model parameters;

[0012] Step 4: Markov Chain Monte Carlo method sampling: Use the Markov Chain Monte Carlo method to sample from the posterior distribution to obtain a new set of model parameters;

[0013] Step 5, predict the airbag adjustment time range and airbag pressure adjustment value: use the sampled model parameters to obtain the expected value and credible interval of the predicted adjustment time range, and then predict the airbag pressure adjustment prediction value through the preset objective function based on the predicted adjustment time range and current physiological parameters.

[0014] As a further solution of the present invention, in step 2, the time prediction function of the preset physiological parameters and the airbag adjustment time is composed of a single influence term and a cross-image term. The single influence term considers the independent influence of each physiological parameter on the adjustment time. The expression of the single influence term is:

[0015]

[0016] Where: S single is a single influencing item, k1, k2, k3, and k4 are the heart rate, blood oxygen saturation, respiratory rate, and blood pressure after pretreatment, respectively;

[0017] The cross-influence term is a vector composed of the interaction term between heart rate and blood oxygen saturation, the interaction term between heart rate and respiratory rate, the interaction term between heart rate and blood pressure, the interaction term between blood oxygen saturation and respiratory rate, the interaction term between blood oxygen saturation and blood pressure, and the interaction term between respiratory rate and blood pressure. The expression of the cross-influence term is:

[0018] S inter =(k1·k2, k1·k3, k1·k4, k2·k3, k2·k4, k3·k4)

[0019] Where: s inter is a cross-effect term.

[0020] As a further solution of the present invention, in step 2, the time prediction function of the preset physiological parameters and the airbag adjustment time is obtained by performing vector operations on the coefficient matrix and the single influence term and the cross-influence term. The time prediction function formula of the preset physiological parameters and the airbag adjustment time is:

[0021] f(k1,k2,k3,k4,θ)=θ single ·S single +θ inter ·S inter

[0022] Where: f(k1, k2, k3, k4, θ) is the time prediction function of the preset physiological parameters and the airbag adjustment time, θ is the coefficient term, θ single is the coefficient matrix of a single influencing term, θ single =(θ1,θ2,θ3,θ4),θ inter is the coefficient matrix of the cross-influence term, θ inter=(θ5,θ6,θ7,θ8),θ single ,θ inter Obtained through numerical analysis of historical data, θ includes θ1, θ2, θ3, θ4, θ5, θ6, θ7, and θ8.

[0023] As a further solution of the present invention, in step 2, the likelihood function of the predicted value of the airbag pressure adjustment is:

[0024]

[0025] Where: P(Δt|k1, k2, k3, k4, θ) is the likelihood function of the predicted value of the airbag pressure adjustment, Δt is the airbag pressure adjustment time, and σ is the standard deviation of the likelihood function.

[0026] As a further solution of the present invention, in step 3, the Bayesian formula is:

[0027] P(θ|Δt,k1,k2,k3,k4)∝P(Δt|k1,k2,k3,k4,θ)·P(θ)

[0028] Where: P(θ|Δt,k1,k2,k3,k4) is the posterior distribution of updated model parameters, and P(θ) is the prior distribution of model parameters.

[0029] As a further solution of the present invention, in step 4, the specific implementation process of using the Markov chain Monte Carlo method to sample from the posterior distribution to obtain a set of new model parameters includes:

[0030] Step 11, parameter initialization: select the initial value of the coefficient term, which is set based on the prior distribution and experience, set the total number of iterations to the number of samples generated, select the step size parameter, and control the range of candidate sample generation;

[0031] Step 12, iterative sampling: Generate new candidate parameters from the symmetric distribution centered on the current parameter, calculate the ratio of the posterior distribution at the candidate sample and the current sample as the acceptance probability, calculate the posterior distribution according to the Bayesian formula, generate a random number, the value range of the random number is 0 to 1, if the random number is less than or equal to the acceptance probability, then accept the candidate sample, otherwise reject the candidate sample and keep the current sample;

[0032] Step 13, repeat iteration: repeat step 12 until the predetermined number of iterations is reached, and record the generated samples in each iteration;

[0033] Step 14, post-processing: determine the length of the burn-in period based on experience, remove the samples with the previously set burn-in period length from the total samples, obtain a new sample sequence, retain one sample for each set interval from the retained sample sequence, obtain the sample sequence after interval sampling and summarize it to form the final sample set, and calculate the mean and variance of the sample set.

[0034] As a further solution of the present invention, the specific implementation process of step 5 includes:

[0035] Step 21, calculating the expected value from the sampling results: for the parameter set of the coefficient term sampled from the posterior distribution, the airbag adjustment time corresponding to each parameter set is calculated by using the preset physiological parameter and the time prediction function of the airbag adjustment time;

[0036] Step 22, calculating the expected value: using the mean of the airbag adjustment time corresponding to all parameter sets as the mean of the airbag adjustment time;

[0037] Step 23, calculate the credible interval: calculate the 95% credible interval of the airbag adjustment interval, where the two ends of the interval are the sample values ​​of the 2.5th percentile and the 97.5th percentile respectively. If the parameter numbers of the 2.5th percentile and the 97.5th percentile of the credible interval are not integers, then the value at the nearest integer position is taken;

[0038] Step 24, obtaining the airbag pressure adjustment value according to the current physiological parameters: according to the preset objective function formula Where J is the objective function value. Under the constraint of the airbag pressure change, the variational method is used to obtain the optimal solution P(t) of the objective function. opt , k 1,opt , k 2,opt , k 3,opt , k 4,opt They are the ideal values ​​of airbag adjustment pressure, heart rate, blood oxygen saturation, respiratory rate, and blood pressure, which are set according to expert experience. The airbag pressure adjustment process is further refined through dynamic optimization. The adjustment amount is set as the difference between the airbag pressure adjustment value at the current time and the previous airbag pressure adjustment time. Combined with the dynamic optimization method, the influence of physiological state parameters on the adjustment amount is considered to obtain the airbag pressure adjustment value.

[0039] As a further solution of the present invention, the specific implementation process of step 24 is as follows:

[0040] Step 31, problem definition: The goal is to obtain the adjustment value of the airbag pressure according to the patient's physiological state parameters and the airbag pressure adjustment time, so as to maximize the patient's comfort and safety during nursing, while avoiding excessive or low airbag pressure;

[0041] Step 32, objective function definition: define an objective function J to represent the patient's comfort and safety, which is a function of the balloon pressure P(t) and physiological state parameters;

[0042] Step 33, constraint condition: the change of the airbag satisfies the condition that it is greater than or equal to the minimum value obtained from the patient's historical data to keep the patient comfortable and safe, and is less than or equal to the maximum value obtained from the patient's historical data to keep the patient comfortable and safe;

[0043] Step 34, Variational solution: Define the Lagrangian function λ is the Lagrange multiplier, which is used to construct the Lagrangian function. The Lagrangian function is solved by the variational method to obtain the equation For P(t), we have α1(P(t)-P opt )+λ=0, then α1 is the influence weight of the airbag pressure adjustment value on the objective function;

[0044] Step 35, dynamic optimization: The adjustment process is further refined through dynamic optimization, and the adjustment amount ΔP = P(t)-P(t-Δt) is defined. In combination with the dynamic optimization method, the influence of physiological state parameters on the adjustment amount is considered, ΔP = β1Δt+β2(k1(t)-k 1,opt )+β3(k2(t)-k 2,opt )+β4(k3(t)-k 3,opt )+β5(k4(t)-k 4,opt ), β1, β2, β3, β4, and β5 are optimization coefficients, which are determined through historical data and experimental analysis.

[0045] Compared with the prior art, the technical effect of the present invention is as follows: the present invention realizes personalized, real-time monitoring and automatic adjustment of endotracheal intubation cuff pressure by introducing edge computing and intelligent algorithms, solves the problems of insufficient real-time performance and delayed response of manual operation in traditional nursing, and uses the data acquisition module to obtain the patient's physiological state data in real time. After cleaning, denoising and normalization processing by the edge computing device, the relationship function between the preset physiological parameters and the cuff pressure adjustment time and the Bayesian update algorithm are used to accurately predict the time and range of the cuff pressure adjustment. Compared with the traditional method, the present invention effectively avoids tracheal damage and air leakage caused by excessive or low cuff pressure during nursing, and reduces the risk of VAP. At the same time, data-driven adjustment based on the patient's individual physiological state realizes dynamic optimization of cuff pressure. Through the distributed processing of edge computing devices, the problems of data transmission delay and information islands between devices are solved, and the real-time and accuracy of cuff pressure management are significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a system block diagram of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solution in the present invention. Obviously, the described technical solution is only a part of the present invention but not all of it. Based on the technical solution in the present invention, all other technical solutions obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0048] like Figure 1 As shown, the present invention proposes an edge computing-based tracheal intubation airbag pressure intelligent nursing system, including a data acquisition module, an edge computing device, a communication module and an airbag pressure automatic regulating valve, the data acquisition module is connected to the edge computing device through the communication module, the edge computing device is connected to the airbag pressure automatic regulating valve through the communication module, the edge computing device cleans, denoises and normalizes the acquired patient physiological state data, based on the processed patient physiological state data, through the predefined relationship function between physiological parameters and airbag pressure adjustment time, obtains the likelihood function, and then updates the posterior distribution of the model parameters according to the Bayesian formula, samples from the posterior distribution, obtains new model parameters, and obtains the expected value and credible interval of the predicted airbag pressure time adjustment range by using the sampled model parameters, and then obtains the airbag pressure adjustment prediction value based on the predicted time range and current physiological parameters and the preset objective function, and the airbag pressure automatic regulating valve adjusts the tracheal intubation airbag pressure connected to the ventilator in real time according to the airbag pressure adjustment time prediction value and the airbag pressure adjustment amount prediction value, and completes the airbag pressure adjustment within the predicted airbag pressure adjustment time prediction value.

[0049] The present invention realizes personalized, real-time monitoring and automatic adjustment of endotracheal intubation cuff pressure by introducing edge computing and intelligent algorithms, solves the problems of insufficient real-time performance and delayed response of manual operation in traditional nursing, and uses a data acquisition module to obtain the patient's physiological status data in real time. After cleaning, denoising and normalization processing by edge computing equipment, the time and range of cuff pressure adjustment are accurately predicted through the relationship function between preset physiological parameters and cuff pressure adjustment time and the Bayesian update algorithm. Compared with traditional methods, tracheal damage and air leakage caused by excessive or low cuff pressure during nursing are effectively avoided, and the risk of VAP is reduced. At the same time, data-driven adjustment based on the patient's individual physiological state realizes dynamic optimization of cuff pressure. Through distributed processing of edge computing equipment, the problems of data transmission delay and information islands between devices are solved, and the real-time and accuracy of cuff pressure management are significantly improved.

[0050] It should be noted that the data acquisition module includes a physiological parameter monitoring device and an airbag pressure sensor. The physiological parameter monitoring data obtains the patient's heart rate, blood oxygen saturation, respiratory rate and blood pressure in real time. The airbag pressure sensor monitors the airbag pressure of the endotracheal tube connected to the ventilator in real time. The edge computing device includes a data preprocessing module, an airbag pressure adjustment parameter prediction module and a real-time decision module. The data preprocessing module is connected to the airbag pressure adjustment parameter prediction module, and the airbag pressure adjustment parameter prediction module is connected to the real-time decision module.

[0051] It should be noted that the specific process of the airbag pressure adjustment parameter prediction module obtaining the airbag pressure adjustment time prediction value and the airbag pressure adjustment amount prediction value includes:

[0052] Step 1, data input: input the pre-processed heart rate, blood oxygen saturation, respiratory rate, blood pressure and the airbag pressure at the previous moment;

[0053] Step 2, calculating the likelihood function: obtaining the likelihood function of the pressure adjustment prediction value according to the preset physiological parameters and the time prediction function of the airbag adjustment time;

[0054] Step 3, verify the distribution after updating: According to the Bayesian formula, verify the probability distribution of the likelihood function of the pressure adjustment prediction value after updating the model parameters;

[0055] Step 4: Markov Chain Monte Carlo method sampling: Use the Markov Chain Monte Carlo method to sample from the posterior distribution to obtain a new set of model parameters;

[0056] Step 5, predict the airbag adjustment time range and airbag pressure adjustment value: use the sampled model parameters to obtain the expected value and credible interval of the predicted adjustment time range, and then predict the airbag pressure adjustment prediction value through the preset objective function based on the predicted adjustment time range and current physiological parameters.

[0057] By introducing the balloon pressure adjustment parameter prediction module, the system realizes the accurate prediction and dynamic adjustment of the balloon pressure of endotracheal intubation. First, the pre-processed heart rate, blood oxygen saturation, respiratory rate, blood pressure and balloon pressure data of the previous moment are input into the system, which reflects the patient's physiological state in real time. Based on these data, the likelihood function of the pressure adjustment prediction value is calculated by the preset physiological parameters and the time prediction function of the balloon adjustment time. Then, the posterior distribution of the model parameters is updated by combining the current physiological state data with the Bayesian formula, and a set of new model parameters are obtained by sampling from the posterior distribution through the Markov chain Monte Carlo method. These parameters reflect the possible value range of balloon pressure adjustment and take into account various uncertainties and change factors. Finally, using these model parameters, the system calculates the expected value and credible interval of the predicted adjustment time range, and predicts the balloon pressure adjustment value based on the current physiological parameters and the preset objective function. Through edge computing and intelligent algorithms, the system realizes real-time monitoring and automatic adjustment of balloon pressure, solving the problems of insufficient real-time performance and delayed response of manual operation. Dynamic adjustments based on the patient's real-time physiological status data avoid the discomfort and risks caused by universal standards in traditional methods, and improve the patient's comfort and safety. Accurately control the cuff pressure to avoid tracheal damage and air leakage caused by excessive or low pressure, reducing the risk of ventilator-associated pneumonia (VAP). Through Bayesian updating and Markov chain Monte Carlo methods, historical data and real-time data are fully utilized to improve the accuracy and reliability of model predictions. Distributed processing of edge computing devices solves the problem of poor information flow between devices, improves the real-time and coordination of cuff pressure management, significantly improves the safety and effectiveness of tracheal intubation, and improves the nursing effect of patient tracheal pressure regulation.

[0058] It should be noted that in step 2, the time prediction function of the preset physiological parameters and the airbag adjustment time is composed of a single influence term and a cross-image term. The single influence term considers the independent influence of each physiological parameter on the adjustment time. The expression of the single influence term is:

[0059]

[0060] Where: S single is a single influencing item, k1, k2, k3, and k4 are the heart rate, blood oxygen saturation, respiratory rate, and blood pressure after pretreatment, respectively;

[0061] The cross-influence term is a vector composed of the interaction term between heart rate and blood oxygen saturation, the interaction term between heart rate and respiratory rate, the interaction term between heart rate and blood pressure, the interaction term between blood oxygen saturation and respiratory rate, the interaction term between blood oxygen saturation and blood pressure, and the interaction term between respiratory rate and blood pressure. The expression of the cross-influence term is:

[0062] S inter=(k1·k2, k1·k3, k1·k4, k2·k3, k2·k4, k3·k4)

[0063] Where: S inter is a cross-effect term.

[0064] In step 2, by introducing the time prediction function of the preset physiological parameters and the airbag adjustment time, which includes single influence terms and cross-influence terms, the system can fully consider the independent influence of each physiological parameter on the adjustment time and the complex relationship between them. reflects the independent effect of each physiological parameter, while the cross-effect term S inter =(k1·k2, k1·k3, k1·k4, k2·k3, k2·k4, k3·k4) captures the interaction between physiological parameters. By analyzing the comprehensive influence of physiological parameters in a multi-dimensional and multi-level manner, the accuracy and reliability of the prediction of the airbag pressure adjustment time are improved, thereby ensuring the precise adjustment of the airbag pressure, minimizing the risk of ventilator-associated pneumonia (VAP), and enhancing the safety and comfort of patients.

[0065] Single Impact Item Reflecting the independent effect of each physiological parameter, heart rate k1 is processed by hyperbolic tangent function, blood oxygen saturation k2 is processed by square, respiratory rate k3 is processed by exponential, and blood pressure k4 is processed by logarithm. These function forms can effectively handle the nonlinear characteristics and different magnitudes of data, and improve the adaptability and generalization ability of the model. Cross-influence term S inter =(k1·k2, k1·k3, k1·k4, k2·k3, k2·k4, k3·k4) reflects the interaction between physiological parameters and captures the comprehensive impact of multiple physiological parameter combinations on the airbag pressure adjustment time. This cross-influence term can reveal complex relationships that cannot be captured by a single parameter, making the model more flexible and accurate.

[0066] It should be noted that in step 2, the time prediction function of the preset physiological parameters and the airbag adjustment time is obtained by performing vector operations on the coefficient matrix and the single influence term and the cross-influence term. The time prediction function formula of the preset physiological parameters and the airbag adjustment time is:

[0067] f(k1,k2,k3,k4,θ)=θ single ·S single +θ inter ·S inter

[0068] Where: f(k1, k2, k3, k4, θ) is the time prediction function of the preset physiological parameters and the airbag adjustment time, θ is the coefficient term, θ singleis the coefficient matrix of a single influencing term, θ single =(θ1,θ2,θ3,θ4),θ inter is the coefficient matrix of the cross-influence term, θ inter =(θ5,θ6,θ7,θ8),θ single ,θ inter Obtained through numerical analysis of historical data, θ includes θ1, θ2, θ3, θ4, θ5, θ6, θ7, and θ8.

[0069] In step 2, vector operations are performed on the coefficient matrix and single and cross-influence terms to construct a time prediction function of physiological parameters and airbag adjustment time, which brings significant technical advantages. The coefficient matrix can flexibly adjust the influence weights of each physiological parameter to accurately predict the airbag adjustment time. At the same time, the independent influence and interaction of each physiological parameter are comprehensively considered to make the prediction results more comprehensive and accurate. Based on the physiological parameters obtained in real time, the model can dynamically adjust the airbag pressure to achieve personalized adjustment and improve the comfort and safety of patients. Using the coefficient matrix obtained by historical data analysis, the model has high robustness and adaptability, and is suitable for different patients and clinical conditions. In addition, through the distributed processing of edge computing devices, the system can quickly respond to changes in physiological parameters, realize real-time monitoring and adjustment of airbag pressure, avoid the delay problem of manual operation, effectively reduce the risk of ventilator-associated pneumonia (VAP), and ensure patient safety.

[0070] It should be noted that in step 2, the likelihood function of the predicted value of the airbag pressure adjustment is:

[0071]

[0072] Where: P(Δt|k1, k2, k3, k4, θ) is the likelihood function of the predicted value of the airbag pressure adjustment, Δt is the airbag pressure adjustment time, and σ is the standard deviation of the likelihood function.

[0073] In step 2, the likelihood function of the balloon pressure-adjusted prediction is defined as The probabilistic modeling of the prediction of the airbag adjustment time is realized, which can quantify the reliability and uncertainty of the model prediction, thereby improving the accuracy and robustness of the prediction. Through the likelihood function, the model can adjust the parameters according to the actual observed data, so that the predicted value is closer to the actual value and the error is reduced. In addition, the introduction of the standard deviation σ enables the model to handle noise and outliers in the data, improving the stability of the prediction results. By optimizing the likelihood function, the model can adaptively adjust the airbag pressure to make it more in line with the individual needs of patients and reduce the risks caused by too high or too low pressure. Ultimately, this method improves the accuracy and reliability of airbag pressure management, significantly reduces the risk of ventilator-associated pneumonia (VAP), and enhances the safety and comfort of patients.

[0074] It should be noted that in step 3, the Bayesian formula is:

[0075] P(θ|Δt,k1,k2,k3,k4)∝P(Δt|k1,k2,k3,k4,θ)·P(θ)

[0076] Where: P(θ|Δt, k1, k2, k3, k4) is the posterior distribution of the updated model parameters, and P(θ) is the prior distribution of the model parameters.

[0077] Through the Bayesian update method, the model parameters can be dynamically adjusted according to the newly acquired data. Each time new data is input, the model will be adaptively adjusted to make the parameters more consistent with the current patient's physiological state, thereby improving the accuracy of the prediction. The Bayesian method naturally handles uncertainty. By combining prior information and observed data to generate a posterior distribution, it can reflect the uncertainty in parameter estimation and provide more reliable prediction results. The prior distribution P(θ) can introduce previous knowledge or experience, and even in the case of insufficient data or large noise, it can ensure the robustness of the model and avoid overfitting problems. The posterior distribution P(θ|Δt, k1, k2, k3, k4) represents the optimal estimate of the parameters under the observed data. By sampling the posterior distribution, a set of optimal parameters can be obtained, which improves the accuracy of the prediction of the airbag pressure adjustment. The Bayesian update method enables the model to be adjusted according to the physiological data of individual patients, providing a personalized airbag pressure adjustment solution, which can significantly improve the comfort and safety of patients. Through accurate parameter updates and pressure adjustments, it is possible to avoid excessive or low airbag pressure and reduce the risk of ventilator-associated pneumonia (VAP) caused by improper airbag pressure.

[0078] It should be noted that in step 4, the specific implementation process of using the Markov Chain Monte Carlo method to sample from the posterior distribution to obtain a set of new model parameters includes:

[0079] Step 11, parameter initialization: select the initial value of the coefficient term, which is set based on the prior distribution and experience, set the total number of iterations to the number of samples generated, select the step size parameter, and control the range of candidate sample generation;

[0080] Step 12, iterative sampling: Generate new candidate parameters from the symmetric distribution centered on the current parameter, calculate the ratio of the posterior distribution at the candidate sample and the current sample as the acceptance probability, calculate the posterior distribution according to the Bayesian formula, generate a random number, the value range of the random number is 0 to 1, if the random number is less than or equal to the acceptance probability, then accept the candidate sample, otherwise reject the candidate sample and keep the current sample;

[0081] Step 13, repeat iteration: repeat step 12 until the predetermined number of iterations is reached, and record the generated samples in each iteration;

[0082] Step 14, post-processing: determine the length of the burn-in period based on experience, remove the samples with the previously set burn-in period length from the total samples, obtain a new sample sequence, retain one sample for each set interval from the retained sample sequence, obtain the sample sequence after interval sampling and summarize it to form the final sample set, and calculate the mean and variance of the sample set.

[0083] By sampling from the posterior distribution, the Markov chain Monte Carlo method can fully explore the parameter space, avoid falling into the local optimal solution, and improve the accuracy and robustness of parameter estimation. The Markov chain Monte Carlo method can effectively handle complex multi-peak distributions and asymmetric distributions. By iterative sampling and calculation of acceptance probability, it gradually approaches the true posterior distribution, making parameter estimation more reliable. By recording a large number of samples and calculating the mean and variance of the samples, a set of model parameters reflecting the characteristics of the posterior distribution are obtained. These parameters can improve the accuracy of the prediction of the airbag pressure adjustment time and adjustment amount, and ensure the accuracy of the adjustment. In the post-processing step, by removing the burn-in period samples and interval sampling, the independence and representativeness of the final sample set are ensured, the autocorrelation in the sampling is reduced, and the model's ability to handle uncertainty and data noise is enhanced. This method enables the model to be dynamically adjusted according to real-time data, providing personalized airbag pressure adjustment solutions, improving patient comfort and safety, and avoiding the risks caused by excessive or low airbag pressure through precise and reliable airbag pressure adjustment, reducing the incidence of ventilator-associated pneumonia (VAP).

[0084] It should be noted that the specific implementation process of step 5 includes:

[0085] Step 21, calculating the expected value from the sampling results: for the parameter set of the coefficient term sampled from the posterior distribution, the airbag adjustment time corresponding to each parameter set is calculated by using the preset physiological parameter and the time prediction function of the airbag adjustment time;

[0086] Step 22, calculating the expected value: using the mean of the airbag adjustment time corresponding to all parameter sets as the mean of the airbag adjustment time;

[0087] Step 23, calculate the credible interval: calculate the 95% credible interval of the airbag adjustment interval, where the two ends of the interval are the sample values ​​of the 2.5th percentile and the 97.5th percentile respectively. If the parameter numbers of the 2.5th percentile and the 97.5th percentile of the credible interval are not integers, then the value at the nearest integer position is taken;

[0088] Step 24, obtaining the airbag pressure adjustment value according to the current physiological parameters: according to the preset objective function formula Where J is the objective function value. Under the constraint of the airbag pressure change, the variational method is used to obtain the optimal solution P(t) of the objective function. opt , k 1,opt , k 2,opt , k 3,opt , k 4,opt They are the ideal values ​​of airbag adjustment pressure, heart rate, blood oxygen saturation, respiratory rate, and blood pressure, which are set according to expert experience. The airbag pressure adjustment process is further refined through dynamic optimization. The adjustment amount is set as the difference between the airbag pressure adjustment value at the current time and the previous airbag pressure adjustment time. Combined with the dynamic optimization method, the influence of physiological state parameters on the adjustment amount is considered to obtain the airbag pressure adjustment value.

[0089] The parameter set sampled from the posterior distribution is used to calculate the airbag adjustment time corresponding to each parameter set through the preset time prediction function, and the mean and credible interval are calculated using these data to ensure the accuracy and reliability of the prediction results. The uncertainty of the prediction is quantified by calculating the 95% credible interval, which provides a clear range and confidence level for the adjustment process and enhances the reliability of the prediction results. The airbag pressure adjustment value is dynamically optimized according to the current physiological parameters and the preset objective function to make it more in line with individual needs and improve the comfort and safety of patients. The airbag pressure is adjusted in real time using the variational method and dynamic optimization method, and the adjustment amount is determined according to the pressure difference between the current and previous moments. The change of physiological state is quickly responded to to avoid delayed and inaccurate adjustments. By accurately controlling the airbag pressure, tracheal damage and air leakage caused by excessive or low pressure during nursing can be avoided, and the risk of ventilator-associated pneumonia (VAP) can be reduced. The airbag pressure adjustment process is refined in combination with the influence of physiological state parameters on the adjustment amount to ensure the refinement and effectiveness of the regulation process.

[0090] It should be noted that the specific implementation process of step 24 is as follows:

[0091] Step 31, problem definition: The goal is to obtain the adjustment value of the airbag pressure according to the patient's physiological state parameters and the airbag pressure adjustment time, so as to maximize the patient's comfort and safety during nursing, while avoiding excessive or low airbag pressure;

[0092] Step 32, objective function definition: define an objective function J to represent the patient's comfort and safety, which is a function of the balloon pressure P(t) and physiological state parameters;

[0093] Step 33, constraint condition: the change of the airbag satisfies the condition that it is greater than or equal to the minimum value obtained from the patient's historical data to keep the patient comfortable and safe, and is less than or equal to the maximum value obtained from the patient's historical data to keep the patient comfortable and safe;

[0094] Step 34, Variational solution: Define the Lagrangian function λ is the Lagrange multiplier, which is used to construct the Lagrangian function. The Lagrangian function is solved by the variational method to obtain the equation For P(t), we have α1(P(t)-P opt )+λ=0, then α1 is the influence weight of the airbag pressure adjustment value on the objective function;

[0095] Step 35, dynamic optimization: The adjustment process is further refined through dynamic optimization, and the adjustment amount ΔP = P(t)-P(t-Δt) is defined. In combination with the dynamic optimization method, the influence of physiological state parameters on the adjustment amount is considered, ΔP = β1Δt+β2(k1(t)-k 1,opt )+β3(k2(t)-k 2,opt )+β4(k3(t)--k 3,opt )+β5(k4(t)-k 4,opt ), β1, β2, β3, β4, and β5 are optimization coefficients, which are determined through historical data and experimental analysis.

[0096] By defining the objective function J and constraints, the optimization goal is clarified, that is, to accurately adjust the airbag pressure while ensuring the comfort and safety of the patient, to avoid the discomfort and risks caused by too high or too low pressure, and to use the variational method to calculate the Lagrangian function Solve and obtain the analytical expression of the airbag pressure P(t) Among them, α1 represents the influence weight of the airbag pressure adjustment value on the objective function, which ensures the accuracy and controllability of the airbag pressure adjustment process. Through the dynamic optimization method, the airbag pressure adjustment process is refined, and the adjustment amount ΔP = P(t)-P(t-Δt) is defined. Combined with the influence of physiological state parameters, the formula is ΔP = β1Δt+β2(k1(t)-k 1,opt )+β3(k2(t)-k 2,opt )+β4(k3(t)-k 3,opt )+β5(k4(t)-k 4,opt ) is the optimization coefficient. This dynamic optimization takes into account the changes in real-time physiological status, improves the accuracy and response speed of the airbag pressure adjustment, and combines the patient's historical data and current physiological status. The system can provide personalized airbag pressure adjustment care plans, improves the adaptability of the adjustment and the patient's comfort. Through precise control and dynamic optimization, it avoids tracheal damage and air leakage caused by excessive or low airbag pressure, and significantly reduces the risk of ventilator-associated pneumonia (VAP).

[0097] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0098] Finally: The above description is only an implementation scheme of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An edge computing-based endotracheal intubation airbag pressure intelligent care system, characterized in that: The invention comprises a data acquisition module, an edge computing device, a communication module and an automatic airbag pressure regulating valve. The data acquisition module is connected to the edge computing device through the communication module. The edge computing device is connected to the automatic airbag pressure regulating valve through the communication module. The edge computing device cleans, denoises and normalizes the acquired patient physiological state data. Based on the processed patient physiological state data, the likelihood function is obtained through the relationship function between the predefined physiological parameters and the airbag pressure adjustment time. Then, the posterior distribution of the model parameters is updated according to the Bayesian formula. The new model parameters are sampled from the posterior distribution. The expected value and the credible interval of the predicted airbag pressure time adjustment range are obtained by using the sampled model parameters. Then, according to the predicted time range and the current physiological parameters, based on the preset objective function, the airbag pressure adjustment prediction value is obtained. The airbag pressure automatic regulating valve adjusts the endotracheal intubation airbag pressure connected to the ventilator in real time according to the airbag pressure adjustment time prediction value and the airbag pressure adjustment amount prediction value, and completes the airbag pressure adjustment within the predicted airbag pressure adjustment time prediction value.

2. According to the edge computing-based endotracheal intubation airbag pressure intelligent care system of claim 1, it is characterized in that: The data acquisition module includes a physiological parameter monitoring device and an airbag pressure sensor. The physiological parameter monitoring data obtains the patient's heart rate, blood oxygen saturation, respiratory rate and blood pressure in real time. The airbag pressure sensor monitors the airbag pressure of the endotracheal tube connected to the ventilator in real time.

3. According to the edge computing-based endotracheal intubation airbag pressure intelligent nursing system of claim 2, it is characterized in that: The edge computing device includes a data preprocessing module, an airbag pressure adjustment parameter prediction module and a real-time decision module. The data preprocessing module is connected to the airbag pressure adjustment parameter prediction module, and the airbag pressure adjustment parameter prediction module is connected to the real-time decision module.

4. According to the edge computing-based endotracheal intubation airbag pressure intelligent nursing system of claim 3, it is characterized in that: The specific process of the airbag pressure adjustment parameter prediction module obtaining the airbag pressure adjustment time prediction value and the airbag pressure adjustment amount prediction value includes: Step 1, data input: input the pre-processed heart rate, blood oxygen saturation, respiratory rate, blood pressure and the airbag pressure at the previous moment; Step 2, calculating the likelihood function: obtaining the likelihood function of the pressure adjustment prediction value according to the preset physiological parameters and the time prediction function of the airbag adjustment time; Step 3, verify the distribution after updating: According to the Bayesian formula, verify the probability distribution of the likelihood function of the pressure adjustment prediction value after updating the model parameters; Step 4: Markov Chain Monte Carlo method sampling: Use the Markov Chain Monte Carlo method to sample from the posterior distribution to obtain a new set of model parameters; Step 5, predict the airbag adjustment time range and airbag pressure adjustment value: use the sampled model parameters to obtain the expected value and credible interval of the predicted adjustment time range, and then predict the airbag pressure adjustment prediction value through the preset objective function based on the predicted adjustment time range and current physiological parameters.

5. According to the edge computing-based endotracheal intubation airbag pressure intelligent care system of claim 4, it is characterized in that: In step 2, the time prediction function of the preset physiological parameters and the airbag adjustment time is composed of a single influence term and a cross-image term. The single influence term considers the independent influence of each physiological parameter on the adjustment time. The expression of the single influence term is: Where: S single is a single influencing item, k1, k2, k3, and k4 are the heart rate, blood oxygen saturation, respiratory rate, and blood pressure after pretreatment, respectively; The cross-influence term is a vector composed of the interaction term between heart rate and blood oxygen saturation, the interaction term between heart rate and respiratory rate, the interaction term between heart rate and blood pressure, the interaction term between blood oxygen saturation and respiratory rate, the interaction term between blood oxygen saturation and blood pressure, and the interaction term between respiratory rate and blood pressure. The expression of the cross-influence term is: S inter =(k1·k2,k1·k3,k1·k4,k2·k3,k2·k4,k3·k4) Where: S inter is a cross-effect term.

6. According to the edge computing-based endotracheal intubation airbag pressure intelligent nursing system of claim 5, it is characterized in that: In step 2, the time prediction function of the preset physiological parameters and the airbag adjustment time is obtained by performing vector operations on the coefficient matrix and the single influence term and the cross-influence term. The time prediction function formula of the preset physiological parameters and the airbag adjustment time is: f(k1,k2,k3,k4,θ)=θ single ·S single +θ inter ·S inter Where: f(k1, k2, k3, k4, θ) is the time prediction function of the preset physiological parameters and the airbag adjustment time, θ is the coefficient term, θ single is the coefficient matrix of a single influencing term, θ single =(θ1,θ2,θ3,θ4),θ inter is the coefficient matrix of the cross-influence term, θ inter =(θ5,θ6,θ7,θ8),θ single ,θ inter Obtained through numerical analysis of historical data, θ includes θ1, θ2, θ3, θ4, θ5, θ6, θ7, and θ8.

7. According to claim 6, an edge computing-based endotracheal intubation airbag pressure intelligent care system is characterized in that: In step 2, the likelihood function of the predicted value of the balloon pressure adjustment is: Where: P(Δt|k1, k2, k3, k4, θ) is the likelihood function of the predicted value of the airbag pressure adjustment, Δt is the airbag pressure adjustment time, and σ is the standard deviation of the likelihood function.

8. The endotracheal intubation airbag pressure intelligent nursing system based on edge computing according to claim 7 is characterized in that: In step 3, the Bayesian formula is: P(θ|Δt,k1,k2,k3,k4)∝P(Δt|k1,k2,k3,k4,θ)·P(θ) Where: P(θ|Δt,k1,k2,k3,k4) is the posterior distribution of updated model parameters, and P(θ) is the prior distribution of model parameters.

9. The endotracheal intubation airbag pressure intelligent nursing system based on edge computing according to claim 8 is characterized in that: In step 4, the specific implementation process of using the Markov Chain Monte Carlo method to sample from the posterior distribution to obtain a new set of model parameters includes: Step 11, parameter initialization: select the initial value of the coefficient term, which is set based on the prior distribution and experience, set the total number of iterations to the number of samples generated, select the step size parameter, and control the range of candidate sample generation; Step 12, iterative sampling: Generate new candidate parameters from the symmetric distribution centered on the current parameter, calculate the ratio of the posterior distribution at the candidate sample and the current sample as the acceptance probability, calculate the posterior distribution according to the Bayesian formula, generate a random number, the value range of the random number is 0 to 1, if the random number is less than or equal to the acceptance probability, then accept the candidate sample, otherwise reject the candidate sample and keep the current sample; Step 13, repeat iteration: repeat step 12 until the predetermined number of iterations is reached, and record the generated samples in each iteration; Step 14, post-processing: determine the length of the burn-in period based on experience, remove the samples with the previously set burn-in period length from the total samples, obtain a new sample sequence, retain one sample for each set interval from the retained sample sequence, obtain the sample sequence after interval sampling and summarize it to form the final sample set, and calculate the mean and variance of the sample set.

10. The endotracheal intubation airbag pressure intelligent nursing system based on edge computing according to claim 8, characterized in that: The specific implementation process of step 5 includes: Step 21, calculating the expected value from the sampling results: for the parameter set of the coefficient term sampled from the posterior distribution, the airbag adjustment time corresponding to each parameter set is calculated by using the preset physiological parameter and the time prediction function of the airbag adjustment time; Step 22, calculating the expected value: using the mean of the airbag adjustment time corresponding to all parameter sets as the mean of the airbag adjustment time; Step 23, calculate the credible interval: calculate the 95% credible interval of the airbag adjustment interval, where the two ends of the interval are the sample values ​​of the 2.5th percentile and the 97.5th percentile respectively. If the parameter numbers of the 2.5th percentile and the 97.5th percentile of the credible interval are not integers, then the value at the nearest integer position is taken; Step 24, obtaining the airbag pressure adjustment value according to the current physiological parameters: according to the preset objective function formula Where J is the objective function value. Under the constraint of the airbag pressure change, the variational method is used to obtain the optimal solution P(t) of the objective function. opt , k 1,opt , k 2,opt , k 3,opt , k 4,opt They are the ideal values ​​of airbag adjustment pressure, heart rate, blood oxygen saturation, respiratory rate, and blood pressure, which are set according to expert experience. The airbag pressure adjustment process is further refined through dynamic optimization. The adjustment amount is set as the difference between the airbag pressure adjustment value at the current time and the previous airbag pressure adjustment time. Combined with the dynamic optimization method, the influence of physiological state parameters on the adjustment amount is considered to obtain the airbag pressure adjustment value.

Citation Information

Cited By

  • Airbag pressure monitoring system in artificial airway based on machine learning

    CN120827669A

  • Intelligent regulation and control method and system for pressure of trachea cannula cuff

    CN121177628A