A method and system for monitoring the status of anesthesia nebulization equipment

Abstract

This invention relates to the field of medical equipment monitoring technology, specifically to an intelligent method and system for monitoring the status of anesthesia nebulizer equipment. First, the invention collects nebulizer flow rate, particle size distribution, motor vibration frequency, drug solution temperature, and tubing pressure to construct a set of equipment status features. Simultaneously, it collects the patient's electroencephalogram (EEG) signals and constructs an EEG state vector. The feature set is then input into a support vector machine (SVM) model to identify the equipment's operating status and determine if any abnormalities exist. When the equipment is in normal condition, a nebulizer efficacy decay equation is constructed based on continuous data to predict future efficacy trends. The prediction results are then input into a dynamic balance model of the anesthesia depth index along with the EEG status to assess the adaptability of the nebulizer output to the patient's sedation response. When the efficacy decay rate is too high or the dynamic error exceeds a threshold, a prompt signal is generated and uploaded to a remote platform. This invention achieves linked control of equipment status monitoring, efficacy prediction, and neurofeedback, improving intraoperative safety and postoperative assessment accuracy.

Description

Technical Field

[0001] This invention relates to the field of medical equipment monitoring technology, specifically to an intelligent method and system for monitoring the status of anesthesia nebulization equipment. Background Technology

[0002] With the widespread application of anesthetic nebulization technology in non-invasive sedation, preoperative induction, and local anesthesia, ensuring the stability and safety of the nebulization process has become a crucial clinical concern. This invention aims to enhance the intelligent monitoring capabilities of the equipment, enabling simultaneous perception and dynamic control of anesthetic efficacy and the patient's physiological state.

[0003] Current anesthetic nebulization equipment generally relies on manual observation of nebulized particles, equipment noise, or simple threshold alarms for status assessment. This makes it difficult to promptly identify critical issues such as abnormal particles, uneven nebulization, or tubing blockage, and lacks predictive maintenance mechanisms for equipment operation trends. Furthermore, while existing monitoring systems can collect physiological parameters such as heart rate and blood oxygenation, they lack a coupling mechanism with equipment status, making it difficult to dynamically assess the suitability of nebulization output for the patient's sedation response. This can easily lead to deviations in anesthesia depth control, posing potential risks.

[0004] Therefore, this invention proposes an intelligent method and system for monitoring the status of anesthesia nebulization equipment. Summary of the Invention

[0005] This invention provides an intelligent method and system for monitoring the status of anesthesia nebulization equipment, aiming to improve the intelligent monitoring capability of equipment operation and realize the synchronous perception and dynamic control of anesthetic efficacy and patient physiological status.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A method for monitoring the status of an intelligent anesthesia nebulizer includes:

[0008] The operating parameters of the anesthesia nebulization device are collected, including nebulization flow rate, particle size distribution, motor vibration frequency, drug solution temperature and pipeline pressure. The device status feature set is constructed, and the patient's electroencephalogram (EEG) signal is collected simultaneously. The spectral features reflecting the sedation state are extracted, and the EEG state vector is constructed.

[0009] The device status feature set is input into the status recognition model to identify the current operating status of the device. When an abnormal status is identified, a prompt signal is generated and the device status information is uploaded.

[0010] When identified as a normal state, atomization-related characteristic indicators are calculated based on continuous operation data and substituted into the atomization efficiency decay equation to predict the future trend of atomization efficiency.

[0011] The nebulization efficiency and EEG state vector are input into the dynamic balance model of the anesthesia depth index to calculate the dynamic error between nebulization output and neural response.

[0012] When the efficiency decay rate exceeds the set threshold, or the dynamic error continues to exceed the target range, a prompt signal is generated, and the nebulization prediction results, EEG state vector, and error information are uploaded to the remote monitoring platform.

[0013] As a preferred technical solution of the invention, the step of constructing the device state feature set includes:

[0014] Multiple sets of sensors installed inside the atomizing device and at the gas path nodes collect atomization flow sensing signals, particle size detection signals, motor acceleration sensing signals, liquid temperature signals, and gas path pressure signals, respectively.

[0015] Each signal is filtered, denoised, and standardized to extract statistical characteristic parameters that characterize the equipment's operating status, including mean flow rate, peak particle size, vibration spectrum energy, temperature gradient, and pressure fluctuation amplitude.

[0016] The above feature parameters are combined into a multi-dimensional state vector, which is then used as the device state feature set input to the subsequent state recognition model.

[0017] As a preferred embodiment of the invention, the step of constructing the EEG state vector includes:

[0018] Real-time brainwave signals are acquired using EEG electrodes;

[0019] The EEG signal is filtered and artifact removal is performed to extract the power spectral density values ​​of the alpha and beta wave frequency bands.

[0020] The sedation index is calculated based on the ratio of alpha wave power to beta wave power, and the sedation index is used to reflect the depth of anesthesia in patients.

[0021] The alpha wave power, beta wave power, and EEG sedation index are used to construct an EEG state vector to evaluate the compatibility between nebulization efficacy and patient neurological responses.

[0022] As a preferred embodiment of the invention, the state recognition model is a support vector machine classification model built based on a support vector set, and its specific implementation includes:

[0023] Collect historical operating parameters of the equipment under various operating conditions, construct a training sample set, and standardize the equipment state feature set in the sample, and attach corresponding operating state labels;

[0024] A set of support vectors for constructing support boundaries is selected from the training sample set, and the support vector machine classification model is trained.

[0025] The currently collected equipment status feature set is input into the support vector machine classification model, and the current operating status label of the equipment is output to determine whether there are conditions such as uneven atomization, abnormal particles, depletion of liquid, pipeline blockage, or dry burning.

[0026] As a preferred technical solution of the invention, the step of predicting the future trend of atomization efficiency includes:

[0027] The vibration frequency, gas pressure and liquid temperature signals collected over a continuous time period were smoothed using the exponential weighted moving average method to obtain the weighted average of each indicator.

[0028] Substitute the weighted average value into the atomization efficiency decay equation to calculate the atomization efficiency estimate for multiple time points;

[0029] Based on the estimated values, a trend curve of atomization efficiency changes over a future time period is constructed to determine whether there is a risk of decreased device efficiency.

[0030] The formula for calculating the atomization efficiency attenuation equation is as follows:

[0031] ;

[0032] in, For initial performance; These are the smoothed motor vibration frequency index, gas pressure index, and liquid temperature index, respectively. These are the attenuation coefficients for each parameter.

[0033] As a preferred embodiment of the invention, the implementation steps of the dynamic balance model of the anesthesia depth index include:

[0034] Within a set time window, the predicted efficacy values ​​at each moment in the nebulization efficacy trend curve are paired with the corresponding EEG state vectors for that time period.

[0035] Extracting the brainwave sedation index from the brainwave state vector Based on the current nebulization efficiency, target nebulization efficiency, EEG sedation index, and target sedation value, the following dynamic equilibrium model is constructed:

[0036] ;

[0037] in, This is a performance prediction value; Target atomization efficiency; The preset target sedation value; and This is an empirical adjustment coefficient used to balance the relative influence of device output and neural response;

[0038] Through calculation The value of is used to obtain the dynamic error rate of change between anesthetic efficacy and neural response, and to determine whether it is within the set adaptation range.

[0039] As a preferred embodiment of the invention, the step of generating a prompt signal and uploading information includes:

[0040] The first-order difference of the predicted nebulization efficacy over a continuous time period is calculated to obtain the efficacy decay rate, which is then compared with a preset threshold. It is also determined whether the dynamic error change rate continuously exceeds the anesthesia depth adaptation range.

[0041] When any judgment condition is met, an abnormal prompt signal is generated. The prompt types include "performance degradation abnormality", "adaptation imbalance" or "combined abnormality".

[0042] Construct a data package containing the current atomization efficiency trend prediction results, EEG state vectors, dynamic error values, and prompt types;

[0043] The data packet is uploaded to the remote monitoring platform via the wireless communication module, along with a timestamp and device ID information.

[0044] The present invention also includes an intelligent anesthesia nebulization device status monitoring system, comprising:

[0045] The parameter acquisition module collects the operating parameters of the anesthesia nebulization device, including nebulization flow rate, particle size distribution, motor vibration frequency, drug solution temperature and pipeline pressure, constructs a device state feature set, and simultaneously collects the patient's electroencephalogram (EEG) signals, extracts spectral features reflecting the sedation state, and constructs an EEG state vector.

[0046] The status recognition module is used to input the device status feature set into the status recognition model, identify the current operating status of the device, and generate a prompt signal and upload the device status information when an abnormal status is identified.

[0047] The performance prediction module, when identified as being in a normal state, calculates atomization-related characteristic indicators based on continuous operating data and substitutes them into the atomization performance decay equation to predict the future trend of atomization performance.

[0048] The error assessment module is used to input the nebulization efficiency and EEG state vector into the dynamic balance model of the anesthesia depth index to calculate the dynamic error between nebulization output and neural response.

[0049] The abnormality alert and upload module generates an alert signal when the efficiency decay rate exceeds a set threshold or the dynamic error continues to exceed the target range. It also uploads the nebulization prediction results, EEG state vector, and error information to the remote monitoring platform to assist in intraoperative judgment and postoperative evaluation.

[0050] The beneficial effects of this invention are:

[0051] 1. This invention constructs a nebulization efficiency decay equation and combines it with multi-dimensional operating parameters to achieve dynamic modeling and trend prediction of anesthesia nebulization performance, which is more continuous and forward-looking than existing schemes that rely on single parameter fluctuations. This mechanism not only improves the accuracy of performance evaluation but also provides data support for equipment maintenance and drug management. Combined with an anomaly early warning mechanism, it significantly enhances the stability and controllability of the nebulization process.

[0052] 2. This invention integrates the patient's EEG state vector with the trend of device nebulization efficiency for analysis, and constructs a dynamic balance model of the anesthesia depth index to achieve closed-loop monitoring of "device output - physiological response". This mechanism effectively compensates for the problem that existing devices only focus on physical parameters and ignore individual differences in patient response, improving the accuracy and individualization of sedation control, and has broad clinical application value. Attached Figure Description

[0053] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0054] Figure 1 This is a flowchart illustrating the intelligent anesthesia nebulization device status monitoring method of the present invention;

[0055] Figure 2 This is a schematic diagram of the structure of an intelligent anesthesia nebulization equipment status monitoring system of the present invention. Detailed Implementation

[0056] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0057] Example 1: As Figure 1 The present invention provides an intelligent method for monitoring the status of anesthesia nebulization equipment, comprising:

[0058] S100: Collect the operating parameters of the anesthesia nebulization device, including nebulization flow rate, particle size distribution, motor vibration frequency, drug solution temperature and pipeline pressure, construct a device state feature set, simultaneously collect the patient's electroencephalogram (EEG) signal, extract spectral features reflecting the sedation state, and construct an EEG state vector.

[0059] Furthermore, the step of constructing the device state feature set includes:

[0060] Multiple sets of sensors installed inside the atomizing device and at the gas path nodes collect atomization flow sensing signals, particle size detection signals, motor acceleration sensing signals, liquid temperature signals, and gas path pressure signals, respectively.

[0061] Each signal is filtered, denoised, and standardized to extract statistical characteristic parameters that characterize the equipment's operating status, including mean flow rate, peak particle size, vibration spectrum energy, temperature gradient, and pressure fluctuation amplitude.

[0062] The above feature parameters are combined into a multi-dimensional state vector, which is then used as the device state feature set input to the subsequent state recognition model.

[0063] Specifically, the following five types of sensor components are deployed inside the anesthesia nebulizer and its key gas path nodes to collect core parameters involved in the operation of the equipment: a thermal mass flow sensor is installed at the spray outlet to obtain the instantaneous nebulization flow rate per second; a laser diffraction particle size detector is installed near the spray outlet to obtain the particle size distribution and main peak position; a triaxial MEMS accelerometer is attached to the nebulizer motor housing to collect the motor vibration frequency and amplitude in real time; a thermocouple temperature sensor is embedded in the middle of the drug flow pipeline to record the temperature change of the drug as it passes through; and a piezoresistive pressure sensor is connected in the middle of the connecting pipeline to collect real-time pressure data inside the pipeline.

[0064] After all acquired signals are sent to the edge processing unit via the data acquisition module, the following operations are performed: bandpass filtering is applied to various sensor signals to filter out interference signals below 0.5Hz and above 50Hz; the average value and rate of change are calculated for each parameter using the sliding window method; and each parameter is normalized using the standard normalization method to facilitate subsequent unified modeling.

[0065] For each type of parameter, representative numerical features are extracted, including: the average atomization flow rate over 5 seconds; the diameter of the main particle peak; the main frequency of motor vibration; the average heating rate of the liquid temperature; and the amplitude of pipeline pressure fluctuation, i.e., the difference between the maximum and minimum values.

[0066] The five types of feature parameters are arranged in a fixed order to form a multidimensional state vector of length 5. This vector is the device state feature set, which will be input into the state recognition model to determine whether the current device is operating normally.

[0067] This feature construction method is portable and real-time, supporting operational status modeling for different models of anesthesia nebulization devices. By fusing multi-dimensional operational parameters to construct a status feature set, a high-resolution input foundation is provided for subsequent model recognition, enhancing the accuracy and generalization ability of device anomaly detection.

[0068] Furthermore, the step of constructing the EEG state vector includes:

[0069] Real-time brainwave signals are acquired using EEG electrodes;

[0070] The EEG signal is filtered and artifact removal is performed to extract the power spectral density values ​​of the alpha and beta wave frequency bands.

[0071] The sedation index is calculated based on the ratio of alpha wave power to beta wave power, and the sedation index is used to reflect the depth of anesthesia in patients.

[0072] The alpha wave power, beta wave power, and EEG sedation index are used to construct an EEG state vector to evaluate the compatibility between nebulization efficacy and patient neurological responses.

[0073] Specifically, in this embodiment, a wireless multi-channel EEG acquisition device is used, configured with 4 input channels, a sampling frequency set to 256Hz, and a quantization precision of 16 bits. Disposable silver chloride gel electrodes are used, and the electrode impedance must be controlled within 5kΩ to reduce noise. The electrode layout follows the international 10-20 system standard, selecting four channels: Fp1, Fp2, T3, and T4. The signal reference electrode is placed on both earlobes, and the grounding electrode is located at the midpoint of the forehead. The acquisition time window is set to update every 10 seconds.

[0074] The acquired raw EEG signals are preprocessed using the following steps:

[0075] A third-order Butterworth bandpass filter was used to truncate the original signal in the frequency domain from 0.5 to 40 Hz, filtering out power line interference and high-frequency electromyographic noise. Independent component analysis (ICA) was used to perform blind source separation on each EEG channel, extracting independent components. By calculating the Pearson correlation coefficient between each component and the EEG channel, highly correlated artifact components (typically with a correlation coefficient greater than 0.8 and a peak voltage exceeding ±100 μV) were identified, nulled, and then the signal was reconstructed, retaining effective brain source components. The weighted average of the Fp1 and T3 signals was used as the left brain signal channel, and the weighted average of the Fp2 and T4 signals was used as the right brain signal channel, avoiding the influence of local biases on subsequent feature extraction.

[0076] Fast Fourier Transform (FFT) was used to process the purified EEG signal, and the power spectral density of the following frequency bands was extracted: alpha band and beta band. The analysis window was sliced ​​using a window function, and the power spectral density of each frequency band was calculated using the integral energy method, with units of μV² / Hz.

[0077] The brainwave sedation index is derived by calculating the β / α ratio (i.e., the ratio of β wave power to α wave power).

[0078] ;

[0079] Let be the EEG sedation index obtained through spectral analysis at time t. The larger the number, the higher the alertness and the lighter the anesthesia. The smaller the value, the deeper the sedation and the more complete the anesthesia. Therefore, this ratio is used as the sedation index. Used for assessing the depth of anesthesia.

[0080] The above calculation results are concatenated into a three-dimensional vector, which serves as the representation of the current EEG state:

[0081] ;

[0082] This vector serves as an important input parameter for subsequent physiological adaptability analysis and anesthetic efficacy feedback of the system.

[0083] By parameterizing the characteristics of EEG frequency bands, a standardized form of neural state expression is constructed, enabling physiological feedback to form a comparable computational basis with device output, supporting individualized adaptation analysis.

[0084] S200: Input the device status feature set into the status recognition model to identify the current operating status of the device. When an abnormal status is identified, generate a prompt signal and upload the device status information.

[0085] Furthermore, the state recognition model is a support vector machine classification model built based on a support vector set, and its specific implementation includes:

[0086] Collect historical operating parameters of the equipment under various operating conditions, construct a training sample set, and standardize the equipment state feature set in the sample, and attach corresponding operating state labels;

[0087] A set of support vectors for constructing support boundaries is selected from the training sample set, and the support vector machine classification model is trained.

[0088] The currently collected equipment status feature set is input into the support vector machine classification model, and the current operating status label of the equipment is output to determine whether there are conditions such as uneven atomization, abnormal particles, depletion of liquid, pipeline blockage, or dry burning.

[0089] Specifically, firstly, during the equipment testing phase, multiple sets of data samples representing typical operating states were obtained by manually setting the equipment's operating conditions under different states. The set operating states included: normal atomization state, uneven atomization state, liquid depletion state, pipeline blockage state, and dry burning state.

[0090] In each state, real-time operating parameters are acquired through various sensors installed within the equipment, including: atomization flow rate, particle size, motor vibration frequency, liquid temperature, and pipeline pressure. The acquired parameter data is processed to extract statistical features that characterize state changes, including the magnitude of flow rate changes, particle size stability, motor frequency fluctuation range, temperature gradient changes, and pressure curve fluctuation trends.

[0091] Subsequently, the extracted feature parameters are combined in a uniform order to form state feature vectors, and a corresponding running state label is added to each feature vector to form a training sample set. To avoid the influence of differences in the scales of different parameters on the model, the feature parameters in the samples are first standardized to ensure that they have similar data scales.

[0092] After constructing the training samples, the labeled state information in the training sample set is used to model the samples using the support vector classification principle. By selecting several of the most representative feature vectors from the samples as support boundary points, a support vector machine model for state classification is established. After the model is trained, it has the ability to determine the current device state based on the input feature vectors.

[0093] During actual operation, when the device is working continuously, it collects various state parameters of the current operating cycle in real time through sensors, and extracts the state features of the current cycle in accordance with the feature extraction method during training, and combines them into a new input vector.

[0094] The input vector is fed into a pre-trained support vector machine classification model. Based on the feature value distribution, the model outputs a label indicating the current operating status of the device. If the output label corresponds to an abnormal state, including normal atomization, uneven atomization, depleted medication, blocked tubing, or dry burning, the system immediately generates a warning signal to notify the operator or activate the device's self-protection logic. Simultaneously, this status information, along with the original parameters, is uploaded to a remote monitoring platform for intraoperative recording and postoperative analysis.

[0095] By using support vector machines to classify and identify device status features, intelligent judgment of various atomization anomalies can be achieved, which has stronger real-time performance and adaptability compared with empirical methods.

[0096] S300: When identified as a normal state, calculates atomization-related characteristic indicators based on continuous operation data, and substitutes them into the atomization efficiency decay equation to predict the future trend of atomization efficiency.

[0097] Furthermore, the step of predicting the future trend of atomization efficiency includes:

[0098] The vibration frequency, gas pressure and liquid temperature signals collected over a continuous time period were smoothed using the exponential weighted moving average method to obtain the weighted average of each indicator.

[0099] Substitute the weighted average value into the atomization efficiency decay equation to calculate the atomization efficiency estimate for multiple time points;

[0100] Based on the estimated values, a trend curve of atomization efficiency changes over a future time period is constructed to determine whether there is a risk of decreased device efficiency.

[0101] The formula for calculating the atomization efficiency attenuation equation is as follows:

[0102] ;

[0103] in, For initial performance; These are the smoothed motor vibration frequency index, gas pressure index, and liquid temperature index, respectively. These are the attenuation coefficients for each parameter.

[0104] Specifically, during continuous operation of the equipment, the system collects data including motor vibration frequency according to a preset sampling period. Pipeline pressure and the temperature of the medicine solution Key operating parameters, including those mentioned above, are included. To enhance the robustness of the data, the system performs an exponentially weighted moving average processing on the original time series data, as shown in the following formula:

[0105] ;

[0106] in, This represents the original sampled value of a certain operating parameter at the current moment; This represents the weighted average value at the current moment; As a smoothing factor, historical nebulization flow data under different patients and working conditions were collected, and a candidate value range (e.g., 0.1 to 0.9) was set. For each candidate value, the mean absolute error between it and the actual efficacy curve was calculated, and the value corresponding to the smallest error was selected. As the final value used.

[0107] Subsequently, the system uses the three characteristic indicators obtained from the above smoothing process: vibration frequency ,pressure With temperature Substitute the values ​​into the atomization efficiency decay equation to calculate the atomization efficiency value at the current moment. The specific formula is as follows:

[0108] ;

[0109] in, This represents the theoretical maximum atomization efficiency under the initial optimal state of the device; These parameters are used to characterize the influence of motor vibration frequency, drug solution temperature, and pipeline pressure on nebulization efficiency. They are obtained by fitting historical operating data and combined with multiple clinical samples. Nonlinear regression analysis is used to determine the intensity of the influence of each operating parameter on nebulization efficiency, ensuring that the parameters are practically representative and repeatable.

[0110] The system continuously calculates multiple time points within a sliding time window. The system uses the value to construct a trend curve for atomization efficiency. A trend curve of atomization efficiency is constructed with values ​​on the vertical axis and time on the horizontal axis. The slope of this curve at several future time points is then fitted and analyzed. If the trend curve shows a continuous downward trend within a set time period, or if the estimated efficiency value corresponding to its endpoint is lower than a preset operational safety threshold obtained from historical efficiency data under normal operating conditions, an early warning mechanism is triggered, marking it as a potential efficiency decline risk state. Each estimated point on the trend curve represents the predicted atomization efficiency value at the corresponding time, serving as input for subsequent adaptation analysis.

[0111] By constructing a time-series continuous atomization efficiency trend model, the system can dynamically perceive the process of equipment performance changes, thereby shifting from passive detection to active prediction and significantly improving the forward-looking control capability of equipment operation stability.

[0112] S400: Input the nebulization efficiency and EEG state vector into the dynamic balance model of anesthesia depth index, and calculate the dynamic error between nebulization output and neural response;

[0113] Furthermore, the implementation steps of the dynamic balance model of the anesthesia depth index include:

[0114] Within a set time window, the predicted efficacy values ​​at each moment in the nebulization efficacy trend curve are paired with the corresponding EEG state vectors for that time period.

[0115] Extracting the brainwave sedation index from the brainwave state vector Based on the current nebulization efficiency, target nebulization efficiency, EEG sedation index, and target sedation value, the following dynamic equilibrium model is constructed:

[0116] ;

[0117] in, This is a performance prediction value; Target atomization efficiency; The preset target sedation value; and This is an empirical adjustment coefficient used to balance the relative influence of device output and neural response;

[0118] Through calculation The value of is used to obtain the dynamic error rate of change between anesthetic efficacy and neural response, and to determine whether it is within the set adaptation range.

[0119] Specifically, within a set time window (e.g., 30 seconds), the system extracts the predicted value of continuous atomization efficiency. , which represents the predicted nebulization efficacy value at time t. Simultaneously, the corresponding EEG sedation index is extracted from the EEG state vector during this time period. ,in The EEG sedation index is obtained from the spectral analysis at time t, and is derived through the steps in S100.

[0120] Let the target atomization efficiency be Target sedation value The formula for calculating the rate of change of dynamic error is:

[0121] ;

[0122] in, This indicates the dynamic error trend between anesthetic output and neural response, reflecting the immediate impact of increased or decreased nebulization efficacy on neural responses. The empirical adjustment coefficient was determined using a regression optimization method. Specifically, it included: selecting 50 cases of actual anesthesia nebulization equipment operation data and EEG sedation response data continuously recorded during surgery, covering two age groups: adult and elderly patients; constructing a sample set using nebulization flow rate, particle mean, motor vibration, and EEG sedation index; and fitting the sample set using gradient descent method with the objective function of minimizing the variance of the dynamic error sequence.

[0123] The system provides data for consecutive time intervals. The calculation results are monitored. When the error rate of change continues to exceed the set adaptation threshold range (e.g., ±0.2) for a certain duration (e.g., 20 seconds), the system determines that there is an adaptation abnormality in the current anesthesia state, indicating a possible risk of sedation that is too shallow or too deep.

[0124] By combining the trend of nebulization efficacy with the changes in EEG response, dynamic error analysis can be performed to achieve collaborative assessment of human-machine status and improve the stability and individual adaptability of the anesthesia process.

[0125] S500: When the efficiency decay rate exceeds the set threshold, or the dynamic error continues to exceed the target range, a prompt signal is generated, and the nebulization prediction results, EEG state vector and error information are uploaded to the remote monitoring platform to assist in intraoperative judgment and postoperative evaluation.

[0126] Furthermore, the step of generating a prompt signal and uploading information includes:

[0127] The first-order difference of the predicted nebulization efficacy over a continuous time period is calculated to obtain the efficacy decay rate, which is then compared with a preset threshold. It is also determined whether the dynamic error change rate continuously exceeds the anesthesia depth adaptation range.

[0128] When any judgment condition is met, an abnormal prompt signal is generated. The prompt types include "performance degradation abnormality", "adaptation imbalance" or "combined abnormality".

[0129] Construct a data package containing the current atomization efficiency trend prediction results, EEG state vectors, dynamic error values, and prompt types;

[0130] The data packet is uploaded to the remote monitoring platform via a wireless communication module, along with a timestamp and device ID information, for use in intraoperative status early warning and postoperative retrospective analysis.

[0131] Specifically, the system constructs a nebulization efficacy decay trend curve using an exponentially weighted moving average method. Within each sampling period, it calculates the change between two adjacent predicted values, obtaining a first-order difference sequence to characterize the rate of change in nebulization efficacy. The mean of the first-order differences within a fixed time window (e.g., 30 seconds) is set as the nebulization efficacy decay rate and compared with a preset critical threshold (determined by device type, dosage form, and clinical data experience). When the decay rate exceeds this threshold, it indicates a significant risk of device performance degradation.

[0132] Simultaneously, based on the dynamic error output by the dynamic equilibrium model, the system records the rate of change of dynamic error over multiple consecutive time periods. If the error continuously deviates from the preset sedation adaptation range, it is judged as an abnormal neural adaptation.

[0133] Next, the system will combine and analyze the two judgment processes mentioned above, and generate prompt information according to the following logic: if only the atomization efficiency decay rate exceeds the threshold, a "efficiency decay abnormal" prompt will be generated; if only the dynamic error continuously exceeds the target range, a "fitting imbalance" prompt will be generated; if both are true, a "joint abnormal" prompt will be generated.

[0134] To facilitate doctors' real-time monitoring of equipment operation status, intraoperative response risks, and postoperative analysis and review, the system constructs the following data into a set of monitoring data packages: the nebulization efficiency trend curve within the current prediction period; the concurrent EEG state vector; the dynamic error value and its changing trend; the abnormality type label and judgment result; and the current system timestamp and equipment identification information.

[0135] The data packet is uploaded to the remote monitoring platform. The platform can receive upload information from each terminal device in real time and trigger a local alarm interface to remind clinicians to respond promptly. Simultaneously, the uploaded data is archived for postoperative status retrospective analysis, equipment maintenance records, and safety reviews of the anesthesia process. By constructing a linkage mechanism between prompt generation and remote upload, the system achieves a synchronous closed loop of local judgment and remote response, ensuring timely communication of abnormal states and the formation of traceable records, thus enhancing the real-time performance and safety of clinical applications.

[0136] Example 2:

[0137] In the thoracic surgery operating room of a hospital, anesthesia maintenance was performed on a 65-year-old male patient undergoing thoracoscopic lobectomy using the ultrasonic nebulization module integrated into a Dräger Perseus A500 anesthesia workstation. The intelligent anesthesia nebulization device status monitoring system of this invention was also employed. Figure 2 ,include:

[0138] The parameter acquisition module collects the operating parameters of the anesthesia nebulization device, including nebulization flow rate, particle size distribution, motor vibration frequency, drug solution temperature and pipeline pressure, constructs a device state feature set, and simultaneously collects the patient's electroencephalogram (EEG) signals, extracts spectral features reflecting the sedation state, and constructs an EEG state vector.

[0139] The status recognition module is used to input the device status feature set into the status recognition model, identify the current operating status of the device, and generate a prompt signal and upload the device status information when an abnormal status is identified.

[0140] The performance prediction module, when identified as being in a normal state, calculates atomization-related characteristic indicators based on continuous operating data and substitutes them into the atomization performance decay equation to predict the future trend of atomization performance.

[0141] The error assessment module is used to input the nebulization efficiency and EEG state vector into the dynamic balance model of the anesthesia depth index to calculate the dynamic error between nebulization output and neural response.

[0142] The abnormality alert and upload module generates an alert signal when the efficiency decay rate exceeds a set threshold or the dynamic error continues to exceed the target range. It also uploads the nebulization prediction results, EEG state vector, and error information to the remote monitoring platform to assist in intraoperative judgment and postoperative evaluation.

[0143] Specifically, data was collected in real time 25 minutes after the start of the surgery using the following sensors in the parameter acquisition module:

[0144] Atomized flow rate: Instantaneous flow rate is collected using a thermal mass flow meter, and after processing by Butterworth low-pass filter, the mean and rate of change within a 5-second sliding window are calculated to obtain the current flow rate value of 0.82 L / min;

[0145] Particle size distribution: Particle groups were detected by laser diffraction, and the main peak diameter was extracted to be 3.2 μm;

[0146] Motor vibration: Vibration signals were collected by a triaxial accelerometer, and the energy integral value of 0.15g² / Hz in the 20-200Hz frequency band was extracted by fast Fourier transform;

[0147] Liquid temperature: The platinum resistance temperature sensor records the temperature gradient of the liquid as it flows through the pipeline, and the current value is 24.3℃;

[0148] Pipeline pressure: Piezoresistive sensors monitor gas pressure fluctuations, with a maximum fluctuation amplitude of 12 mbar within a 10-second window.

[0149] After standard processing of the above data, the atomized flow rate, vibration energy, temperature gradient, pipeline pressure, and peak particle size are -0.25, 0.75, -1.0, 0.4, and 3.2, respectively. These five characteristic parameters are arranged in a fixed order to form a multidimensional state vector of length 5. This vector is the equipment state feature set: [-0.25, 3.2, 0.75, -1.0, 0.4].

[0150] The patient's EEG signals were acquired synchronously using a four-channel EEG acquisition system (with Fp1, Fp2, T3, and T4 electrodes deployed in a 10-20 system). After bandpass filtering and independent component analysis to remove oculomotor artifacts from the 0.5-40Hz band, the power spectral density of the alpha band was calculated to be 12.4 μV² / Hz, and the power spectral density of the beta band was calculated to be 4.7 μV² / Hz, thus yielding a sedation index B(t) of 0.38. The time synchronization error between the equipment parameters and the EEG signals was controlled within ±50ms.

[0151] The aforementioned 5-dimensional equipment state feature set is input into a pre-trained support vector machine classification model. This model is generated based on 200 sets of historical samples and uses a radial basis function kernel (penalty factor C=1.0). The model output shows: normal state probability 91%, pipeline blockage probability 5%, and dry burning probability 2%. The current equipment is determined to be in normal operating condition, and the performance prediction process begins.

[0152] The vibration frequency, pipeline pressure, and liquid temperature data over a continuous 30-second period were smoothed using an exponentially weighted moving average method (smoothing factor λ=0.2), yielding weighted averages: vibration frequency 0.73, pressure 0.38, and temperature -0.98. These values ​​were then substituted into the atomization efficiency attenuation equation:

[0153] ;

[0154] Among them, the initial efficiency = 1.0, and the attenuation coefficient , , (determined by regression fitting of 100 groups of historical data).

[0155] The current efficiency value is calculated as . Based on this equation, further prediction is made: the efficiency drops to 0.83 after 5 minutes, to 0.80 after 10 minutes, and to 0.77 after 15 minutes. The predicted efficiency value and the electroencephalogram sedation index are input into the dynamic balance model:

[0156] ;

[0157] Among them, the target atomization efficiency , the target sedation index , and the adjustment coefficient , (determined by verification based on 50 cases of clinical data).

[0158] The dynamic error change rate is calculated as . Continuously monitor within three calculation cycles (15 seconds) and if it is less than -0.02, it is determined as "imbalance between anesthesia depth and device output adaptation".

[0159] The system triggers the following linkage responses:

[0160] Local alarm: Display a red-yellow dual-color alarm logo on the screen of the anesthesia workstation. The red area is marked with "Continuous attenuation of atomization efficiency (current 0.85)", and the yellow area is marked with "Too shallow anesthesia depth (B(t)=0.38)";

[0161] Remote push: Send a structured data packet to the remote monitoring platform through the communication module, including:

[0162] Atomization efficiency prediction curve (0.85 → 0.83 → 0.80 → 0.77);

[0163] Electroencephalogram state vector [12.4, 4.7, 0.38];

[0164] Dynamic error value = -0.024;

[0165] Device ID "Perseus-A500-034" and timestamp;

[0166] Decision Recommendation: The platform automatically generates the following operation prompt: "Recommendation: Check the wear of the atomizing components and increase the propofol infusion rate by 10%."

[0167] In this embodiment, the system simultaneously acquires device parameters and EEG signals in a dual-modal manner. When the device's mechanical performance is normal, it predicts 15 minutes in advance that the nebulization efficiency will fall below the safe threshold (0.80), and simultaneously identifies the hidden risk of insufficient anesthesia depth in the patient. After the clinical team adjusts the anesthesia plan accordingly, the patient's sedation index recovers to the target range (0.35-0.40) within 8 minutes, effectively avoiding the risk of intraoperative awareness. Compared to traditional single-parameter monitoring methods, this approach shortens the abnormal response time to 8 seconds and reduces the number of intraoperative interventions.

[0168] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent anesthesia nebulization equipment status monitoring system, characterized in that, include: The parameter acquisition module is used to acquire the operating parameters of the anesthesia nebulization device, including nebulization flow rate, particle size distribution, motor vibration frequency, drug solution temperature and pipeline pressure, to construct a device state feature set, and simultaneously acquire the patient's electroencephalogram (EEG) signal, extract spectral features reflecting the sedation state, and construct an EEG state vector. The status recognition module is used to input the device status feature set into the status recognition model, identify the current operating status of the device, and generate a prompt signal and upload the device status information when an abnormal status is identified. The performance prediction module is used to calculate atomization-related characteristic indicators based on continuous operation data when the normal state is identified, and substitute them into the atomization performance decay equation to predict the future trend of atomization performance. The error assessment module is used to input the nebulization efficiency and EEG state vector into the dynamic balance model of the anesthesia depth index to calculate the dynamic error between nebulization output and neural response. The abnormality alert and upload module is used to generate an alert signal when the efficiency decay rate exceeds a set threshold or the dynamic error continues to exceed the target range. It also uploads the nebulization prediction results, EEG state vector and error information to the remote monitoring platform to assist in intraoperative judgment and postoperative evaluation.

2. The intelligent anesthesia nebulization equipment status monitoring system according to claim 1, characterized in that, The steps for constructing the device state feature set include: Multiple sets of sensors installed inside the atomizing device and at the gas path nodes collect atomization flow sensing signals, particle size detection signals, motor acceleration sensing signals, liquid temperature signals, and gas path pressure signals, respectively. Each signal is filtered, denoised, and standardized to extract statistical characteristic parameters that characterize the equipment's operating status, including mean flow rate, peak particle size, vibration spectrum energy, temperature gradient, and pressure fluctuation amplitude. The above feature parameters are combined into a multi-dimensional state vector, which is then used as the device state feature set input to the subsequent state recognition model.

3. The intelligent anesthesia nebulization equipment status monitoring system according to claim 1, characterized in that, The steps for constructing the EEG state vector include: Real-time brainwave signals are acquired using EEG electrodes; The EEG signal is filtered and artifact removal is performed to extract the power spectral density values ​​of the alpha and beta wave frequency bands. The sedation index is calculated based on the ratio of alpha wave power to beta wave power, and the sedation index is used to reflect the depth of anesthesia in patients. The alpha wave power, beta wave power, and EEG sedation index are used to construct an EEG state vector to evaluate the compatibility between nebulization efficacy and patient neurological responses.

4. The intelligent anesthesia nebulization equipment status monitoring system according to claim 1, characterized in that, The state recognition model is a support vector machine classification model built based on a set of support vectors, and its specific implementation includes: Collect historical operating parameters of the equipment under various operating conditions, construct a training sample set, and standardize the equipment state feature set in the sample, and attach corresponding operating state labels; A set of support vectors for constructing support boundaries is selected from the training sample set, and the support vector machine classification model is trained. The currently collected equipment status feature set is input into the support vector machine classification model, and the current operating status label of the equipment is output to determine whether there are conditions such as uneven atomization, abnormal particles, depletion of liquid, pipeline blockage, or dry burning.

5. The intelligent anesthesia nebulization equipment status monitoring system according to claim 1, characterized in that, The steps for predicting future trends in atomization efficiency include: The vibration frequency, gas pressure and liquid temperature signals collected over a continuous time period were smoothed using the exponential weighted moving average method to obtain the weighted average of each indicator. Substitute the weighted average value into the atomization efficiency decay equation to calculate the atomization efficiency estimate for multiple time points; Based on the estimated values, a trend curve of atomization efficiency changes over a future time period is constructed to determine whether there is a risk of decreased device efficiency. The formula for calculating the atomization efficiency attenuation equation is as follows: ； in, For initial performance; These are the smoothed motor vibration frequency index, gas pressure index, and liquid temperature index, respectively. These are the attenuation coefficients for each parameter.

6. The intelligent anesthesia nebulization equipment status monitoring system according to claim 1, characterized in that, The implementation steps of the dynamic balance model of the anesthesia depth index include: Within a set time window, the predicted efficacy values ​​at each moment in the nebulization efficacy trend curve are paired with the corresponding EEG state vectors for that time period. Extracting the brainwave sedation index from the brainwave state vector Based on the current nebulization efficiency, target nebulization efficiency, EEG sedation index, and target sedation value, the following dynamic equilibrium model is constructed: ； in, This is a performance prediction value; Target atomization efficiency; The preset target sedation value; and This is an empirical adjustment coefficient used to balance the relative influence of device output and neural response; Through calculation The value of is used to obtain the dynamic error rate of change between anesthetic efficacy and neural response, and to determine whether it is within the set adaptation range.

7. The intelligent anesthesia nebulization equipment status monitoring system according to claim 6, characterized in that, The steps of generating a prompt signal and uploading information include: The first-order difference of the nebulization efficacy prediction value within a continuous time period is calculated to obtain the efficacy decay rate, which is then compared with a preset threshold; and it is determined whether the dynamic error change rate continuously exceeds the anesthesia depth adaptation range. When any judgment condition is met, an abnormal prompt signal is generated. The prompt type includes "performance degradation abnormality", "adaptation imbalance" or "combined abnormality". Construct a data package containing the current atomization efficiency trend prediction results, EEG state vectors, dynamic error values, and prompt types; The data packet is uploaded to the remote monitoring platform via the wireless communication module, along with a timestamp and device ID information.

Images