Anesthesia mask tube internal pressure electric regulation system

Abstract

The application relates to the technical field of medical instrument, and discloses an electric pressure regulation system for an anesthesia mask pipe. The system comprises a physiological data acquisition module, which acquires multi-source physiological data to generate a synchronous data set; a multi-modal feature modeling module, which generates a multi-dimensional joint probability feature matrix by using a Bayesian probability graph model; a dynamic pressure decision module, which constructs an adaptive controller based on deep reinforcement learning to output a pressure regulation target value; and a multi-target collaborative control module, which generates a steady-state pressure control strategy by using a mixed integer programming model and an improved ant colony algorithm. In addition, the system is also provided with a virtual pressure estimation module for pressure leakage diagnosis and redundant valve switching, and an abnormal interruption protection module for guaranteeing the safety of a patient. The system realizes accurate and intelligent regulation of the pressure in the anesthesia mask pipe, and improves the safety and stability of the anesthesia process.

Description

Technical Field

[0001] This invention relates to the field of medical device technology, specifically to an electric pressure regulating system for the tube of an anesthesia mask. Background Technology

[0002] In the field of modern medical anesthesia, the anesthesia mask is a crucial device for maintaining a patient's airway patency, delivering anesthetic gases, and ensuring respiratory support. Precise control of the pressure within the anesthesia mask tubing is essential for ensuring patient safety and improving anesthetic efficacy. However, existing anesthesia mask pressure regulation technologies have many shortcomings and are insufficient to meet clinical needs.

[0003] Traditional anesthesia mask pressure regulation relies heavily on manual operation by medical staff. This method is not only inefficient but also highly susceptible to human error. Medical staff need to constantly monitor the patient's physiological state and pressure data, manually adjusting the valve opening to control the pressure. In emergencies, the speed and accuracy of manual adjustment are difficult to guarantee, easily leading to excessive pressure fluctuations that can affect the patient's respiratory function and the effectiveness of anesthesia. For example, during surgery, the patient's physiological state may suddenly change, requiring rapid adjustment of the anesthesia mask pressure; however, manual adjustment may be delayed, failing to meet the patient's needs in a timely manner.

[0004] While some anesthesia masks employing simple electronic control systems achieve a degree of automation, their functionality remains limited. These systems typically adjust pressure based on preset, fixed parameters, failing to consider individual patient differences and dynamic changes in physiological states in real time. Different patients vary in age, weight, condition, and breathing patterns, resulting in different pressure requirements for the anesthesia mask. Existing simple control systems cannot flexibly adjust to these differences, potentially leading to excessively high or low pressure, causing discomfort or even harm to the patient. For example, pediatric patients, whose respiratory systems are more vulnerable, require more precise pressure regulation, a need that existing simple control systems cannot meet.

[0005] Furthermore, during anesthesia, a patient's respiratory rate, blood oxygen saturation, and other physiological indicators constantly change, and these changes are closely related to the pressure inside the anesthesia mask tube. However, most current regulation systems lack effective integration and analysis of multi-source physiological data, making it impossible to establish a dynamic correlation model between pressure and physiological indicators. This results in a lack of scientific basis for pressure regulation, making precise control difficult. For example, when a patient's respiratory rate increases, existing regulation systems may not be able to adjust the pressure in time to meet the patient's respiratory needs, thus affecting gas exchange and the anesthetic effect.

[0006] Current pressure monitoring technologies also have limitations. Pressure monitoring at critical locations such as the endotracheal tube connection is not accurate enough, failing to detect potential pressure leaks in a timely manner. Pressure leaks not only lead to wasted anesthetic gas and increased medical costs, but can also affect the depth of anesthesia and respiratory function, endangering the patient's life. Furthermore, existing protection mechanisms are inadequate when the system malfunctions, failing to provide rapid and effective measures to ensure patient safety. Summary of the Invention

[0007] The purpose of this invention is to provide an electric pressure regulating system for anesthesia mask tubes to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: an electric pressure regulating system for anesthesia mask tubes, the system comprising:

[0009] The physiological data acquisition module is used to collect real-time data on the pressure signal inside the anesthesia mask, the patient's respiratory rate, and blood oxygen saturation, and generate a multi-source synchronous physiological dataset.

[0010] The multimodal feature modeling module is used to input the multi-source physiological synchronization dataset into the Bayesian probabilistic graphical model, and extract the correlation features of respiratory phase, pressure fluctuation and physiological state through node conditional probability distribution and dynamic latent variable inference algorithm to generate a multidimensional joint probability feature matrix.

[0011] The dynamic stress decision module is used to construct an adaptive controller based on deep reinforcement learning, perform real-time policy optimization on the multi-dimensional joint probability feature matrix, and output the stress regulation target value.

[0012] The multi-objective collaborative control module is used to establish a mixed integer programming model based on the pressure regulation target value, and to optimize the opening gradient, response delay and energy consumption of the electric valve using an improved ant colony algorithm to generate a steady-state pressure control strategy.

[0013] Preferably, noise suppression and time alignment processing of the multi-source physiological synchronization dataset based on Kalman filtering and dynamic time warping algorithms specifically includes:

[0014] The high-frequency noise component and low-frequency trend component in the pressure signal are separated, and the high-frequency component is recursively filtered using a Kalman gain matrix.

[0015] The dynamic time warping algorithm is used to align the respiratory rate sequence with the pressure waveform, eliminating the time offset of physiological signals.

[0016] The blood oxygen saturation data are locally smoothed using the sliding window interpolation method to generate a time-domain consistent multi-source dataset.

[0017] Preferably, the generation of the multidimensional joint probability feature matrix includes:

[0018] Define respiratory phase nodes, pressure state nodes, and physiological index nodes as vertices of a Bayesian network, and construct a conditional probability transition table.

[0019] The variational inference algorithm is used to calculate the posterior distribution of latent variables and iteratively update the dependency weights between nodes;

[0020] A joint probability density function is generated using a Gibbs sampling strategy, and the output is a feature matrix containing nonlinear interaction relationships.

[0021] Preferably, the construction of the adaptive controller based on deep reinforcement learning includes:

[0022] The design state space is a tripartite of current pressure deviation, respiratory cycle phase angle, and historical regulatory actions;

[0023] The reward function is defined as a linear combination of pressure stability weights, valve action smoothness, and energy consumption penalty terms;

[0024] A dual-depth Q-network architecture is used to train the policy network and the value network in parallel, and the optimal valve control command is output.

[0025] Preferably, the construction steps of the mixed integer programming model include:

[0026] The objective function is set as a Pareto tradeoff between minimizing pressure fluctuation variance, minimizing valve life loss coefficient, and maximizing energy efficiency.

[0027] Add constraints such as the maximum permissible pressure deviation threshold, valve mechanical stroke limit, and breathing cycle synchronization tolerance range;

[0028] A pheromone evaporation factor is introduced to adaptively adjust the exploration capability of the ant colony algorithm, and a control strategy for screening the generation of non-dominated solution sets is adopted.

[0029] Preferably, the multi-source physiological synchronization dataset includes:

[0030] The system integrates real-time pressure waveforms acquired by an integrated piezoelectric sensor, tidal volume curves recorded by an infrared respiratory sensor, and physiological parameters monitored by a photoelectric blood oxygen probe.

[0031] The tensor decomposition algorithm is used to perform low-rank approximation completion on the missing data to construct a complete spatiotemporal data tensor.

[0032] Adaptive quantile normalization eliminates dimensional differences and distribution shifts between sensors.

[0033] Preferably, the system further includes:

[0034] A virtual pressure estimation module based on generative adversarial networks is constructed to simulate and reconstruct the turbulent pressure distribution at the endotracheal tube connection.

[0035] The reconstruction results are compared with the measured pressure data to generate a residual vector, which triggers the logic for pressure leakage diagnosis and redundant valve switching.

[0036] Preferably, the system further includes:

[0037] Design an abnormal interruption protection module based on fuzzy logic. When a sudden change in respiratory rate or blood oxygen saturation exceeding the limit is detected, the risk level is calculated using a membership function.

[0038] The preset safety pressure curve is switched according to the risk level, and the buzzer alarm signal and valve emergency lock-up command are triggered.

[0039] Preferably, the present invention also includes an electronic device, the device comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to realize the functions of each module in the above-described electric pressure regulation system for anesthesia mask tubes.

[0040] Compared with the prior art, the beneficial effects of the present invention are:

[0041] The electric pressure regulation system for anesthesia masks of this invention exhibits significant benefits in multiple aspects, greatly improving the accuracy, safety, and intelligence of pressure regulation during anesthesia. Regarding precise data acquisition and processing, the physiological data acquisition module integrates multiple sensors to collect real-time pressure signals inside the anesthesia mask, patient respiratory rate, and blood oxygen saturation data. Missing data is supplemented using tensor decomposition algorithms, and adaptive quantile normalization is employed to eliminate dimensional differences. Simultaneously, Kalman filtering and dynamic time warping algorithms are used for noise suppression and time-series alignment. This ensures that the acquired multi-source physiological synchronous dataset is accurate, complete, and temporally consistent, providing a solid foundation for subsequent precise analysis and decision-making. For example, Kalman filtering effectively suppresses high-frequency noise components in the pressure signal, avoiding noise interference with pressure monitoring and regulation, enabling the system to more accurately perceive the patient's true pressure needs. The dynamic time warping algorithm aligns the respiratory rate sequence with the pressure waveform, eliminating time offsets in physiological signals and ensuring the correlation and synchronization between data, allowing the system to perform precise pressure regulation based on the patient's real-time respiratory status.

[0042] The multimodal feature modeling module inputs multi-source synchronous physiological datasets into a Bayesian probabilistic graphical model. It constructs a conditional probability transition table by defining nodes for respiratory phase, pressure state, and physiological indicators. A variational inference algorithm is used to calculate the posterior distribution of latent variables and update node dependency weights. A Gibbs sampling strategy is then employed to generate a multidimensional joint probability feature matrix. This process deeply explores the complex nonlinear interactions between respiratory phase, pressure fluctuations, and physiological states, providing rich and accurate feature information for pressure decision-making. Compared to traditional methods, it goes beyond simple linear relationship analysis, comprehensively capturing the potential connections between data. For example, when analyzing the relationship between pressure fluctuations and blood oxygen saturation at different respiratory phases, it can more accurately grasp the underlying patterns, thus providing a more targeted basis for pressure regulation.

[0043] The dynamic pressure decision-making module constructs an adaptive controller based on deep reinforcement learning, designs a reasonable state space and reward function, and employs a dual-deep Q-network architecture to train the policy network and value network in parallel. In this way, the system can optimize its strategy in real time based on real-time pressure deviation, respiratory cycle phase angle, and historical adjustment actions, outputting a more accurate pressure regulation target value. The reward function comprehensively considers pressure stability weights, valve action smoothness, and energy consumption penalties, ensuring that the system not only maintains pressure stability during pressure regulation but also optimizes valve action and reduces energy consumption. For example, during surgery, the patient's respiratory status may change at any time. This module can quickly adjust the pressure regulation strategy based on real-time data to ensure that the pressure inside the anesthesia mask tube remains within the appropriate range, while reducing unnecessary frequent valve actions, extending valve life, and reducing energy consumption.

[0044] The multi-objective collaborative control module establishes a mixed-integer programming model based on the pressure regulation target value, sets reasonable objective functions and constraints, and uses an improved ant colony algorithm to optimize the opening gradient, response delay, and energy consumption of the electric valve, generating a steady-state pressure control strategy. This module achieves collaborative optimization of multiple objectives, minimizing pressure fluctuation variance while reducing valve lifespan loss coefficient and improving energy efficiency. For example, by optimizing the valve opening gradient, pressure regulation becomes more stable, avoiding adverse effects on patients from sudden pressure changes; reasonable control of response delay ensures the system can respond promptly to pressure changes; and optimized energy consumption reduces medical costs.

[0045] The system also includes a virtual pressure estimation module based on generative adversarial networks and an anomaly interruption protection module based on fuzzy logic. The virtual pressure estimation module simulates and reconstructs the turbulent pressure distribution at the endotracheal tube connection point. By comparing the reconstructed results with measured pressure data, a residual vector is generated to realize pressure leak diagnosis and redundant valve switching logic, significantly improving the system's reliability and safety. When a pressure leak is detected, it can promptly switch to redundant valves to ensure the continuity of the anesthesia process and prevent the patient's anesthesia effect from being affected or life-threatening due to pressure leakage. The anomaly interruption protection module, upon detecting a sudden change in respiratory rate or excessive blood oxygen saturation, uses a membership function to calculate the risk level, switches to a preset safety pressure curve based on the risk level, and triggers a buzzer alarm signal and an emergency valve locking command, effectively ensuring the patient's safety during anesthesia. For example, when a patient experiences an emergency such as a sudden increase in respiratory rate or a sharp drop in blood oxygen saturation, this module can react quickly and take appropriate measures to prevent further escalation of the danger. Attached Figure Description

[0046] Figure 1 This is a schematic diagram illustrating the working principle of an electric pressure regulating system for anesthesia mask tubes according to the present invention.

[0047] Figure 2 A schematic diagram for constructing an adaptive controller based on deep reinforcement learning;

[0048] Figure 3 A flowchart illustrating the steps involved in processing multi-source physiological synchronization datasets;

[0049] Figure 4 This is a flowchart of the virtual pressure estimation and related logic. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Please see Figures 1-4 This invention provides an electrically adjustable system for the internal pressure of an anesthesia mask tube, aiming to achieve precise and intelligent adjustment of the internal pressure of the anesthesia mask tube to meet the physiological needs of patients during anesthesia. The overall implementation scheme is as follows:

[0052] The physiological data acquisition module is responsible for acquiring real-time data on the pressure signal inside the anesthesia mask, the patient's respiratory rate, and blood oxygen saturation, generating a multi-source synchronous physiological dataset. By integrating various sensors, such as a piezoelectric sensor for acquiring real-time pressure waveforms, an infrared respiratory sensor for recording tidal volume curves, and a photoelectric pulse oximeter for monitoring physiological parameters, it ensures the acquisition of comprehensive and accurate physiological information.

[0053] The multimodal feature modeling module inputs multi-source synchronous physiological datasets into a Bayesian probabilistic graphical model. In this model, respiratory phase nodes, pressure state nodes, and physiological indicator nodes are defined as vertices of the Bayesian network, and a conditional probability transition table is constructed. Next, a variational inference algorithm is used to calculate the posterior distribution of latent variables, iteratively updating the dependency weights between nodes. Finally, a multidimensional joint probability feature matrix containing correlation features of respiratory phase, pressure fluctuation, and physiological state is generated using a Gibbs sampling strategy.

[0054] A dynamic pressure decision-making module is constructed, featuring an adaptive controller based on deep reinforcement learning. The state space is designed as a triplet of current pressure deviation, respiratory cycle phase angle, and historical regulation actions. The reward function is defined as a linear combination of pressure stability weights, valve action smoothness, and energy consumption penalty terms. A dual-deep Q-network architecture is used to train the policy network and value network in parallel, thereby optimizing the multi-dimensional joint probability feature matrix in real time and outputting the target pressure regulation value.

[0055] The multi-objective collaborative control module establishes a mixed-integer programming model based on the pressure regulation target value. The objective function is defined as a Pareto tradeoff between minimizing pressure fluctuation variance, minimizing valve life loss coefficient, and maximizing energy efficiency. Constraints such as the maximum allowable pressure deviation threshold, valve mechanical stroke limit, and respiratory cycle synchronization tolerance range are also added. An improved ant colony algorithm is employed to optimize the opening gradient, response delay, and energy consumption of the electric valve, generating a steady-state pressure control strategy to effectively regulate the pressure inside the anesthesia mask tube.

[0056] The implementation of the present invention will be further described below with reference to Examples 1 to 6.

[0057] Example 1:

[0058] In practical applications, a series of preprocessing operations are required to ensure the quality of the collected multi-source physiological synchronization datasets.

[0059] During the data acquisition phase, integrated piezoelectric sensors, infrared respiratory sensors, and photoelectric pulse oximeters were used to collect real-time pressure waveforms, tidal volume curves, and physiological parameters, respectively. Since the sensor-acquired data may contain missing values, a tensor decomposition algorithm was used to perform low-rank approximation completion of the missing data. Assume the acquired data tensor is X∈R. I×J×KWhere I represents the time dimension, J represents the sensor type dimension, and K represents the sample dimension. X is decomposed into a core tensor G and multiple factor matrices U using a tensor decomposition algorithm. i (i = 1, 2, ...), i.e., X ≈ G × 1 U1 × 2 U2 × 3 U3. In this way, missing data is filled in to construct a complete spatiotemporal data tensor.

[0060] To eliminate dimensional differences and distribution shifts between sensors, an adaptive quantile normalization method is employed. For each sensor's acquired data sequence x = [x1, x2, ..., x...],... n We calculate its quantile q(x), and then normalize the data according to the adaptive rule to make the data from different sensors comparable.

[0061] In the data processing stage, noise suppression and time alignment are performed on the multi-source physiological synchronization dataset based on Kalman filtering and dynamic time warping algorithms.

[0062] For a pressure signal, it is separated into a high-frequency noise component and a low-frequency trend component. Assuming the pressure signal is p(t), it can be expressed as p(t) = p through a specific filtering algorithm. high (t)+p low (t), where p high (t) represents the high-frequency noise component, p low (t) represents the low-frequency trend component. The Kalman gain matrix K is used to analyze the high-frequency component p. high (t) Recursive filtering is performed. The Kalman gain matrix K is calculated based on the system's state transition matrix A, observation matrix H, process noise covariance matrix Q, and observation noise covariance matrix R. In each recursive step, the Kalman gain matrix K is updated according to the current estimated state and observations, thereby effectively suppressing high-frequency noise components.

[0063] A dynamic time warping algorithm is used to align respiratory rate sequences and pressure waveforms. The respiratory rate sequence r(t) and pressure waveform p(t) may have time offsets due to physiological differences or limitations of the acquisition equipment. The dynamic time warping algorithm finds an optimal time warping path to align the two sequences in time, eliminating the time offset of the physiological signals. Specifically, the distance matrix D(i,j) between the two sequences is calculated, where i and j represent the time points in the two sequences. Using a dynamic programming algorithm, an optimal path is found from the starting point (1,1) to the ending point (m,n) (where m and n are the lengths of the two sequences, respectively), minimizing the sum of the distances along the path, thus achieving alignment between the respiratory rate sequence and the pressure waveform.

[0064] For blood oxygen saturation data, local smoothing is performed using sliding window interpolation. Assuming the blood oxygen saturation data sequence is o(t), a sliding window w is set, with a window size of N. Within each window, an interpolation algorithm is used to smooth the data. For example, for the data [o(t)] within the window... i ,o i+1 ,…,o i+N-1 By employing methods such as linear interpolation or spline interpolation, smoothed intermediate values ​​are calculated to generate a time-domain consistent multi-source dataset, providing a high-quality data foundation for subsequent data analysis and processing.

[0065] Example 2:

[0066] Generating a multidimensional joint probability feature matrix in the multimodal feature modeling module is a key step in realizing intelligent decision-making in the system.

[0067] In a Bayesian probabilistic graphical model, respiratory phase nodes, pressure state nodes, and physiological index nodes are defined as vertices of the Bayesian network. Respiratory phase nodes represent different stages in a patient's respiratory process, such as the beginning, middle, and end of the inspiratory and expiratory phases; pressure state nodes reflect changes in pressure within the anesthesia mask, such as high pressure, low pressure, and stable pressure; and physiological index nodes cover the states corresponding to physiological parameters such as the patient's respiratory rate and blood oxygen saturation.

[0068] A conditional probability transition table is constructed to describe the probabilistic dependencies between nodes. Assume there are *m* states for the respiratory phase node, *n* states for the pressure state node, and *k* states for the physiological indicator node. For each respiratory phase state *i* (i = 1, ..., m), pressure state *j* (j = 1, ..., n), and physiological indicator state *l* (l = 1, ..., k), a conditional probability P(j|i, l) is defined, representing the probability that the pressure state is *j* given the respiratory phase and physiological indicator states. Through extensive statistical analysis of historical data, the values ​​of these conditional probabilities are determined, and a complete conditional probability transition table is constructed.

[0069] A variational inference algorithm is used to compute the posterior distribution of latent variables and iteratively update the dependency weights between nodes. The variational inference algorithm approximates the true posterior distribution P(z|x) of the latent variable z by introducing a variational distribution q(z), where x represents the observed data. The parameters of the variational distribution are continuously optimized by minimizing the KL divergence KL(q(z)||P(z|x)) between the variational distribution q(z) and the true posterior distribution P(z|x). In each iteration, the dependency weights between nodes are updated based on the current variational distribution and the observed data, enabling the model to better reflect the underlying relationships in the data.

[0070] A joint probability density function is generated using a Gibbs sampling strategy. Gibbs sampling is a sampling method based on Markov chain Monte Carlo (MCMC). Starting from the initial state, each node is sampled sequentially. The sampling probability of that node is calculated based on the current states of other nodes and the conditional probability transition table. For example, for the breathing phase node i, with other nodes fixed, the sampling probability is calculated based on the conditional probability P(i|j1,…,j…). s (j1,…,j) s The states of other nodes are sampled. Through multiple samplings, a series of samples are obtained, the distribution of which approximates the joint probability density function. Finally, based on these samples, a feature matrix containing nonlinear interaction relationships is output. This matrix comprehensively reflects the complex correlation between respiratory phase, pressure fluctuations, and physiological state, providing rich feature information for subsequent dynamic pressure decision-making.

[0071] Example 3:

[0072] The dynamic pressure decision-making module achieves optimized output of the pressure regulation target value by constructing an adaptive controller based on deep reinforcement learning.

[0073] The design state space is a tripartite relationship between the current pressure deviation, the respiratory cycle phase angle, and historical regulatory actions. The current pressure deviation Δp is defined as the current actual pressure p. actual With the preset target pressure p target The difference, i.e., Δp = p actual -p target The respiratory cycle phase angle θ describes the patient's current position within the respiratory cycle, typically ranging from [0, 2π], and can be calculated by monitoring respiratory rate and time. Historical adjustment actions record previous adjustments to the electric valve, such as changes in valve opening and adjustment time.

[0074] The reward function is defined as a linear combination of pressure stability weights, valve action smoothness weights, and energy consumption penalty terms. Let the reward function be R, the pressure stability weight be w1, the valve action smoothness weight be w2, and the energy consumption penalty term weight be w3. The pressure stability weight w1 reflects the importance placed on pressure stability; the larger w1 is, the more the system tends to maintain pressure stability. The valve action smoothness weight w2 measures the smoothness of valve action, avoiding frequent and drastic valve adjustments. The energy consumption penalty term weight w3 penalizes high-energy-consuming adjustments, prompting the system to reduce energy consumption while adjusting pressure. The specific reward function can be expressed as R = w1 × S - w2 × |aa| prev |-w3×E, where S is the pressure stability index, for example, the reciprocal of the pressure deviation; a is the current valve adjustment action, a prevE represents the previous valve adjustment action; E is the energy consumption indicator, such as the electrical energy consumption during the adjustment process of an electric valve.

[0075] A dual-deep Q-network architecture is employed to train the policy network and value network in parallel. Dual-deep Q-networks (DDQNs) reduce the overestimation problem present in deep Q-networks (DQNs) by decoupling action selection and action evaluation. The policy network π(s; θ) π Output an action a based on the current state s, where θ π These are the parameters of the policy network. Value network Q(s,a; θ) Q Evaluate the value of performing action a in state s, where θ Q Here are the parameters of the value network. During training, an action 'a' is first selected based on the policy network, and then the value of that action is evaluated using the value network. This is achieved by minimizing the loss function L(θ). Q The parameters of the value network are updated using a method called [missing information - likely a function or method], and the loss function is typically calculated based on temporal difference error. Simultaneously, the parameters of the value network are periodically copied to the target value network for calculating the target value, improving training stability. Through continuous training, the policy network gradually learns the optimal valve control strategy and outputs the optimal valve control command to effectively regulate the pressure inside the anesthesia mask tube.

[0076] Example 4:

[0077] The multi-objective collaborative control module generates a steady-state pressure control strategy by constructing a mixed integer programming model and using an improved ant colony algorithm to optimize the opening gradient, response delay, and energy consumption of the electric valve.

[0078] The steps for constructing a mixed-integer programming model are as follows:

[0079] The objective function is defined as a Pareto tradeoff between minimizing pressure fluctuation variance, minimizing valve life loss coefficient, and maximizing energy efficiency. Let the pressure fluctuation variance be... The valve life loss coefficient is c, and the energy efficiency is η. The objective function can be expressed as: ω1, ω2, and ω3 are weighting coefficients used to balance the importance of different objectives. By adjusting these weighting coefficients, different objectives can be optimized according to actual needs. For example, when high pressure stability is required, the value of w1 can be appropriately increased.

[0080] Add constraints as the maximum permissible pressure deviation threshold, valve mechanical stroke limit, and breathing cycle synchronization tolerance range. Maximum permissible pressure deviation threshold Δp max The range of pressure fluctuations is limited to ensure that the pressure inside the anesthesia mask does not exceed the safe range. The valve's mechanical stroke limit specifies the maximum and minimum opening degree of the electric valve; let the maximum valve opening degree be a.max The minimum opening is a min Then the valve opening degree 'a' must satisfy 'a'. min ≤a≤a max The respiratory cycle synchronization tolerance range is used to ensure the synchronization of valve adjustment with the patient's respiratory cycle. Let the respiratory cycle be T, and the synchronization tolerance range be [T]. min ,T max During the adjustment process, the valve's operating time must be within this tolerance range to avoid adverse effects on the patient's breathing.

[0081] A pheromone evaporation factor (ρ) is introduced to adaptively adjust the exploration capability of the ant colony algorithm, and a control strategy is generated by screening non-dominated solution sets. The ant colony algorithm is an optimization algorithm that simulates the foraging behavior of ants. Ants release pheromones along their paths to guide other ants to find the optimal path. In the improved ant colony algorithm, a pheromone evaporation factor (ρ) is introduced. As the number of iterations increases, the pheromone gradually evaporates, preventing premature convergence. Simultaneously, the value of the pheromone evaporation factor is adaptively adjusted according to the characteristics of the problem. During the algorithm's operation, ants choose paths based on pheromone concentration and heuristic information, and the pheromone concentration is updated after each iteration. Through multiple iterations, a non-dominated solution set is selected, which is a set of solutions that satisfy multiple objectives. A suitable solution is selected from these non-dominated solution sets as a steady-state pressure control strategy to optimize the opening gradient, response delay, and energy consumption of the electric valve, ensuring that the pressure inside the anesthesia mask tube remains stable within a suitable range.

[0082] Example 5:

[0083] This system introduces a virtual pressure estimation module based on generative adversarial networks, which aims to simulate and reconstruct the turbulent pressure distribution at the endotracheal tube connection, thereby enabling pressure leakage diagnosis and redundant valve switching, thus improving the stability and reliability of the system.

[0084] Generative Adversarial Networks (GANs) primarily consist of a generator and a discriminator. The generator receives a random noise vector as input. In practical applications, this random noise vector can be a set of data randomly sampled from a specific distribution (such as a normal distribution), denoted by the symbol z. The generator processes the input random noise vector through its complex neural network structure, ultimately outputting simulated turbulent pressure distribution data at the endotracheal tube connection point, denoted as z. The discriminator's role is to determine the authenticity of the input data. Its inputs include real pressure data p and simulated data generated by the generator. The discriminator's internal neural network extracts and analyzes features from the input data, and then outputs a judgment result indicating whether the input data is real data or simulated data.

[0085] In training a generative adversarial network (GAN), the generator and discriminator engage in adversarial training. The generator aims to continuously adjust its neural network parameters to generate simulated data that increasingly resembles real stress distribution data, thereby deceiving the discriminator into classifying its output as real data. The discriminator, on the other hand, strives to improve its ability to distinguish between real and simulated data. During this process, both continuously optimize their parameters, gradually reaching an equilibrium. As training continues, the generator becomes capable of generating simulated data that closely resembles the real stress distribution.

[0086] After completing the simulation reconstruction, the reconstruction results are compared with the measured pressure data to generate a residual vector. Assume the measured pressure data is p. real The simulated reconstructed pressure data is p sim Then the residual vector r is calculated as r = p real -p sim To determine whether a pressure leak exists, a preset threshold ∈ needs to be set. When a component of the residual vector exceeds this preset threshold ∈, the system will determine that a pressure leak may exist.

[0087] Once a pressure leak is detected, the system quickly triggers pressure leak diagnosis and redundant valve switching logic. First, the system performs in-depth analysis of the residual vector, studying its characteristics (such as magnitude and trend) and its temporal and spatial distribution to determine the location and extent of the leak. For example, if the component of the residual vector in a specific area increases sharply at a certain moment and remains at a high level for a period of time, a pressure leak can be preliminarily identified in that area. After determining the leak, the system quickly switches to the backup valve based on pre-set redundant valve configuration information. During the switchover, the system ensures that the opening degree and adjustment strategy of the backup valve match the original valve, thereby maintaining stable pressure within the anesthesia mask tubing and ensuring the patient's normal anesthesia status. Simultaneously, the system records detailed information about the pressure leak event, including the time of occurrence, leak location, and relevant pressure data, for subsequent troubleshooting and system maintenance.

[0088] Example 6:

[0089] The system is specially designed with an abnormal interruption protection module based on fuzzy logic, which plays a key role in ensuring the safety of patients during anesthesia.

[0090] When the system detects a sudden change in respiratory rate or an excess of blood oxygen saturation, it uses a membership function to calculate the risk level. Assuming respiratory rate is represented by f, and the preset normal respiratory rate range is [f...] min ,f max Blood oxygen saturation is used for... This indicates that the normal blood oxygen saturation range is [S]. min ,S max For cases of sudden changes in respiratory rate and excessive blood oxygen saturation, membership functions μ are defined respectively. f (f) and Taking the membership function for abrupt changes in respiratory rate as an example, a Gaussian membership function can be used, the expression of which is: Where f center This represents the center value of the normal respiratory rate, and it can take the range of the normal respiratory rate [f]. min ,f max The middle value of ], that is σ is the standard deviation, which determines the shape and rate of change of the membership function and can be adjusted according to the actual situation. When the respiratory rate f exceeds the normal range, (ff...) center ) 2 The value of μ will increase, causing μ to... f The value of (f) also increases accordingly, which means that the possibility of sudden changes in respiratory rate increases. Similarly, the membership function for blood oxygen saturation exceeding the limit can also be defined in a similar form to accurately reflect the degree to which blood oxygen saturation deviates from the normal range.

[0091] Based on the calculated membership function values ​​for respiratory rate mutations and oxygen saturation exceeding limits, the system uses fuzzy inference rules to calculate the risk level. Generally, the risk level is divided into three levels: low risk, medium risk, and high risk. For example, if both the membership function values ​​for respiratory rate mutations and oxygen saturation exceeding limits are low, it indicates that although the patient's physiological state has changed to some extent, it is still within a relatively safe range, and the risk level is determined to be low risk. If one membership function value is high and the other is at a medium level, it indicates that the patient's physiological state faces some risk, and the risk level is medium risk. When both membership function values ​​are high, it means that the patient's physiological state faces significant risk, and the risk level is high risk.

[0092] The system takes corresponding measures based on different risk levels. When the risk level is low, the system will issue a prompt to remind medical staff to pay attention to changes in the patient's physiological state so as to detect potential problems in a timely manner. If the risk level is medium, the system will immediately switch to a preset safety pressure curve. This safety pressure curve is preset based on a large amount of clinical data and professional knowledge. Under the premise of ensuring patient safety, it appropriately adjusts the pressure inside the anesthesia mask tube to maintain the patient's physiological stability. At the same time, the system will trigger a buzzer alarm signal to attract the attention of medical staff so that they can take further measures in a timely manner. When the risk level reaches high, in addition to switching the safety pressure curve and triggering the buzzer alarm signal, the system will immediately trigger an emergency valve closure command. This command will quickly close the relevant valves to prevent abnormal pressure from causing further harm to the patient, thereby effectively ensuring the patient's safety during anesthesia.

[0093] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0094] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An electrically adjustable pressure system for the tube of an anesthesia mask, characterized in that, The system includes: The physiological data acquisition module is used to collect real-time data on the pressure signal inside the anesthesia mask, the patient's respiratory rate, and blood oxygen saturation, and generate a multi-source synchronous physiological dataset. The multimodal feature modeling module is used to input the multi-source physiological synchronization dataset into the Bayesian probabilistic graphical model, and extract the correlation features of respiratory phase, pressure fluctuation and physiological state through node conditional probability distribution and dynamic latent variable inference algorithm to generate a multidimensional joint probability feature matrix. The dynamic stress decision module is used to construct an adaptive controller based on deep reinforcement learning, perform real-time policy optimization on the multi-dimensional joint probability feature matrix, and output the stress regulation target value. The multi-objective collaborative control module is used to establish a mixed integer programming model based on the pressure regulation target value, and to optimize the opening gradient, response delay and energy consumption of the electric valve using an improved ant colony algorithm to generate a steady-state pressure control strategy. Ant colony optimization (ACO) is an optimization algorithm that simulates the foraging behavior of ants. Ants release pheromones along their paths to guide other ants to find the optimal path. The improved ACO introduces a pheromone evaporation factor. As the number of iterations increases, the pheromone gradually evaporates, preventing premature convergence. The pheromone evaporation factor is also adaptively adjusted according to the characteristics of the problem. During the operation of the improved ACO, ants select paths based on pheromone concentration and heuristic information. The pheromone concentration is updated after each iteration, and through multiple iterations, a set of non-dominated solutions—that is, a set of solutions that satisfies multiple objectives—is selected. The construction of the adaptive controller based on deep reinforcement learning includes: The design state space is a tripartite of current pressure deviation, respiratory cycle phase angle, and historical regulatory actions; The reward function is defined as a linear combination of pressure stability weights, valve action smoothness, and energy consumption penalty terms; A dual-depth Q-network architecture is used to train the policy network and the value network in parallel, and the optimal valve control command is output. The construction steps of the mixed integer programming model include: The objective function is set as a Pareto tradeoff between minimizing pressure fluctuation variance, minimizing valve life loss coefficient, and maximizing energy efficiency. Add constraints such as the maximum permissible pressure deviation threshold, valve mechanical stroke limit, and breathing cycle synchronization tolerance range; A pheromone evaporation factor is introduced to adaptively adjust the exploration capability of the ant colony algorithm, and a control strategy for screening the generation of non-dominated solution sets is adopted.

2. The electric pressure regulating system inside the anesthesia mask tube as described in claim 1, characterized in that, The multi-source physiological synchronization dataset is subjected to noise suppression and time alignment processing based on Kalman filtering and dynamic time warping algorithms, specifically including: The high-frequency noise component and low-frequency trend component in the pressure signal are separated, and the high-frequency component is recursively filtered using a Kalman gain matrix. The dynamic time warping algorithm is used to align the respiratory rate sequence with the pressure waveform, eliminating the time offset of physiological signals. The blood oxygen saturation data are locally smoothed using the sliding window interpolation method to generate a time-domain consistent multi-source dataset.

3. The electric pressure regulating system inside the anesthesia mask tube as described in claim 1, characterized in that, The generation of the multidimensional joint probability feature matrix includes: Define respiratory phase nodes, pressure state nodes, and physiological index nodes as vertices of a Bayesian network, and construct a conditional probability transition table. The variational inference algorithm is used to calculate the posterior distribution of latent variables and iteratively update the dependency weights between nodes; A joint probability density function is generated using a Gibbs sampling strategy, and the output is a feature matrix containing nonlinear interaction relationships.

4. The electric pressure regulating system inside the anesthesia mask tube as described in claim 1, characterized in that, The multi-source physiological synchronization dataset includes: The system integrates real-time pressure waveforms acquired by an integrated piezoelectric sensor, tidal volume curves recorded by an infrared respiratory sensor, and physiological parameters monitored by a photoelectric blood oxygen probe. The tensor decomposition algorithm is used to perform low-rank approximation completion on the missing data to construct a complete spatiotemporal data tensor. Adaptive quantile normalization eliminates dimensional differences and distribution shifts between sensors.

5. The electric pressure regulating system inside the anesthesia mask tube as described in claim 1, characterized in that, The system also includes: A virtual pressure estimation module based on generative adversarial networks is constructed to simulate and reconstruct the turbulent pressure distribution at the endotracheal tube connection. The reconstruction results are compared with the measured pressure data to generate a residual vector, which triggers the logic for pressure leakage diagnosis and redundant valve switching.

6. The electric pressure regulating system inside the anesthesia mask tube as described in claim 1, characterized in that, The system also includes: Design an abnormal interruption protection module based on fuzzy logic. When a sudden change in respiratory rate or blood oxygen saturation exceeding the limit is detected, the risk level is calculated using a membership function. The preset safety pressure curve is switched according to the risk level, and the buzzer alarm signal and valve emergency lock-up command are triggered.

7. An electronic device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the functions of the various modules of the anesthesia mask tube pressure electric adjustment system as described in any one of claims 1-6.

Images