Intelligent early warning system for airway management of otolaryngology department patient

Abstract

The invention discloses an intelligent early warning system for airway management of otolaryngological patients, which relates to the technical field of airway management and is configured to execute the following steps: acquiring physiological data such as upper airway breathing audio signals representing airflow and airway tissue states and disturbance data such as sleep body positions; the method comprises the following steps: generating a first risk index representing an airflow turbulence effect and a second risk index representing a tissue support capability, performing individual adaptive calibration parameter fusion calculation to obtain an airway stability margin, generating a basic risk index in combination with disturbance data, and performing dynamic covariant correction factor correction to obtain a final risk index. And an early warning signal is output after comparison with a preset threshold value. Closed-loop management is formed through hierarchical intervention, the early warning accuracy and intervention effectiveness are improved, and the problems that a traditional method lacks predictability, is insufficient in individuation, lags in intervention and the like are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of airway management, and in particular to an intelligent early warning system for airway management of otolaryngology patients. Background Art

[0002] In the field of airway management, instability risk assessment involves complex physiological mechanisms, requiring the integration of multimodal physiological signals such as respiratory audio, bioimpedance, and sleep position. Airway collapse is a nonlinear mechanical event caused by fluid-structure interaction. The mapping between external physiological signals and the internal tissue mechanical state is highly nonlinear, requiring precise quantification of airflow dynamics and tissue biomechanical characteristics to predict risk.

[0003] Traditional risk assessment methods often diagnose respiratory events, such as apnea, only after they occur, lacking predictive capabilities. Furthermore, they struggle to effectively integrate different modal signals, such as acoustic and bioelectrical, without disrupting the patient's sleep, and to decouple changes in airflow dynamics and upper airway structural mechanical properties in real time. Consequently, they are unable to accurately predict "functional" airway instability. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent early warning system for airway management of otolaryngology patients, which solves the problems existing in the background technology.

[0005] To solve the above technical problems, the present invention provides an intelligent early warning system for airway management of otolaryngology patients, which is configured to perform the following steps: Acquiring first physiological data representing an airflow state, second physiological data representing an airway tissue state, and disturbance data representing a sleep state of the individual; Processing the first physiological data to generate a first risk indicator for characterizing airflow turbulence effects , and processing the second physiological data to generate a second risk indicator for characterizing tissue support capacity ; Through individual adaptive calibration parameters , for the first risk indicator and the second risk indicator Perform fusion calculations to determine airway stability margins , and combined with the airway stability margin and the disturbance data to generate the basic risk index ; According to the first risk indicator With the second risk indicator The amount of change within a preset time window determines the dynamic covariance correction factor used to correct the contradictory risk indicators ; By With the dynamic covariance correction factor Perform combination operations to generate the final risk index ; The final risk index The final risk index is compared with a preset risk threshold, and an early warning signal is output when the final risk index meets the preset conditions.

[0006] Preferably, the first physiological data is set to an upper airway breathing audio signal; The second physiological data is set to be a multi-frequency bioimpedance spectrum of parapharyngeal tissue; The disturbance data is set to include a set of sleeping position data and sleep stage data.

[0007] Preferably, the step of processing the first physiological data and the second physiological data further includes: Performing a short-time Fourier transform on the first physiological data to calculate spectral entropy and high-order moments, and setting the calculation results of the spectral entropy and high-order moments as the first risk index ; A Cole-Cole model is established based on the second physiological data to solve the equivalent tissue impedance modulus, and the equivalent tissue impedance modulus is set as the second risk indicator. .

[0008] Preferably, the individual adaptive calibration parameters The determination process includes: Perform the initial calibration step to set the basic physiological information input by the user when using it for the first time. Initial baseline value of Perform continuous adaptive correction steps to analyze the correlation between the model prediction results and the actual effects of the intervention measures. Perform closed-loop feedback fine-tuning.

[0009] Preferably, the basic risk index The generation of is further restricted to the following computation: ; and For specific users The individual adaptive sensitivity weights representing the effects of body position and sleep stage respectively; is a body position factor extracted from the sleeping body position data; is a staging factor extracted from the sleep staging data.

[0010] Preferably, the dynamic covariance correction factor The determination of is further defined as the following calculation: ; and is the rate of change of the first risk indicator and the second risk indicator calculated within a preset time window; is a preset covariance correction sensitivity coefficient used to adjust the model's penalty intensity for contradictory signals. The value of is determined based on offline optimization of historical clinical datasets.

[0011] Preferably, the combination operation is further defined as: With the dynamic covariance correction factor Multiply.

[0012] Preferably, the individual adaptive sensitivity weight and The determination process includes: Assigning a universal benchmark value to the weight as an initial calibration based on large-scale clinical statistical data; In long-term use, the weights are individually and continuously fine-tuned automatically by analyzing the correlation between interventions and actual changes in risk indicators.

[0013] Preferably, the preset risk threshold includes: The first level risk threshold is used to identify the risk state of significantly decreased airway stability; The second level risk threshold, whose value is set to be higher than the first level risk threshold, is used to identify a critical risk state of impending apnea.

[0014] Preferably, the output of the early warning signal is configured to trigger a hierarchical intervention strategy, the strategy comprising: When the final risk index When the first level risk threshold is exceeded, the flexible intervention module is activated to guide the user to unconsciously change their sleeping position; When the final risk index When the second-level risk threshold is exceeded, it is used to start the enhanced intervention module to promote the user to a lighter sleep stage or adjust the pressure of the continuous positive airway pressure device.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. By constructing a prediction model that integrates acoustic and bioimpedance dual-domain information, it is possible to capture precursor signals before airway instability occurs, transforming passive responses into advance predictive control.

[0016] 2. By introducing individual adaptive calibration parameters and dynamic covariance correction factors, combined with initial calibration and continuous adaptive correction, it can adapt to individual physiological differences, correct misjudgments of conflicting signals, and improve model personalization and prediction robustness.

[0017] 3. By establishing a graded intervention strategy, triggering flexible or enhanced intervention based on the risk index, forming a "monitoring-assessment-intervention-reassessment" closed loop, active airway management can be achieved, improving safety while avoiding excessive interference with sleep. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention, and those skilled in the art can derive other drawings based on these drawings without inventive effort. Figure 1 It is a logic block diagram of the system of the present invention. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0020] Example 1:

[0021] See also Figure 1 The present invention provides an intelligent early warning system for airway management of otolaryngology patients, characterized in that the system is configured to perform the following steps: obtaining first physiological data representing airflow status, second physiological data representing airway tissue status, and disturbance data representing individual sleep status; processing the first physiological data to generate a first risk index for representing airflow turbulence effects , and processing the second physiological data to generate a second risk indicator for characterizing tissue support capacity ; Through individual adaptive calibration parameters , for the first risk indicator and the second risk indicator Perform fusion calculations to determine airway stability margins , and combined with the airway stability margin and disturbance data to generate the basic risk index ; According to the first risk indicator and the second risk indicator The amount of change within a preset time window determines the dynamic covariance correction factor used to correct the contradictory risk indicators ; By the basic risk index Dynamic covariance correction factor Perform combination operations to generate the final risk index ; The final risk index Compare with the preset risk threshold and output a warning signal when the final risk index meets the preset conditions; The intelligent early warning system for airway management of otolaryngology patients in this embodiment realizes early warning of the risk of airway instability of patients through its configured specific execution steps. The execution steps of the system begin with the acquisition of three types of core data: first physiological data representing the state of airflow, second physiological data representing the state of airway tissue, and disturbance data representing the sleep state of the individual. The system uses these input data to generate a final risk index through a series of calculations. The core of this process is that the system not only performs simple status monitoring, but also processes the first physiological data and the second physiological data and combines them with individual adaptive calibration parameters. , perturbation data, and dynamic covariance correction factors , quantitatively modeling the intrinsic physical process of airway collapse; ultimately, the final risk index It can compare with the preset risk threshold in real time. Once the trigger conditions are met, the system will output an early warning signal. The purpose of designing this early warning signal is to provide a time window for taking intervention measures before a respiratory event (such as asphyxia) is fully formed, thereby upgrading traditional diagnostic monitoring to predictive, interventional active airway management.

[0022] Example 2: The first physiological data is set as an upper airway breathing audio signal; the second physiological data is set as a parapharyngeal tissue multi-frequency bioimpedance spectrum; the disturbance data is set as a set including sleeping posture data and sleep stage data; The step of processing the first physiological data and the second physiological data further includes: performing a short-time Fourier transform on the first physiological data to calculate spectral entropy and high-order moments, and setting the calculation results of the spectral entropy and high-order moments as the first risk indicator ; Establish a Cole-Cole model based on the second physiological data to solve the equivalent tissue impedance modulus, and set the equivalent tissue impedance modulus as the second risk indicator ; Among them, the Cole-Cole model is established: the collected multi-frequency bioimpedance spectrum data points Plotted on the complex plane. Using nonlinear least squares, these data points are fitted to the Cole-Cole equation , the resistance is solved through this fitting process 、 , time constant and the distribution coefficient , and finally, the equivalent tissue impedance modulus is set to the difference between the low-frequency resistance and the high-frequency resistance, i.e. ; : at the i-th angular frequency Below, the real part of the measured bioimpedance; : at the i-th angular frequency Below, the imaginary part of the measured bioimpedance; : The i-th angular frequency point for impedance spectrum measurement; : As the angular frequency The complex impedance of the function is given by ; : Resistance value when frequency tends to infinity; : Resistance value when the frequency approaches zero; : Imaginary unit, defined as ; : angular frequency; : The time constant in the Cole-Cole model, related to the characteristic frequency; : the distribution coefficient in the Cole-Cole model, ranging from 0 to 1, characterizing the breadth of the relaxation time distribution; : Equivalent tissue impedance modulus, defined as the second risk indicator in the present invention; The short-time Fourier transform (STFT) was configured to use a Hamming window with a window length of 256 samples and a 50% overlap between windows. After each frame of signal was transformed, its power spectral density was used for subsequent spectral entropy and high-order moment calculations.

[0023] The intelligent early warning system for airway management of otolaryngology patients in this embodiment has specific settings for data sources and processing steps to achieve accurate calculation of risk indicators. The first physiological data is defined as upper airway respiratory audio signals acquired via a clip-on acoustic sensor array; the second physiological data is defined as a multi-frequency bioimpedance spectrum of parapharyngeal tissue acquired via a flexible neck electrode patch; and the disturbance data is defined as a combination of sleep position data and sleep stage data acquired via a body position sensor and a portable EEG module. In the step of processing physiological data, the system performs short-time Fourier transform (STFT) on the respiratory audio signal as the first physiological data to analyze its time-frequency spectrum. To quantify the signal uncertainty and energy distribution changes caused by the intensification of airflow turbulence, the system further calculates two key feature quantities: spectral entropy ( ), which is used to measure the uniformity of spectral energy distribution; and high-order moments (in terms of spectral kurtosis For example, it is used to measure the concentration of spectrum energy. In order to integrate the characteristics of these two different dimensions into a single risk indicator, the system performs a linear weighted summation using preset weights. The calculation is limited to: ; is the first risk indicator at time t; and are the dimensionless values ​​of the spectral entropy and spectral kurtosis at time t after normalization (e.g., maximum and minimum normalization) to eliminate the influence of different feature dimensions; and The weight coefficient is preset and optimized by analyzing historical clinical data sets, which is used to balance the relative importance of spectral entropy and kurtosis in assessing airflow turbulence risk. and As input features, the binary result of whether airway collapse occurs later is used as a label to build a logistic regression model. and This is the best fitting coefficient obtained after the model training, and it also satisfies Normalization constraints.

[0024] At the same time, based on the second physiological data (multi-frequency bioimpedance spectrum), the system establishes a Cole-Cole model for fitting analysis to solve an equivalent tissue impedance modulus; this modulus reflects the biomechanical properties of the pharyngeal wall tissue, such as water content and muscle tension, and is set as the second risk indicator to characterize the tissue support capacity. .

[0025] Example 3: Individual adaptive calibration parameters The determination process includes: performing an initial calibration step for setting the basic physiological information input by the user when using it for the first time The initial baseline value of the model is used to perform continuous adaptive correction steps to analyze the association between the model prediction results and the actual effects of the intervention measures. Perform closed-loop feedback fine-tuning; Basic Risk Index The generation of is further restricted to the following computation: ; and are the individual adaptive sensitivity weights for a specific user u, representing the influence of body position and sleep stage respectively; is the body position factor extracted from the sleeping body position data; is the staging factor extracted from the sleep staging data; The combination operation is further limited to the basic risk index Dynamic covariance correction factor multiplication; Individual adaptive sensitivity weight and The determination process includes: assigning a universal benchmark value to the weight as an initial calibration based on large-scale clinical statistical data; in long-term use, by analyzing the correlation between intervention measures and actual changes in risk indicators, the weight is personalized and continuously automatically fine-tuned; the individual adaptive calibration parameters The initial baseline value of can be determined by a linear regression model, and the individual adaptive calibration parameters Initial baseline value Calculated using the following model: ; represents height (in meters), represents body weight (in kilograms), Represents neck circumference (in centimeters). It is an empirical constant pre-calibrated based on previous clinical trial data. The specific determination method is as follows: collect a basic data set containing at least 500 subjects, which contains the height of each subject. ,weight , neck circumference and the optimal sleep apnea score jointly determined by at least two sleep medicine experts based on their overnight polysomnography (PSG) data. The optimal value The calibration goal of the value is to make the final risk index obtained in the subsequent calculation The most accurate prediction of apnea events. As the independent variable, the expert-calibrated As the dependent variable, multiple linear regression analysis was used to solve the regression coefficient and intercept .

[0026] The pair The specific algorithm for closed-loop feedback fine-tuning is limited to an online learning algorithm based on gradient descent. During the continuous use of the system, when a clear respiratory event (such as apnea or hypopnea, which can be confirmed by signals such as decreased pulse oximetry) or a successful intervention (such as normal breathing sounds after a change in body position) is detected, the system will generate an error signal. The error signal is defined as: ,in is a target risk value (e.g., for respiratory events, the target value is a higher threshold; for normal breathing, the target value is a lower baseline value). The update rules are as follows:

[0027] in, is the preset learning rate, for example By using the chain rule, partial derivatives can be calculated, thus achieving Iterative updates to make it more adaptable to the physiological characteristics of individual users; represents the individual adaptive calibration parameter at time t The value of Represents the individual adaptive calibration parameters at time t+1 (i.e., the next update time) The value of .

[0028] In order to ensure the applicability of the model to different individuals and the accuracy of prediction, the system of this embodiment introduces individual adaptive calibration parameters and individual adaptive sensitivity weights and ; The determination process reflects the personalized learning ability of the model: in the initial calibration step, the system will use the basic physiological information such as height, weight, neck circumference, etc. input by the user for the first time to Set an initial baseline value; in subsequent use, the continuous adaptive correction step will be started, and the system will correlate the triggering of the model warning with the actual effect of the intervention measures (such as position change), and use the closed-loop feedback algorithm to Fine-tune it so that it evolves dynamically to match the uniqueness of the user; similarly, the weights and The determination process also follows the principle of "initial calibration + continuous optimization": based on large-scale clinical statistical data, the system will preset a universal benchmark value for these two weights; then, during long-term use, the system will continuously and automatically fine-tune the weights in a personalized manner by learning the user's sensitivity to sleeping posture and different sleep stages (such as REM sleep); Based on these personalized parameters, the system calculates the basic risk index A specific calculation model is used to generate the airway stability margin. is defined as:

[0029] is the airway stability margin at time t; Risk indicators for organizational support capabilities; is a risk indicator of airflow turbulence effects; Calibrate parameters for individual adaptation; Basic Risk Index The generation of is limited to the following calculation:

[0030] and are the individual adaptive sensitivity weights for a specific user u, representing the influence of body position and sleep stage respectively; u is the user; The body position factor is extracted from the sleeping position data, with the supine position defined as 0° and the fully side-lying position as 90°. This body position factor is obtained through a three-axis acceleration sensor. By calculating the projection of the gravity vector in the human torso coordinate system, the angle between the torso and the vertical direction can be calculated in real time. This angle is defined as The sensor deployment location is fixed on the sternum or neck to ensure consistent measurement.

[0031] The dimensionless staging factor extracted from the sleep staging data is quantified into different values ​​based on real-time EEG signal recognition. The specific definition of the quantification process is: The value was set to 0, non-rapid eye movement (N1) to 0.5, non-rapid eye movement (N2) to 0.8, non-rapid eye movement (N3) to 1.0, and rapid eye movement (REM) to 1.2 due to its high correlation with respiratory events. These values ​​reflect the differences in the degree of muscle relaxation and the risk of apnea in different sleep stages.

[0032] The combination operation is further limited to the basic risk index Dynamic covariance correction factor Multiply them together to get the final risk index;

[0033] The effect of this design is that by introducing and dynamically adjusting 、 and The system can evolve from a general model to a dedicated prediction model tailored for each user; the calculation formula of the basic risk index clearly integrates the core physiological stability ( ) and external disturbance factors (sleep posture and sleep stage), making risk assessment more comprehensive and dynamic.

[0034] Example 4: Dynamic covariance correction factor The determination of is further defined as the following calculation:

[0035] and The change rates of the first risk indicator and the second risk indicator calculated within the preset time window are normalized to become dimensionless values ​​before being substituted into the formula; is the preset covariance correction sensitivity coefficient used to adjust the model's penalty intensity for contradictory signals. The value of is determined based on offline optimization of historical clinical datasets; The offline optimization process is specifically implemented as follows: using a labeled dataset containing physiological signal changes before and after sleep breathing events of thousands of patients, a grid search combined with cross-validation is used. By systematically testing a series of candidate The area under the receiver operating characteristic curve (AUC) of each value on the validation set was evaluated, and the value that maximized the AUC was finally selected. value as the optimal parameter.

[0036] In order to improve the prediction robustness in special physiological scenarios, the system of this embodiment adjusts the dynamic covariance correction factor The determination of is limited to a specific mathematical operation; the purpose of this limitation is to solve a class of "contradictory signal" problems, that is, when the turbulence of the airflow is characterized by and characterize tissue support When atypical increases occur simultaneously (both increases indicate risk, but their ratio is may remain unchanged or change only slightly), conventional risk assessments may mask the true risk of collapse; Dynamic covariance correction factor The determination of is further defined as the following calculation:

[0037] is the dynamic covariance correction factor at time t; and The rate of change of the first risk indicator and the second risk indicator calculated within a very short preset time window; Covariation-corrected sensitivity coefficient It is a preset covariation-corrected sensitivity coefficient used to adjust the model's penalty strength for contradictory signals. This coefficient is a model hyperparameter whose value is determined by offline optimization based on historical clinical datasets through machine learning model tuning technology to achieve a balance between prediction accuracy and sensitivity. The effect of this embodiment is that when and When the product of is positive (i.e., the two change in the same direction, especially when they increase abnormally at the same time), The value of will be greater than 1, thus affecting the basic risk index Punitive amplification is carried out to reveal the real risks behind this "illusion of stability"; when the changing trends of the two are normal (reverse changes) or there is no significant change, The value of approaches 1 and has no impact on the basic risk index; this mechanism significantly enhances the prediction accuracy and stability of the system under the influence of short-term atypical physiological events such as micro-arousal and swallowing.

[0038] Example 5: The preset risk thresholds include: a first-level risk threshold for identifying a risk state of significantly decreased airway stability; a second-level risk threshold, which is set higher than the first-level risk threshold for identifying a critical risk state of impending apnea; The output of the early warning signal is configured to trigger a graded intervention strategy, which includes: When the first level risk threshold is exceeded, the flexible intervention module is used to guide the user to change the sleeping position unconsciously; when the final risk index When the second level risk threshold is exceeded, it is used to start the enhanced intervention module to promote the user to a lighter sleep stage or adjust the pressure of the continuous positive airway pressure device; After the system outputs the warning signal, in order to achieve closed-loop management, the system of this embodiment specifically defines the preset risk threshold and graded intervention strategy. The preset risk threshold is designed to be two levels: Level 1 risk threshold: This threshold is set to identify a risk state where airway stability has significantly decreased, corresponding to a precursor to a hypoventilation event; for example, when When , this threshold is met; Second level risk threshold: Its value is set higher than the first level risk threshold and is used to identify the critical risk state of impending apnea; for example, when The specific values ​​of these two thresholds are determined based on the receiver operating characteristic (ROC) curve analysis of historical clinical data to achieve the best early warning classification in clinical significance; Accordingly, the output of early warning signals is configured to trigger a graded intervention strategy that provides different responses depending on the level of risk: When the final risk index When the first-level risk threshold is exceeded, the system activates the flexible intervention module. This module uses a micro-vibration unit worn on the neck or chest to provide patients with sub-arousal-level somatosensory stimulation, guiding the user to unconsciously change sleeping positions (for example, from supine to side-lying), thereby naturally improving airway patency while avoiding deep interference with the user's sleep structure. When the final risk index When the second-level risk threshold is exceeded, the system will activate the enhanced intervention module; this module's intervention measures are more powerful and are specifically divided into two situations: for patients using continuous positive airway pressure devices, the system instructs the device through wireless signals to adjust the pressure of the continuous positive airway pressure device, instantly increasing it by 1-2cmH2O to provide pneumatic support; for patients not using CPAP, the system uses acoustic or stronger vibration stimulation to raise the user to a lighter sleep stage to restore some pharyngeal muscle tone, thereby avoiding the occurrence of apnea; In this way, the graded intervention strategy achieves a shift from "passive monitoring" to "active intervention"; this strategy conducts precise and appropriate intervention based on the severity of the risk, which can not only effectively resolve sleep apnea events before they occur, but also maximize the user's sleep quality, forming a complete closed loop of "real-time monitoring-risk assessment-graded intervention-effect re-evaluation".

[0039] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0040] The present invention is described with reference to flowcharts and / or block diagrams of systems, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0041] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0042] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0043] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0044] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. An intelligent early warning system for airway management of otolaryngology patients, characterized by: is configured to perform the following steps: Acquiring first physiological data representing an airflow state, second physiological data representing an airway tissue state, and disturbance data representing a sleep state of the individual; Processing the first physiological data to generate a first risk indicator for characterizing airflow turbulence effects , and processing the second physiological data to generate a second risk indicator for characterizing tissue support capacity ; Through individual adaptive calibration parameters , for the first risk indicator and the second risk indicator Perform fusion calculations to determine airway stability margins , and combined with the airway stability margin and the disturbance data to generate the basic risk index ; According to the first risk indicator With the second risk indicator The amount of change within a preset time window determines the dynamic covariance correction factor used to correct the contradictory risk indicators ; By With the dynamic covariance correction factor Perform combination operations to generate the final risk index ; The final risk index The final risk index is compared with a preset risk threshold, and an early warning signal is output when the final risk index meets the preset conditions.

2. The intelligent early warning system for airway management of otolaryngology patients according to claim 1 is characterized in that: The first physiological data is set to an upper airway breathing audio signal; The second physiological data is set to be a multi-frequency bioimpedance spectrum of parapharyngeal tissue; The disturbance data is set to include a set of sleeping position data and sleep stage data.

3. The intelligent early warning system for airway management of otolaryngology patients according to claim 1 is characterized in that: The step of processing the first physiological data and the second physiological data further includes: Performing a short-time Fourier transform on the first physiological data to calculate spectral entropy and high-order moments, and setting the calculation results of the spectral entropy and high-order moments as the first risk index ; A Cole-Cole model is established based on the second physiological data to solve the equivalent tissue impedance modulus, and the equivalent tissue impedance modulus is set as the second risk indicator. .

4. The intelligent early warning system for airway management of otolaryngology patients according to claim 1 is characterized in that: The individual adaptive calibration parameters The determination process includes: Perform the initial calibration step to set the basic physiological information input by the user when using it for the first time. Initial baseline value of Perform continuous adaptive correction steps to analyze the correlation between the model prediction results and the actual effects of the intervention measures. Perform closed-loop feedback fine-tuning.

5. The intelligent early warning system for airway management of otolaryngology patients according to claim 2 is characterized in that: The underlying risk index The generation of is further restricted to the following computation: ； and For specific users The individual adaptive sensitivity weights representing the effects of body position and sleep stage respectively; is a body position factor extracted from the sleeping body position data; is a staging factor extracted from the sleep staging data.

6. The intelligent early warning system for airway management of otolaryngology patients according to claim 1 is characterized in that: The dynamic covariance correction factor The determination of is further defined as the following calculation: ； and is the rate of change of the first risk indicator and the second risk indicator calculated within a preset time window; is a preset covariance correction sensitivity coefficient used to adjust the model's penalty intensity for contradictory signals. The value of is determined based on offline optimization of historical clinical datasets.

7. The intelligent early warning system for airway management of otolaryngology patients according to claim 1 is characterized in that: The combination operation is further defined as the basic risk index With the dynamic covariance correction factor Multiply.

8. The intelligent early warning system for airway management of otolaryngology patients according to claim 1 is characterized in that: The individual adaptive sensitivity weight and The determination process includes: Assigning a universal benchmark value to the weight as an initial calibration based on large-scale clinical statistical data; In long-term use, the weights are individually and continuously fine-tuned automatically by analyzing the correlation between interventions and actual changes in risk indicators.

9. The intelligent early warning system for airway management of otolaryngology patients according to claim 1, characterized in that: The preset risk thresholds include: The first level risk threshold is used to identify the risk state of significantly decreased airway stability; The second level risk threshold, whose value is set to be higher than the first level risk threshold, is used to identify a critical risk state of impending apnea.

10. The intelligent early warning system for airway management of otolaryngology patients according to claim 9, characterized in that: The output of the early warning signal is configured to trigger a graded intervention strategy, which includes: When the final risk index When the first level risk threshold is exceeded, the flexible intervention module is activated to guide the user to unconsciously change their sleeping position; When the final risk index When the second-level risk threshold is exceeded, it is used to start the enhanced intervention module to promote the user to a lighter sleep stage or adjust the pressure of the continuous positive airway pressure device.

Images