Swallow-triggered eustachian tube inflation control system based on signal recognition

Abstract

This invention relates to the field of medical device control technology and discloses a swallowing-triggered Eustachian tube inflation control system based on signal recognition. The system includes an electromyography (EMG) acquisition module, a nasal actuation module, a pressure sensing module, a gas storage module, an air pump module, a solenoid valve module, and a main control module. The EMG acquisition module acquires electrophysiological signals from the swallowing muscles; the pressure sensing module acquires nasal cavity pressure fluctuation signals and actual pressure data; the main control module confirms the effective swallowing cycle based on the received electrophysiological signals and pressure fluctuation signals, calculates the optimal ventilation timing, and determines the target conduction pulse width time. It then controls the solenoid valve module to open according to the target conduction pulse width time, quantitatively releasing compressed gas from the gas storage module into the user's nasal cavity. This invention achieves synchronous airflow triggering through physiological signal fusion and self-calibrated dynamic thresholds, and utilizes a bypass blind cavity and interruption depressurization mechanism to prevent cross-contamination and pressure damage.

Description

Technical Field

[0001] This invention relates to the field of medical device control technology, specifically to a swallowing-triggered Eustachian tube inflation control system based on signal recognition. Background Technology

[0002] The Eustachian tube is a physiological passage connecting the middle ear and the nasopharynx, mainly used to maintain the air pressure balance in the middle ear. In physical interventions for Eustachian tube dysfunction, a Eustachian tube inflation device is usually used to introduce positive pressure airflow into the nasal cavity, forcing the Eustachian tube to open and achieve ventilation.

[0003] Since the Eustachian tube is only physiologically open at the moment of swallowing, the introduction of positive pressure airflow must be synchronized with the swallowing action. Most existing ventilation devices rely on medical staff to manually control the airflow release or use a continuous air supply mode. This method makes it difficult to accurately capture the patient's true swallowing time window, often resulting in airflow being output when the Eustachian tube is closed. This not only fails to achieve effective ventilation but also causes ear pain in patients. Some devices attempt to trigger ventilation by monitoring a single change in nasal pressure, but due to physiological differences in the swallowing force habits of different patients, using a fixed air pressure trigger threshold can lead to false triggering or missed detection by the device.

[0004] Existing equipment also has limitations in terms of air pressure control and pipeline structure design. During positive pressure airflow output, if the equipment lacks real-time monitoring and hardware-level safety blocking mechanisms, once airflow control fails, the continuous high-pressure airflow will cause air pressure damage to the tympanic membrane. At the same time, the pressure sensing elements inside the existing inflation system are usually directly exposed in the main air passage. During ventilation or exhaust, nasal secretions carried back by the airflow will adhere to the sensor surface. This will not only interfere with the accuracy of subsequent data acquisition by the sensor, but also increase the hygiene risks when the equipment is reused. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a swallowing-triggered Eustachian tube inflation control system based on signal recognition, which solves the problems of asynchronous ventilation timing with effective swallowing action, fixed trigger thresholds that are prone to misjudgment, and lack of air pressure safety blocking mechanism in existing Eustachian tube inflation devices.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a swallowing-triggered Eustachian tube inflation control system based on signal recognition, the system comprising: An electromyography (EMG) acquisition module, placed on the user's skin surface, is used to acquire electrophysiological signals from the swallowing muscles. The nasal end execution module, the outer surface of which is used to physically fit with the user's anterior nasal nostrils to form an airtight seal, and to introduce positive pressure airflow into the user's nasal cavity; A pressure sensing module is installed in the air passage inside the nasal end execution module to collect the air pressure fluctuation signal inside the nasal cavity and the actual pressure data inside the nasal cavity in real time. Gas storage module, used to store compressed gas; An air pump module, the output end of which is connected to the input end of the air storage module, is used to inject air into the air storage module; The solenoid valve module has its input end connected to the output end of the gas storage module and its output end connected to the airflow input end of the nose actuator module. The main control module is electrically connected to the electromyography acquisition module, the pressure sensing module, the air pump module, and the solenoid valve module, respectively. The main control module is used to receive the electrophysiological signals of the swallowing muscles and the air pressure fluctuation signals inside the nasal cavity, and based on the received signals, confirm that the current swallowing cycle is in place, calculate the optimal ventilation timing, and then calculate the target conduction pulse width time that matches the optimal ventilation timing. The output level signal controls the solenoid valve module to open according to the target conduction pulse width time, so as to release the compressed gas in the air storage module into the user's nasal cavity in a quantitative manner.

[0007] Furthermore, the main control module is internally configured with a baseline self-calibration program and a logic judgment program: During the baseline self-calibration procedure, the main control module extracts the envelope features of the calibration period reference signal obtained during a single empty swallowing action of the user, calculates the reference peak value and the reference integral area, and generates a dynamic trigger threshold based on the reference peak value. When the system enters the real-time monitoring phase, the main control module performs rectification and filtering on the real-time raw electromyography signal to generate an envelope signal, and performs differentiation on the nasal cavity air pressure fluctuation signal to generate a first derivative sequence. When the main control module determines that the envelope signal is greater than or equal to the dynamic trigger threshold, and the nasal cavity air pressure fluctuation signal and the first derivative sequence are both greater than or equal to zero, the main control module confirms that it is currently in an effective swallowing cycle.

[0008] Furthermore, the main control module also deploys a deep prediction model based on a fusion of one-dimensional convolution and long short-term memory structures: The main control module will identify the time window that is in an effective swallowing cycle as a candidate time window to wake up the depth prediction model. When the main control module determines that the probability of the prediction output by the depth prediction model is greater than the preset action trigger threshold, it extracts the start and end time nodes of the effective swallowing cycle. The main control module extracts the real-time action integral area within the start and end time nodes, calculates the force ratio parameter based on the ratio of the real-time action integral area to the reference integral area, and converts the force ratio parameter into the target work amount to calculate the target conduction pulse width time corresponding to the target work amount.

[0009] Furthermore, the nasal end actuation module is configured with a housing unit, a sealing adapter unit, and an internal air guiding unit: The outer contour of the outer shell unit is a radially tapered frustum, and the outer surface of the outer shell unit is fixed with the elastic sealing adapter unit. The outer shell unit has an internal air guiding unit that runs through both ends. The hollow part of the side wall of the internal air guiding unit extends to form a bypass blind cavity that communicates with the main air passage. The other end of the bypass blind cavity is closed and isolated. The pressure sensing module is independently installed in the bypass blind cavity to collect the air pressure fluctuation signal inside the nasal cavity without direct physical contact with the user.

[0010] Furthermore, the electromyography (EMG) acquisition module is equipped with a surface electrode unit, a signal conditioning unit, an analog-to-digital conversion unit, and a data transmission unit. The surface electrode unit includes a measuring electrode and a reference electrode, and the electrophysiological signal of the swallowing muscle group is obtained by extracting the potential difference between the two. The signal conditioning unit is equipped with a differential amplifier circuit and a cascaded filter network, which are used to perform differential amplification, fixed-band filtering and notch filtering on the acquired signal in sequence to generate a continuous analog signal. The analog-to-digital conversion unit is used to discretize the continuous analog signal into a digital sequence, which is then packaged by the data transmission unit and sent to the main control module via a serial communication bus.

[0011] Furthermore, the system incorporates a security blocking mechanism in its execution architecture: During the process of the solenoid valve module opening to release gas, the main control module uses time-division multiplexing technology to read in real time the actual pressure data in the nasal cavity collected and generated by the pressure sensing module; When the main control module determines that the target conduction pulse width time has ended, or detects that the actual pressure data in the nasal cavity triggers the preset safety upper limit, the main control module outputs an interrupt command through hardware interrupt logic to pull down the drive level and control the solenoid valve module to shut down and close.

[0012] Furthermore, a closed-loop regulation mechanism and a passive safety structure are configured between the air pump module and the air storage module: The main control module monitors the feedback gas pressure data in the gas storage module in real time. Based on the difference between the real-time pressure and the target pressure, it dynamically adjusts the pulse width modulation duty cycle of the motor in the gas pump module drive unit in combination with the proportional-integral-derivative-increment algorithm to maintain the basic gas storage pressure value. The gas storage module is equipped with a mechanical pressure exhaust unit on its outer shell. When the internal gas pressure of the gas storage module exceeds the preset physical elastic deformation limit resistance point, the mechanical pressure exhaust unit is passively opened to perform physical-level pressure relief and exhaust.

[0013] Furthermore, the main control module is externally connected to a communication and storage module, and the communication and storage module is internally configured with a treatment efficacy evaluation unit. After the system performs a single positive pressure deflation operation, the efficacy assessment unit reads the dynamic decay sequence of nasal pressure within the ventilation time window continuously collected by the pressure sensing module. The efficacy assessment unit extracts the characteristics of the trough of the sudden drop in air pressure, and calculates a comprehensive scoring parameter by weighting the integral value of the pressure waveform in the time domain with the extreme pressure drop steep change rate signal according to a preset weighting coefficient. If the comprehensive scoring parameter is greater than the lower limit of the verification threshold, an effectiveness label indicating physical connection is added to the physiological waveform data of this intervention.

[0014] Furthermore, the system is configured with a power management module that provides independent power supply, which internally includes a power status monitoring unit: The power status monitoring unit calculates the remaining number of complete ventilation intervention cycles that can be safely maintained based on the integral of the load change current in the battery discharge circuit and the ambient temperature data fed back by the external thermistor. When the number of remaining ventilation intervention cycles is lower than the preset minimum threshold, the power management module triggers a hard interrupt interception command to actively lock the opening permission of the solenoid valve module.

[0015] Furthermore, the main control module is externally connected to a human-machine interface module, which is equipped with a status alarm unit. The status alarm unit integrates adaptive volume adjustment control logic based on audio feedback. The status alarm unit uses a built-in sound acquisition circuit to continuously extract the background noise level in decibels of the current space. When the system is in an early warning state and drives an external buzzer to output an early warning audio, the system controls the audio frequency gain of the early warning audio to perform dynamic sound pressure compensation adjustment in accordance with the background noise level.

[0016] The principle of this invention is as follows:

[0017] The system synchronously acquires electrophysiological signals of the swallowing muscles on the body surface and air pressure fluctuation signals inside the nasal cavity through the electromyography acquisition module and the pressure sensing module. Based on the above signals, the timing of the swallowing action is determined. The main control module introduces a baseline self-calibration program to dynamically generate trigger thresholds based on the individual physiological differences of the user, avoiding false triggers or missed triggers caused by fixed thresholds. Combined with a deep prediction model with a one-dimensional convolution and long short-term memory composite structure, the system extracts the integral area of ​​the real-time action, compares it with the reference integral area during the calibration period, calculates the force ratio parameter, and then calculates the target conduction pulse width time corresponding to the target work done. This control logic synchronizes the opening time of the solenoid valve module with the physiological opening window of the Eustachian tube to achieve quantitative release of positive pressure airflow.

[0018] In addition, the bypass blind cavity structure of the nasal end actuator module allows the pressure sensing module to collect air pressure data in a non-contact state, isolating contaminants in the main air path. The system is equipped with a safety blocking mechanism based on hardware interruption, which forcibly cuts off power and closes the solenoid valve when the pressure in the nasal cavity reaches the safety limit. The air storage module is equipped with a dual structure of closed-loop regulation and mechanical passive pressure relief to prevent airflow pressure overload.

[0019] This invention provides a swallowing-triggered Eustachian tube inflation control system based on signal recognition. It has the following beneficial effects:

[0020] 1. This invention dynamically generates trigger thresholds through the baseline self-calibration program built into the main control module, extracts the real-time action integral area and calculates the force ratio parameter by combining the depth prediction model, and then calculates the target conduction pulse width time. This technical feature enables the system to adapt to the physiological differences of different users and simultaneously open the solenoid valve during the physiological opening period of the Eustachian tube, realizing the quantitative release of positive pressure airflow and improving the accuracy of airflow triggering and release.

[0021] 2. The nasal end execution module of the present invention extends to the side wall of the internal air guiding unit to construct a closed and isolated bypass blind cavity, and independently sets the pressure sensing module in the bypass blind cavity. This technical feature enables the pressure sensing module to continuously collect the air pressure fluctuation signal inside the nasal cavity without directly contacting the main airflow, physically isolating the secretions present in the main airway and reducing the risk of cross-contamination during the use of the device.

[0022] 3. The present invention introduces a safety blocking mechanism in the control architecture by using hardware interrupt logic to output interrupt instructions to pull the drive level low, and equips the air storage module with a mechanical pressure exhaust unit. This technical feature enables the system to cut off the airflow by actively cutting off the power and closing the valve and passively depressurizing when it detects that the actual pressure in the nasal cavity triggers the safety limit or the internal air pressure exceeds the physical limit resistance point, so as to prevent air pressure overload from causing damage to the user. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the system architecture of the present invention; Figure 2 This is a schematic diagram of the usage method of the present invention; Figure 3 This is a schematic diagram of the signal flow of the electrical acquisition module of the present invention; Figure 4 This is a schematic diagram of the main installation layout and signal acquisition of the pressure sensing module of the present invention; Figure 5 This is a schematic diagram of the control architecture of the positive pressure gas supply layer of the present invention; Figure 6 This is a schematic diagram of the operational architecture of the human-computer interaction module and the power management module of the present invention; Figure 7 This is a schematic diagram of the communication and storage module and the corresponding closed-loop evaluation architecture of the present invention; Figure 8 This is a waveform diagram illustrating the electromyographic envelope extraction and dynamic threshold triggering determination of the present invention. Figure 9 This is a schematic diagram of the curves representing the probability of occurrence of the prediction model and the evaluation of the optimal ventilation timing in this invention. Figure 10 This is a schematic diagram of the pressure response characteristics of a single targeted blowing operation according to the present invention. Figure 11 This is a bar chart illustrating the comparison and evaluation of the intervention effectiveness and core clinical indicators of the present invention. Detailed Implementation

[0024] Please see the appendix Figure 1 , Figure 1 This is a schematic diagram of a system architecture according to an embodiment of the present invention. The present invention provides a swallowing-triggered Eustachian tube inflation control system based on signal recognition, including: an electromyography acquisition module, a pressure sensing module, an air pump module, an air storage module, a solenoid valve module, a main control module, and a nasal end execution module.

[0025] The electromyography (EMG) acquisition module is positioned on the skin surface of the user's neck to acquire electrophysiological signals from the swallowing muscles. The nasal actuation module is configured with a nasal olive-shaped head structure, with its outer surface physically attached to the edge of the user's anterior nostrils to form an airtight seal, which is used to introduce positive pressure airflow into the user's nasal cavity. The pressure sensing module is located in the air passage inside the nasal actuation module, without directly contacting the human skin, and is used to acquire in real time the air pressure fluctuation signals inside the nasal cavity under airtight sealing conditions.

[0026] The main control module is electrically connected to the electromyography (EMG) acquisition module, pressure sensing module, air pump module, and solenoid valve module. It receives the electrophysiological signals of the swallowing muscle group acquired by the EMG acquisition module and the air pressure fluctuation signals inside the nasal cavity acquired by the pressure sensing module. It is responsible for signal processing, logic judgment, and execution command output. The output end of the air pump module is connected to the input end of the air storage module. The output end of the air storage module is connected in series with the solenoid valve module through a pipeline and finally connected to the airflow input end of the nasal execution module.

[0027] Please see the appendix Figure 2 , Figure 2 This is a schematic flowchart of a method according to an embodiment of the present invention. The present invention provides a method for using a swallowing-triggered Eustachian tube inflation control system based on signal recognition, comprising the following steps:

[0028] S10, the main control module drives the air pump module to inject air into the air storage module. The main control module monitors the air pressure feedback data in the air storage module in real time and adjusts the output power of the air pump module to keep the air pressure in the air storage module at the preset basic air storage pressure value. The system enters the background high-pressure standby state.

[0029] S20, execute the baseline self-calibration procedure. The user performs a single empty swallowing action. The electromyography acquisition module acquires the calibration period reference signal and sends it to the main control module. The main control module extracts the envelope features of the calibration period reference signal and calculates the reference peak value and reference integral area. The main control module generates a dynamic trigger threshold based on the reference peak value and stores the dynamic trigger threshold and reference integral area as the reference data for subsequent judgment in the main control module.

[0030] S30, the system enters the real-time monitoring stage. The electromyography acquisition module and the pressure sensing module simultaneously acquire real-time raw electromyography signals and nasal air pressure fluctuation signals and transmit them to the main control module. The main control module performs rectification and filtering on the real-time raw electromyography signals and generates envelope signals. At the same time, the main control module performs differentiation on the nasal air pressure fluctuation signals to generate a first derivative sequence.

[0031] S40, the main control module calls the dynamic trigger threshold generated in the baseline self-calibration program to perform logical judgment. When the main control module determines that the envelope signal generated in the real-time monitoring stage is greater than or equal to the dynamic trigger threshold, and the nasal air pressure fluctuation signal and the first derivative sequence are both greater than or equal to zero, the system confirms that the current monitoring period is in the effective swallowing cycle and non-inspiratory phase, and uses this as a candidate time window to wake up the internal depth prediction model for secondary probability calculation.

[0032] When the main control module determines that the probability of the predicted event output by the depth prediction model is greater than the preset action trigger threshold, it determines the target ventilation timing when the Eustachian tube is about to open and extracts the start and end time nodes of the effective swallowing cycle.

[0033] S50, the main control module extracts the real-time motion integral area within the effective swallowing cycle based on the start and end time nodes of the cycle. The main control module calculates the force ratio parameter based on the ratio of the real-time motion integral area to the reference integral area stored in the baseline self-calibration program. The main control module converts the force ratio parameter into the target work amount for this ventilation through an internal mapping function, and calculates the target conduction pulse width time required for the solenoid valve module to achieve the work amount. The calculated target conduction pulse width time is limited to the dynamic instantaneous cavity pressure corresponding to the calculated dynamic instantaneous cavity pressure not exceeding the preset safety upper limit.

[0034] S60, the main control module outputs a level signal to control the solenoid valve module to open according to the generated target conduction pulse width time. The pressurized gas with constant pre-pressure in the gas storage module enters the user's nasal cavity quantitatively through the nasal end execution module according to the target conduction pulse width time. During the ventilation and gas release process, the main control module collects and generates the actual pressure data in the nasal cavity in real time through the pressure sensing module.

[0035] S70, when the main control module determines that the target conduction pulse width time has ended or the preset safety upper limit value has been triggered, the main control module outputs an interrupt command to control the solenoid valve module to close, the single blowing process ends, the air pump module resumes the air injection operation to the air storage module until the air pressure in the air storage module is restored to the basic air storage pressure value, and the system cyclically executes the signal acquisition and logic judgment process.

[0036] Please see the appendix Figure 3 , Figure 3 This is a schematic diagram of the signal flow of an electromyography (EMG) acquisition module according to an embodiment of the present invention.

[0037] In this embodiment, the electromyography (EMG) acquisition module is equipped with a surface electrode unit, a signal conditioning unit, an analog-to-digital conversion unit, and a data transmission unit. The EMG acquisition module is used to non-invasively acquire the electrophysiological signals of the user's neck swallowing muscles, and to establish a stable transmission path after digital processing of the acquired physical signals in order to send digital sequences to the main control module.

[0038] Regarding the starting point for generating the aforementioned digital sequence, the surface electrode unit is attached to the skin surface of the user's neck. The surface electrode unit includes a measuring electrode and a reference electrode. In a preferred embodiment, the measuring electrode is positioned on the exposed skin above the user's submental muscles, and the reference electrode is positioned on the exposed skin at the edge of the user's clavicle. The surface electrode unit obtains the electrophysiological signal of the swallowing muscles by extracting the weak potential difference between the measuring electrode and the reference electrode.

[0039] For the conductive gel ratio and anti-sensitivity adhesive material of the disposable patch in the surface electrode unit, those skilled in the art can select according to the physical standards of conventional medical surface electrodes. Its impedance matching and breathable insulation isolation are well known technologies in the field and will not be described in detail here.

[0040] After acquiring the above electrophysiological signals, the system focuses on different aspects of signal processing at different operating stages. When the system is in the baseline self-calibration program, the user performs a complete empty swallowing action. The surface electrode unit collects the potential difference changes of the swallowing muscle group throughout the time window of the action and generates a calibration period reference signal. At this time, the calibration period reference signal is input to the subsequent signal conditioning unit in the form of an analog voltage signal.

[0041] Corresponding to the aforementioned state, when the system is in the real-time monitoring stage, the surface electrode unit continuously collects and inputs the real-time raw electromyographic signals to the signal conditioning unit in the same manner. Considering that the electrical signals extracted from the human body surface are extremely weak and accompanied by environmental noise interference, the signal conditioning unit is equipped with a differential amplifier circuit with high input impedance and a cascaded filter network. In order to preserve the core electromyographic energy distribution of swallowing action and filter out power frequency interference and basic physiological migration baseline, the signal conditioning unit performs differential amplification, fixed frequency band filtering and notch filtering on the real-time raw electromyographic signals in sequence.

[0042] For example, the cutoff frequency of a fixed-band filter ranges from 20Hz to 500Hz to accurately cover the main discharge frequency range of the swallowing muscles, thereby eliminating invalid noise outside the band and outputting a smooth, noise-free continuous analog signal.

[0043] The continuous analog signal processed by the signal conditioning unit needs to be converted into a discrete form recognizable by the microprocessor. This continuous analog signal is transmitted to the analog-to-digital converter (ADC) via hardware circuitry. The ADC discretizes the continuous analog signal into a digital sequence in the time and amplitude domains according to a set high-frequency sampling rate and quantization resolution. The specific discretization generation process is implemented based on the following definition: ; In the formula, This is a discrete sequence of pharyngeal electromyographic digital signals acquired by the electromyography acquisition module and output after analog-to-digital conversion; The continuous analog electromyography signal is captured by the front-end patch electrode and amplified and filtered by the underlying hardware conditioning circuitry. This refers to the current sequence index in the discretization sampling process; The reference sampling time interval for the analog-to-digital conversion unit to perform signal discretization processing ranges from 1ms to 5ms.

[0044] In order for the discretized data sequence to enter the control terminal and play a judgment role, the data transmission unit is connected to the internal data register of the analog-to-digital conversion unit to read the generated digital sequence in real time. The external pins of the data transmission unit establish a physical electrical connection network with the main control module of the system.

[0045] The data transmission unit packages the read digital sequence according to bit order and sends it stably to the main control module via the serial communication bus. The main control module receives the digital sequence as the front-end basis for the baseline self-calibration program and subsequent complex logic operations. For the baud rate negotiation configuration and anti-backlash design of the serial communication bus between the data transmission unit and the main control module, those skilled in the art can refer to the microcontroller communication manual to perform register read and write deployment. The underlying timing level conversion and data frame verification are well-known technologies in the field and will not be described in detail here.

[0046] In this embodiment, the nasal actuator module serves as the terminal physical interface of the system's airway. It is mainly equipped with a housing unit, a sealing adapter unit, and an internal air guiding unit. The nasal actuator module as a whole is used to introduce positive pressure airflow into the user's nasal cavity and establish the necessary sealed working environment for the stable acquisition of subsequent air pressure signals.

[0047] To accommodate the differences in anterior nasal nostrils among users with different physiological characteristics, the main configuration of the nasal end execution module is configured as a nasal olive-shaped structure. As a preferred approach, the outer contour of the outer shell unit is a radially gradually changing frustum to physically fit the inner diameter of the nostrils of different sizes. The outer surface of the outer shell unit is fixed with a sealing adapter unit through an encapsulation process. The sealing adapter unit is made of medical-grade soft elastic material, and the material hardness distribution on its radial cross-section shows the characteristics of being hard on the inside and soft on the outside.

[0048] The base layer material near the central axis has a relatively high hardness to maintain the support strength of the internal air guiding unit during ventilation and prevent airflow interruption or deformation distortion due to pressure; while the material in direct contact with the surface layer has a lower hardness, which can undergo elastic deformation under pressure in order to conform to the surface contour of the nasal cavity wall.

[0049] During actual ventilation, the user pushes the nasal actuator module in and presses it against the nose. The outer surface of the sealing adapter unit then presses against the edge of the user's anterior nasal cavity. Due to the deformation properties of the outer material, the rigid gaps between the contact surfaces are effectively blocked, thus forming a stable airtight seal. Ensuring this seal is a prerequisite for preventing leakage when the system injects high-frequency positive pressure airflow into the human body. Regarding the fluid-physical reverse thrust during the introduction of positive pressure airflow, the static friction generated between the sealing adapter unit and the anterior nasal cavity tissue must overcome and balance this reverse thrust. The mechanical equilibrium boundary that prevents slippage between the two is constrained by the following formula: ; In the formula, To prevent static friction between the outer surface of the sealing adapter unit and the edge tissue of the user's anterior nasal cavity; The real-time fluid pressure accumulated inside the gas storage module and associated gas supply pipeline as the micro air pump performs work; This refers to the effective cross-sectional area of ​​the nasal actuator module subjected to the reverse thrust of the airflow.

[0050] As long as the above mechanical boundary conditions are met, the original airtight sealing structure can be maintained and will not fail due to air pressure impact. The system uses the user's physical thrust to increase the effective contact cross-sectional area and static friction, thereby ensuring smooth inflation and deflation of the internal channel.

[0051] For the internal flow logic of the controlled gas, an internal gas guiding unit is provided inside the outer shell unit, which runs through both ends of the unit. The internal gas guiding unit extends longitudinally along the central axis and forms a straight hollow flow channel. Its gas supply input end is connected to the output interface of the solenoid valve module via an external conduit. When a gas supply command is issued, the internal gas guiding unit guides the high-pressure airflow released by the gas storage module and passing through the solenoid valve module to achieve directional delivery.

[0052] To reduce eddy separation and aerodynamic noise caused by surface roughness when gas flows through the restricted channel at high speed, the inner wall cavity of the internal air guiding unit has been smoothed and polished. This reduces pressure loss along the flow path and improves the transient response of airflow loading from a fluid dynamics perspective.

[0053] While constructing a direct airway, in order to achieve parallel extraction of air pressure fluctuation signals, a bypass blind cavity is constructed in the hollow part of the side wall of the internal air guiding unit, which is connected to the main airway. The other end of the bypass blind cavity is completely sealed, thereby isolating it from the external physical space that directly contacts the human body. Based on this internal structure, the pressure sensing module is independently set in the bypass area of ​​the airway channel inside the nasal end execution module.

[0054] This concealed installation method avoids direct contact between the pressure sensing module and the exposed skin of the human body, reducing the risk of cross-infection in clinical settings. The pressure sensing module, placed in the bypass blind cavity, collects the air pressure fluctuation signal inside the nasal cavity under airtight sealing conditions.

[0055] Regarding the aforementioned hot pressing molding process and mold injection channel design, those skilled in the art can set the structure according to conventional polymer molding fluid dynamics standards. The stress relief and sealant bonding process are well-known technologies in this field and will not be elaborated here.

[0056] Please see the appendix Figure 5 , Figure 5 This is a schematic diagram of the main body installation layout and signal acquisition of a pressure sensing module according to an embodiment of the present invention. In this embodiment, the pressure sensing module is used to perform full-cycle quantitative acquisition of air pressure changes inside the nasal cavity. In order to achieve non-destructive detection at the fluid physics level and avoid redundancy in the front-end structure space, the pressure sensing module is independently set in the bypass probe blind cavity extending from the side wall of the internal air guiding unit of the nasal end execution module.

[0057] As a preferred approach, this concealed assembly structure makes full use of the uniform pressure distribution characteristics of gas within a confined micro-cavity. The pressure-sensitive surface of the pressure sensing module is parallel to the mainstream airflow channel of the internal air guiding unit, thereby establishing a connected fluid dynamic detection node without obstructing the directional delivery of positive pressure airflow. This assembly layout ensures that the pressure sensing module does not come into direct physical contact with the user's exposed skin, thus blocking potential cross-infection routes during clinical use.

[0058] Meanwhile, thanks to the mechanical support of the robust outer shell of the nose-end actuator module, the pressure sensing module is effectively isolated from the mechanical transmission path of external force collisions, thereby ensuring the stability of the measurement reference of the internal sensitive components.

[0059] Based on the aforementioned interconnected detection structure, the system executes appropriate acquisition logic at different operating stages. During the real-time monitoring stage, the system has not yet output positive pressure airflow. Since the front-end solenoid valve module is normally closed, relying on the airtight seal established between the nasal end execution module and the user's nostrils, the hollow flow channel inside the nasal end execution module and the user's nasal cavity together constitute a relatively sealed physical detection cavity. Based on the basic principles of fluid mechanics and the gas state equation, within a fixed amount of sealed space, the gas pressure is inversely proportional to the volume change of the cavity.

[0060] When a user performs normal breathing or swallowing, the physiological contraction and relaxation of the airway muscles cause minute changes in tissue volume, which in turn triggers minute airflow migration and pressure fluctuations within the sealed detection cavity. The pressure sensing module senses and extracts these pressure fluctuations through the lateral blind cavity, indirectly acquiring the nasal air pressure fluctuation signal caused by the phase transition between exhalation and inhalation. In order to quantify the fluid pressure changes into a digital sequence that the subsequent main control module can recognize, the pressure sensing module records instantaneous air pressure parameters at a fixed high-frequency sampling trigger rate. The specific physical mapping conversion relationship is described based on the following formula: ; In the formula, Indicates that it is in the detection time The instantaneous value of the nasal cavity air pressure fluctuation signal collected and output by the pressure sensing module; The voltage-to-gas pressure conversion sensitivity coefficient is physically calibrated at the factory and written into the register section for the piezoresistive sensing unit. The value of this coefficient ranges from 0.5 kPa / V to 2.0 kPa / V. Indicates in The transient continuous output voltage value after differential amplification processing by the module's internal conditioning circuit; This indicates the zero-point drift reference bias voltage of the hardware under absolute standard atmospheric pressure conditions without fluid excitation.

[0061] As part of the parameter determination logic, the system pre-reads the ambient background voltage and stores it in an internal register during the single power-on initialization phase, thereby dynamically assigning... With the current accurate values, the main control module can output a continuous respiratory phase time fluctuation sequence by reading and solving the above difference amplification formula in real time.

[0062] When the system's main control logic determines that the preset deflation conditions are met and enters the ventilation stage, the aforementioned passive monitoring state undergoes an adaptive change. During this stage, the solenoid valve module opens instantly, releasing the pre-set high-pressure gas in the gas storage module and introducing it into the user's nasal cavity. In response to this drastic change in physical state, the pressure sensing module switches to dynamic monitoring mode, changing its actual monitoring target from weak physiological respiratory waves to high-intensity dynamic inflation pressure. Relying on its wide-range detection properties, the pressure sensing module generates and reports the actual dynamic pressure data within the nasal cavity reflecting the current inflation state to the main control module in real time on the underlying data bus. This creates a closed-loop control network with high-frequency data updates between the main control module and the front-end airway.

[0063] The main control module continuously reads in the actual dynamic pressure data inside the nasal cavity and compares and verifies it with the preset target pressure and safety protection threshold generated by the algorithm. This closed-loop detection mechanism ensures that the fluid work done by the system during positive pressure blowing is always within the controlled boundary, providing the underlying data calculation basis for the solenoid valve module to perform emergency shut-off protection actions.

[0064] For the microcomputer voltage resistance physical sensing principle used in the pressure sensing module and the specific board layout and wiring of the Wheatstone bridge network amplifier circuit, those skilled in the art can refer to the application specifications and deployment requirements of standard industrial-grade micro pressure sensors to draw the circuit diagram. Its electromagnetic shielding design and bidirectional temperature drift compensation calibration are well-known technologies in this field and will not be described in detail here.

[0065] In this embodiment, the main control module, as the key computing platform for the system to implement closed-loop control and low-level data parsing, is mainly configured with a data receiving unit, a feature preprocessing unit, a neural network unit, and an instruction issuing unit. The main control module establishes a stable physical communication link with the aforementioned electromyography acquisition module, pressure sensing module, and the back-end solenoid valve module responsible for airflow release through the internal system data bus. This is used to fuse multi-source heterogeneous physiological records of the human body in real time and accurately determine the optimal ventilation timing for controlling the opening of the airway.

[0066] To address the frequency asynchrony issue between electromyography (EMG) feedback signals and air pressure fluctuation signals at the underlying acquisition hardware, the data receiving unit continuously reads the EMG digital sequence reported by the EMG acquisition module and the nasal air pressure fluctuation signal submitted by the pressure sensing module after the system is woken up and enters the real-time monitoring phase. Considering the objective differences between the internal clock references and analog-to-digital conversion sampling rates of the two sets of physical sensors, the feature preprocessing unit internally incorporates digital signal processing logic based on timestamp synchronization alignment. This processing logic, based on linear interpolation, resamples the two discretely acquired time-domain sequences to an absolutely consistent time dimension.

[0067] As a preferred approach, the feature preprocessing stage after synchronization uses the current trigger time of the system clock as the zero point reference, traces back along the time axis and extracts data segments of a fixed time window length, and then splices them together to synthesize a two-dimensional input feature matrix. The horizontal span parameter of the time window ranges from 400ms to 600ms. Within this hardware extraction range, the acquired data array is sufficient to completely capture the pre-swallowing electromyographic latent waveform at the beginning of a single swallowing action, while also taking into account the transient phase change trajectory caused by respiratory airflow traction.

[0068] Assume the uniform sampling frequency determined after resampling is Then the extracted and constructed two-dimensional input feature matrix has (0.5⋅ The specific numerical dimension is 2×2, and its two columns of channels are respectively mapped to the aligned electromyographic amplitude and the relative pressure value of the air chamber.

[0069] Based on the two-dimensional input feature matrix extracted by resampling, the system's internal bus imports it into the neural network unit for deep state analysis. In order to establish the correlation features between multi-source time-series signals and the physical state of the Eustachian tube being open, the neural network unit deploys a deep prediction model based on a fusion of one-dimensional convolution and long short-term memory in its internal runtime memory.

[0070] The processing topology of this depth prediction model is defined sequentially from front to back as a spatial feature extraction layer, a temporal dependency layer, and a mapping output layer. In the flow mechanism of the underlying computation, the first end of the two-dimensional input feature matrix is ​​imported into the one-dimensional convolutional unit encompassed by the spatial feature extraction layer. This level of operation is equipped with a sliding tensor convolution kernel with a field of view size of 1×5, and performs local inner product processing on the dual-channel data source step by step along the set temporal vertical axis. This feedforward process is specifically used to identify high-amplitude energy mutation signals and small pressure drop troughs in the electromyography frequency band amidst complex physiological background interference.

[0071] The high-dimensional implicit feature map output by the convolutional layer is subjected to information dimensionality reduction operation by the internal max pooling layer and input into the long short-term memory network structure built into the temporal dependency layer. The long short-term memory network structure quantifies the long-term causal transmission relationship of the accumulation process of electromyographic potential energy at different time points and the resulting changes in the local flow field of the nasal cavity through the internal parameterized forgetting threshold structure and sequence state update gating. The hidden state vector matrix of each time step after iterative purification is finally converged and handed over to the terminal mapping output layer for processing.

[0072] This layer utilizes a fully connected network and a Sigmoid activation function to perform non-linear dimensionality collapse operations. The final result of its single output node is mapped to the predicted probability that the user's current Eustachian tube anatomy is in a state of imminent opening.

[0073] To ensure that the applied deep prediction model possesses reliable generalization reasoning capabilities when dealing with complex and individual physiologically diverse environments, the system relies on the underlying firmware of the main control module to execute an offline supervised training program during the factory pre-installation and configuration phase. The reference samples used during model training are derived from a legally acquired large-scale clinical multi-channel physiological waveform validation set, and the corresponding observational physical label benchmarks are established as follows:

[0074] By synchronously acquiring middle ear pressure troughs and impedance abrupt changes using a tympanic acoustic impedance meter or an external auditory canal micro-pressure sensor, the objective physical indicators of whether the Eustachian tube is truly patent are independently calibrated and confirmed. In the multi-round iterative optimization process of the hidden layer node weight parameters of the internal model, the system introduces a cross-entropy loss function to quantitatively measure the distribution divergence between the positive prediction output probability and the actual physical state verification label. The mathematical expression bounds of this network loss iteration logic are as follows: ; In the formula, This represents the total cross-entropy loss of the entire network structure topology after a single batch calculation of forward and backward gradient propagation; This represents the total number of physiological feature samples extracted in batches during the forward propagation of a single network parameter update. To load the first data set Each sample corresponds to a real medical expert's confirmation label (here, 1 indicates that the objective physiological optimal ventilation window has appeared, and 0 indicates that the window has not appeared). This represents the output of the dynamic decoding of the neural network unit and corresponds to the first... The probability of a sample being quantitatively predicted.

[0075] The entire optimization process employs an adaptive moment estimation network optimization scheme to specifically correct the gradient shift of the convolution kernel weights and the coefficients inside the memory cells until the overall cross-model validation loss converges and remains within the preset global minimum confidence domain for multiple consecutive generations, ultimately forming the non-volatile solidification of the weights.

[0076] When entering the clinical auxiliary judgment stage of actual ventilation after the model is solidified, the command issuing unit first performs basic screening of the envelope signal and pressure derivative sequence based on the aforementioned dynamic trigger threshold. Under the premise that the initial screening determines that the action has occurred, the predicted probability of occurrence is actively read at high frequency. If the recorded predicted probability of occurrence shows an upward trend with the calculation time axis and exceeds the action trigger threshold solidified in the system, the command issuing unit then determines that the user's optimal ventilation intervention window has been successfully captured.

[0077] The aforementioned action trigger threshold for error detection is usually limited to a range of 0.85 to 0.95. The specific actual calibration value can be obtained in the baseline self-test program executed during a single power-on. It absorbs and performs adaptive dynamic level offset compensation based on the user's current resting state signal floor noise intensity. As long as the ventilation trigger requirements are matched, the command issuing unit will then transmit a pulse width modulation action command mapped to the force ratio parameter to the solenoid valve module through the interrupt direct connection pin, transiently driving the physical connection of the entire positive pressure air passage.

[0078] Meanwhile, in order to test and prevent uncontrollable operational risks caused by the lack of underlying decision algorithm logic, and to avoid the deep network evaluation falling into an infinite waiting logical calculation dead zone due to extremely weak sensor detection signals in individual testers due to pathological reasons, the main control module internal circuit is additionally equipped with a hardware-level foolproof timeout watchdog timing unit.

[0079] Assuming that the time range of the single-cycle monitoring action is limited to 10s to 20s, and the predicted probability generated by the algorithm remains suppressed at a low level due to external factors and fails to cross the set gating trigger threshold, the watchdog timer unit will release a hard interrupt to interrupt the current analysis instruction cycle and force the system into a safe sleep state.

[0080] This control mechanism then sends high-priority abnormal alarm signals to the external interactive terminal, aiming to guide operators to check the contact resistance reference of the surface sensing electrodes, or to guide and allow clinicians to manually switch to the safety-limited protection mode. For the copper pouring rules of the edge power supply of the embedded main control chip running inside the main control module, and the wiring theory of the core crystal oscillator clock tree frequency multiplication phase-locked loop, those skilled in the art can implement them according to the standard design specifications released by the supplier. The multi-task system scheduling stack partitioning and electromagnetic shielding immunity design strategies covered here are all well-known technologies in this field, and will not be described in detail here.

[0081] Please see the appendix Figure 6 , Figure 6 This is a schematic diagram of the control architecture of the positive pressure air supply layer according to an embodiment of the present invention. In this embodiment, the positive pressure air supply layer control architecture of the system serves as the hardware foundation for performing the air pressure blowing intervention task. It is mainly constructed by an air source module, an air storage module, and a solenoid valve module. The air source module is responsible for drawing and compressing air from the external environment to continuously supply a stable airflow with a target pressure to the subsequent air storage module. When the system is in the real-time monitoring stage of multi-source physiological signals, the air storage module has a buffer cavity with a fixed clearance for pre-stocking compressed high-pressure fluid.

[0082] As a preferred approach, this pre-pressurized energy storage mechanism effectively avoids the problem that the ventilation response is limited by the physical delay of the underlying air pump pressurization, thereby ensuring the transient response of the fluid at high flow rates when the main control module issues control commands.

[0083] To ensure that the actual airflow pressure of the fluid stored inside the gas storage module remains stable within the set standard threshold range, the system is equipped with digital incremental feedback regulation logic in the main control module to achieve closed-loop control of the output power of the gas source module.

[0084] Specifically, in terms of fluid pressure regulation principle, continuous unidirectional pumping will cause the pressure inside the gas storage module to rise continuously and even exceed the limit. The system needs to dynamically adjust the motor speed according to the difference between the real-time pressure and the target pressure: when the actual pressure is close to the target limit, the pumping frequency is reduced to prevent over-inflation from causing damage to the gas circuit components, and rapid gas replenishment is provided when the pressure leaks or drops. Based on this, a closed-loop feedback unit for monitoring the pumping status is added inside the gas source module.

[0085] The main control module continuously reads the real-time feedback pressure data in the buffer chamber along the time axis through the closed-loop feedback unit, compares it with the target predetermined reference pressure value set at the bottom layer, and then dynamically adjusts the pulse width modulation duty cycle of the motor in the air source module drive unit. The above-mentioned mathematical increment calculation process for dynamic adjustment of duty cycle output is based on the following formula for bounded constraints: ; In the formula, For the solenoid valve module drive control link in the first stage The pulse width modulation duty cycle increment for each discrete time period; Indicates that it is currently in the first stage. The absolute deviation between the target predetermined reference pressure value and the real pressure value recovered in real time within each calculation cycle; and They respectively refer to the first one immediately preceding the first one. and the Historical data of pressure deviation retained in each discrete cycle at the bottom layer; The proportional control gain coefficient for the feedback link is used to respond instantly to the current transient pressure deviation of the ventilation network, and its value ranges from 0.5 to 1.2. The integral control gain coefficient of the feedback link is used to dynamically accumulate and eliminate the steady-state static error caused by pipeline gas source leakage, and its value ranges from 0.01 to 0.05. The differential control gain coefficient of the feedback link is used to sense the pressure rise slope in advance to suppress transient aerodynamic overshoot in the early stage of pressure application, and its value ranges from 0.1 to 0.3.

[0086] The specific reference values ​​of the above three gain coefficients are initialized and tuned by the critical proportional method before the system leaves the factory, and parameter compensation is supported according to pipeline aging damping in subsequent operating cycles. The main control module sums and updates the calculated duty cycle increment value with the actual state value in the current motor drive register, so that the internal air pressure of the gas storage module is dynamically shifted and maintained within the set pressure balance range, accumulating kinetic energy for subsequent positive pressure inflation and release.

[0087] When the neural network running at the bottom layer of the main control module determines that the physiologically optimal ventilation window is open, the high-pressure fluid stored in the gas storage module will be directed to the inside of the user's nasal cavity through the solenoid valve module. In the specific air duct layout, the solenoid valve module acts as the control node for connecting and blocking the airflow. Its high-pressure input end is connected to the gas storage module through an airtight flange, while its output end is connected to the air duct unit inside the nasal end execution module.

[0088] Upon receiving the action pulse trigger level from the interrupt pin of the main control module and amplified by cascade, the excitation coil inside the solenoid valve module is electromagneticated within milliseconds. The resulting electromagnetic attraction directly overcomes the mechanical reset reaction force of the internal limit spring, causing the movable iron core to produce axial displacement. This causes the fluid channel, which was originally in a normally closed isolated state, to open instantly. The accumulated high-pressure gas then flows into the user's nasal cavity along the cleared pipe network. The total capacity of a single airflow injection and the amount of work done by the fluid are determined by the pulse width time during which the solenoid valve module is continuously configured to be in the conducting state.

[0089] To address potential uncontrollable sudden changes in the gas supply network and avoid logical dead zones caused by abnormal stagnation of the underlying control algorithm, the system deeply integrates a hardware-software collaborative safety blocking mechanism into its execution architecture. While the solenoid valve module receives instructions and is in a long-term gas release cycle, the main control module uses time-division multiplexing technology to read in real time the actual dynamic pressure of the measuring point returned by the pressure sensing module.

[0090] Once a local airflow pressure increase is detected and exceeds the preset inflation safety upper limit threshold, the highest priority hardware interrupt logic configured in the main control module will be activated immediately and take over the underlying bus, forcibly pulling down and clearing the drive duty cycle level of the solenoid valve coil. This logic causes the solenoid valve to be de-energized and passively reset by relying on the internal mechanical spring, quickly closing the latch and cutting off the main air path.

[0091] As a preferred approach, the threshold value for determining the upper limit of inflation safety is set within the range of 5 kPa to 6 kPa, thereby effectively blocking the path of abnormal high-pressure airflow that could cause barotrauma to the fragile tympanic membrane when the microcontroller software or hardware malfunctions or the user's airway is accidentally blocked by foreign objects.

[0092] As a further supplement to the purely physical hardware redundancy defense, the system has a mechanical pressure exhaust unit on the rigid sidewall shell of the gas storage module that does not rely on electrical drive. In the event of equipment failure due to multiple extreme and complex faults such as severe short circuit in the system power supply line and main control logic crash, if the outward mechanical thrust generated by the uncontrolled gas pressure inside the gas storage module exceeds the physical elastic deformation limit resistance point set by the calibration spring inside the exhaust unit, the micro valve of the mechanical pressure exhaust unit will be forcibly opened, and redundant high-pressure fluid will be directly discharged into the outer space through the passive pressure relief hole.

[0093] This completely decoupled passive load relief method eliminates the risk of high-pressure bursts caused by system electrical control malfunctions and imbalances from the root of physical and mechanical design. For the micro oil-free vacuum pump diaphragm linkage mechanical components used in the gas source module, and the pilot diaphragm fatigue-resistant elastic composite material used in the solenoid valve module, those skilled in the art can refer to the design specifications of general minimally invasive medical pneumatic actuators for selection and replacement. The threaded damping sealing structure and shock-absorbing pad assembly process of the related polyurethane medical gas delivery tube are well known technologies in this field and will not be described in detail here.

[0094] Please see the appendix Figure 7 , Figure 7 This is a schematic diagram of the operating architecture of the human-computer interaction module and the power management module according to an embodiment of the present invention. In this embodiment, in order to ensure the real-time visualization of multi-source physiological signals and the independence of the system's offline operation, the main control module is externally connected to the human-computer interaction module and the power management module. The human-computer interaction module is used as an interactive interface for medical staff and users to monitor the underlying operating status of the system. Its internal components are further divided into a touch display unit and a status alarm unit.

[0095] Based on the data stream reported in real time by the main control module through the internal system bus, the touch display unit synchronously plots the pharyngeal electromyography timing waveform acquired by the electromyography acquisition module and the continuous curve of dynamic pressure change inside the nasal cavity continuously acquired by the pressure sensing module on its display panel. This synchronous display mechanism of multi-source physiological parameters provides the system operator with a visual reference for intuitively judging the changes in the user's current physiological characteristics.

[0096] In order to make the inference results of the deep network architecture inside the main control module explicit, the touch display unit synchronously reads and updates the numerical information of the predicted probability of occurrence in real time. When the system detects that the predicted probability of occurrence is close to the preset action trigger threshold in the main control module register, the touch display unit will adaptively adjust the backlight brightness and background color of the interface to prompt the operator that the system is about to automatically switch to the positive pressure venting stage.

[0097] To address the physical background noise interference present in the clinical intervention environment, the status alarm unit integrates adaptive volume adjustment control logic based on audio feedback. Specifically, in terms of working principle, a fixed-level warning volume is easily masked by ambient noise in a noisy environment, thus losing its warning effect, while in an extremely quiet ward, it can cause psychological fright to the user. Therefore, the system needs to dynamically adjust the output warning sound pressure according to the intensity of the ambient background noise.

[0098] Based on this, the status alarm unit uses the built-in sound acquisition circuit to continuously extract the background noise level in decibels of the current space, and is configured with an RC hardware filter circuit to filter out transient spike noise interference. When the system experiences a complex abnormal state such as internal pipeline air pressure exceeding the limit causing forced airflow blockage, or the bottom sensor electrode falling off causing physical communication disconnection, the status alarm unit will immediately drive the external piezoelectric buzzer to output a warning audio.

[0099] In the audio trigger output stage, as a preferred method, the output alarm volume is not rigidly defined to a constant level. Instead, it follows the calculated background noise level in decibels and performs smooth dynamic audio gain compensation. This ensures that the actual alarm sound pressure effectively covers the normal background noise while avoiding excessive transient roar that could cause a sense of urgency to the human body.

[0100] The hardware carrier that maintains the independent operation of the entire system and is independent of the fixed power supply network is the power management module. Its internal structure includes a large-capacity lithium battery pack and a power status monitoring unit. When the miniature vacuum-driven air pump configured in the system's front flow path is frequently started and stopped, it will cause a wide transient load surge on the power supply network, resulting in a sharp drop and bounce in the terminal voltage of the battery cells.

[0101] If the core charge of a battery cell is estimated by relying solely on a conventional and single physical voltage lookup table algorithm, false low-charge alarms are very likely to occur. In some cases, the main control system may crash and reset due to transient undervoltage at critical ventilation nodes where the gas path is open, directly triggering the hardware low-voltage lockout mechanism. To build an electrical safety defense line at the underlying hardware level and eliminate logic malfunctions and decision dead zones caused by this, the power status monitoring unit incorporates power prediction control logic that integrates ampere-hour integral parameters and ambient temperature compensation mechanisms.

[0102] The operating mechanism of this logic is to continuously integrate the instantaneous load change current in the battery discharge circuit through high-frequency sampling, and combine it with the reference temperature data of the current cell surface obtained by an external thermistor, thereby accurately calculating the current effective absolute charge of the system, and finally converting it into the remaining number of complete ventilation intervention cycles that the system can safely maintain with its current energy. The underlying mathematical operation formula of the above multi-parameter fusion calculation is defined as follows: ; In the formula, The number of positive pressure blowing cycles that the system's current available battery capacity can fully support; The actual remaining value of the system's available power, obtained by the power management module through the underlying coulomb integral verification. This indicates the minimum limit of the forced hibernation capacity preset by the system in the read-only memory area to ensure that basic processes are not lost during operation; This indicates the average time span of the entire process of a single gas source intervention and pressure holding work, calculated and extracted by the power status monitoring unit based on historical operation log feature data. This represents the sequence of transient discharge currents required by the entire system to execute the external electrical loads within the aforementioned integral time span. This indicates the monitoring feedback temperature relative to the current cell casing. Matching discharge efficiency loss factor.

[0103] Based on the aforementioned comprehensive closed-loop charge measurement and evaluation mechanism, when the charge status monitoring unit detects the number of cycles supported by the remaining capacity... As the discharge process continues to decline and touches the minimum safety tolerance threshold of the system, the safety program embedded in the power management module is activated. This program injects the highest priority interception command into the physical communication bus of the main control module through the hard interrupt pin, actively locks the long-term permission configuration of the solenoid valve module, forcibly blocks the start of a new round of ventilation cycle, and switches the entire system into a low-power protection standby state.

[0104] This safety interlocking mechanism preemptively eliminates the hardware hazard of mechanical jamming of the solenoid valve's moving iron core due to insufficient excitation electromagnetic force when the driving voltage drops sharply due to the increase in the internal resistance of the battery cell. This completely prevents patients from being forced to be exposed to continuous high voltage risks for a long time. For the micro boost-buck chopper topology circuit layout used in the power management module and the anti-static multi-point capacitor layer bonding process used in the LCD display panel of the human-machine interaction module, those skilled in the art can refer to the safety regulations for general medical electrical equipment to select and adapt the relevant capacitor and inductor components. The related communication isolation and anti-interference processing scheme and the touch panel anti-side light leakage structure design are well known technologies in this field and will not be described in detail here.

[0105] Please see the appendix Figure 8 , Figure 8 This is a schematic diagram of a communication and storage module and a related closed-loop evaluation architecture according to an embodiment of the present invention. In this embodiment, to achieve long-term traceability of user intervention records and objective evaluation of the effectiveness of system intervention, the communication interface of the main control module is connected to the communication and storage module. The main control module serves as a data relay and analysis layer between the underlying physical airflow intervention action and the external communication network, and internally includes a therapeutic effect evaluation unit, a local storage unit, and a wireless transceiver unit.

[0106] Most existing positive pressure ventilation solutions can only control the constant output of the air source. They usually cannot quantify whether the applied high-pressure airflow has actually opened the user's Eustachian tube and achieved effective ventilation. To make up for the lack of this closed-loop data, the system uses a efficacy assessment unit to quantitatively calibrate the effect of a single ventilation. Specifically, in terms of physical assessment principle, when the user's Eustachian tube is successfully opened by high-pressure fluid during the set optimal ventilation window, the originally closed and restricted posterior nasal airway will be affected by the transient expansion of local volume, resulting in a characteristic pressure drop trough in the air supply line.

[0107] After the system completes a single positive pressure deflation operation, the efficacy assessment unit immediately reads the dynamic decay sequence of nasal pressure continuously collected by the pressure sensing module through the internal communication bus, and performs numerical calculations on the true physiological effectiveness of this air pressure intervention by extracting the physical characteristics of the specific air pressure trough.

[0108] To objectively quantify the aforementioned pressure change characteristics, the efficacy assessment unit employs a comprehensive scoring logic that combines the integral of the pressure waveform with the extreme values ​​of the pressure change rate. The time-domain integral of the pressure waveform can fully reflect the total work done by the pressure applied to the human body within a single inflation time window; the differential extreme values ​​of the pressure can pinpoint and amplify the signal of the abrupt pressure drop rate generated at the moment the Eustachian tube opens. The mathematical definitions of the relevant efficacy scoring parameters and their constraint boundary calculation formulas are as follows: ; In the formula, This represents the comprehensive score parameter of intervention efficacy calculated by the system based on a single positive pressure blow to the user; and These represent the physical moment when the bottom solenoid valve module is triggered and turned on by a high level, and the moment when the air passage is locked and the internal air pressure decays and stabilizes again. In the designated to Within the intervention time window, the micro-pressure sensor recorded the continuous measured dynamic air pressure curves within the nasal passage. To obtain and solidify the user's stable resting nasal baseline atmospheric pressure value before the positive pressure pulse is executed; The area-limited integral weighting coefficient, set for empirical purposes, characterizes the cumulative correlation between total fluid intake during the inflation period and the maintenance of Eustachian tube dilation, with a value ranging from 0.6 to 0.8. The value is a weighting coefficient for the microdispersion characteristic, used to amplify the extreme pressure drop rate signal caused by the local airway opening due to the effective ventilation transient. The value ranges from 0.2 to 0.4.

[0109] After calculating the specific evaluation value based on the above formula, the system does not set it aside as hidden background data, but instead compares it logically with the lower limit of the intervention efficacy verification threshold fixed in the non-volatile register segment.

[0110] As a preferred approach, the upper and lower limits of the efficacy verification threshold are usually defined as the range of 60% to 75% of the peak value of the effective swallowing measurement in a normal healthy population. If the evaluation module confirms that the comprehensive score parameter of the intervention efficacy in the current cycle successfully exceeds the verification threshold, the efficacy assessment unit will attach an efficacy label indicating physical connectivity to the physiological waveform data packet sampled in this session.

[0111] Subsequently, the multi-source data stream carrying the evaluation label is transmitted to the internal local storage unit. In response to the problem that high-frequency continuous sampling by multi-source sensors can easily cause internal memory space overflow or overwrite loss, the local storage unit abandons the indiscriminate fixed-frequency disk writing mechanism and embeds a state-aware dynamic downsampling storage logic.

[0112] When it is detected that the user is in a resting state for a long time and no obvious swallowing electromyographic signals are generated, the downsampling compression mapping of the data storage link is performed.

[0113] Once the system detects a potential swallowing signal or enters the positive pressure release execution phase, the local storage unit forcibly removes the sampling rate limit and resumes the storage and input of high-frequency lossless original data.

[0114] After compressing, reducing the dimensionality of, and solidifying the validity of local data, the wireless transceiver unit reads the packaged physiological characteristic data packets through the system's serial bus port and sends them to the cloud data center or medical staff terminal using a radio frequency modulation protocol. To cope with harsh communication conditions such as electromagnetic shielding from thick walls in wards or radio frequency channel fluctuations during movement, the wireless transceiver unit is equipped with a communication guarantee mechanism that supports breakpoint resumption. This mechanism allows data packets that are blocked by network transmission to be retained in a local static cache pool for a range of 48 to 72 hours.

[0115] During this period, the wireless transceiver unit will maintain low-power network frequency listening until the base station handshakes and confirms that the network connection has been re-established. Then, it will automatically upload the remaining backlog queue at full speed to prevent the loss of medical records caused by network blind spots from the root of data link interaction. As for the low-power radio frequency antenna impedance-capacitance matching structure used inside the wireless transceiver unit, those skilled in the art can select and configure it in accordance with the design specifications of conventional wireless pass-through frequency bands. The antenna anti-static breakdown shielding design associated with its circuit pins and the baseband differential clock calibration mechanism are well known technologies in the field and will not be described in detail here.

[0116] To further disclose the technical means and execution logic of the present invention, the following detailed description of the specific implementation process of this system within a complete ventilation intervention cycle is provided, using a case of a user suffering from Eustachian tube dysfunction as an example:

[0117] 1. The operator places the electromyography (EMG) acquisition module on the exposed skin of the user's neck (the measuring electrode is located in the submental muscle group, and the reference electrode is located at the clavicle end), and pushes the nasal end execution module with a sealing adapter into one side of the user's anterior nostril. Relying on the plastic deformation of the external covering material, an airtight seal is formed with the inner wall of the nasal cavity. After the system is started, the main control module first drives the air pump module to inject air into the air storage module. By monitoring the feedback pressure in real time and adjusting the pulse width modulation duty cycle of the air pump, a basic air storage pressure of 5 kPa is established and maintained in the air storage module.

[0118] Subsequently, the system initiates the baseline self-calibration procedure. The user performs a single empty swallowing action, and the electromyography acquisition module acquires the reference signal of the calibration period of this action cycle. The main control module extracts the envelope features of the signal, calculates the reference peak value and the reference integral area, and generates the current user's dynamic trigger threshold based on the reference peak value (the value range is 0.88 in this embodiment). The threshold value and the reference integral area are stored in the internal memory as the benchmark data for subsequent judgment.

[0119] 2. The system enters the real-time monitoring stage. The electromyography (EMG) acquisition module and the pressure sensing module begin to synchronously acquire real-time raw EMG signals and nasal air pressure fluctuation signals at high frequency. The main control module rectifies and filters the EMG signals to generate envelope signals and differentiates the air pressure fluctuation signals to generate a first derivative sequence.

[0120] When the user makes a real swallowing motion, the main control module detects that the envelope signal is greater than the calibrated dynamic trigger threshold, and at the same time, the nasal air pressure fluctuation signal and the first derivative sequence are both greater than or equal to zero (physically indicating that the airway is currently in the non-inspiratory phase). The system determines that it is currently in an effective swallowing cycle and uses this as a candidate time window to simultaneously wake up the internal deep prediction model. This model integrates one-dimensional convolution and long short-term memory networks to perform local feature extraction and long-term dependency calculation on the synchronized electromyography and air pressure two-dimensional input matrices.

[0121] In this embodiment, at 210 milliseconds after the start of the swallowing action, the predicted probability of occurrence output by the model reaches 0.91, which is greater than the preset action trigger threshold. The main control module thus accurately locks the current time as the optimal ventilation time when the Eustachian tube is about to open, and extracts the start and end time nodes of the effective swallowing cycle.

[0122] 3. Based on the above start and end time nodes, the main control module extracts the real-time action integral area within the time period, and calculates the ratio of its real-time action integral area to the aforementioned reference integral area to obtain the current force ratio parameter (the calculated value in this embodiment is 1.15).

[0123] The main control module then uses an internal mapping function to convert the force ratio parameter into the target work required for this ventilation, and calculates the target conduction pulse width time required for the solenoid valve module to achieve this work (calculated as 120 milliseconds in this embodiment).

[0124] Before outputting the command, the system further verifies whether the estimated dynamic instantaneous cavity pressure corresponding to the pulse width exceeds the safety limit of 6 kPa. After confirming that there is no safety hazard, the main control module outputs a level signal to control the solenoid valve module to open instantly for 120 milliseconds. The pressurized gas pre-stored in the gas storage module is released quantitatively according to the target conduction pulse width time and enters the user's nasal cavity through the internal air guiding unit. During this inflation period, the pressure sensing module monitors and records in real time that the highest impact air pressure in the nasal cavity is 4.6 kPa, and the safety interruption logic is not triggered.

[0125] 4. After a certain amount of high-pressure airflow is released, the user's Eustachian tube is successfully physically opened at the optimal opening time. Due to the instantaneous expansion of the local volume of the posterior nasal airway, the pressure sensing module located in the bypass blind cavity collects a characteristic pressure drop trough in the actual nasal air pressure (the measured drop amplitude is 1.8 kPa).

[0126] After a single inflator procedure is completed, the efficacy assessment unit extracts the air pressure data within that time window. By weighting and summing the total pressure time integral value and the differential extreme value of the pressure drop characteristic within the ventilation time window according to preset empirical weights, the system finally obtains a comprehensive score of 82 points for the efficacy of this intervention. The system compares this score with the set effectiveness threshold (65 points) to determine that the current ventilation is effective.

[0127] The communication and storage module then attaches a valid objective tag to the multi-source data packets of this cycle, performs downsampling and compression, stores them in the local storage unit, and establishes an radio frequency transmission link based on the wireless transceiver unit to upload the data to the background system for long-term medical traceability. The single work cycle ends here, and the air pump module automatically replenishes the air storage module to prepare for the next action trigger.

[0128] To further verify the effectiveness and safety of the control system provided by this invention in practical applications, a clinical parameter verification experiment was conducted on sixty users diagnosed with mild to moderate eustachian tube dysfunction.

[0129] 1. Experimental grouping and conditions: Sixty users were randomly and equally divided into two groups: Control group (30 people): A traditional fixed positive pressure continuous blowing device was used. The airflow output parameter of the device was set to a constant 4 kPa. The intervention was triggered by the user manually pressing a button and coordinating with swallowing.

[0130] Experimental group (30 people): The swallowing-triggered Eustachian tube inflation control system based on signal recognition provided by this invention was used. The timing of intervention was automatically determined by the internal depth prediction model, and the conduction pulse width and work volume were adaptively calculated and output according to the electromyographic force characteristics.

[0131] 2. Results and Statistical Analysis: During the experiment, the objective patency status and force data of the Eustachian tubes of both groups of users were simultaneously recorded using an external middle ear micro-pressure monitoring device. The comparative statistical results of various core evaluation indicators are shown in Table 1 below:

[0132] Table 1: Comparison of test results between the control group and the experimental group: Evaluation dimensions and test indicators control group experimental group The initial intervention had a high success rate. 43.3%（13 / 30） 86.7%（26 / 30） Mean effective ventilation response delay time 850±120 milliseconds 210±45 milliseconds Intranasal cavity measured peak instantaneous impact pressure 5.8 ± 0.6 kPa 4.4 ± 0.3 kPa User Subjective Discomfort Rating 6.2±1.5 2.1±0.8 Comprehensive score of intervention efficacy for barometric trough extraction 48.5±12.3 81.2±9.4

[0133] From Table 1 and Appendix Figure 9 -Appendix Figure 11 We can obtain: The control system disclosed in this invention exhibits significant advantages in all performance indicators: Firstly, in terms of target hit rate, this system automatically identifies the optimal open window by collecting swallowing electromyographic signals and differential air pressure, combined with a depth prediction model, thus avoiding ineffective air delivery during the closure period caused by asynchronous operation in the traditional manual button mode. The success rate of initial intervention was 43.4 percent higher than that of the control group, and the effective ventilation response delay time was shortened by about 75 percent.

[0134] Secondly, in terms of safety and compliance, traditional equipment uses a fixed air supply mode, which can easily cause airflow obstruction and local pressure overshoot when the Eustachian tube is not open; the present invention uses closed-loop force ratio calculation and adaptive pulse width limitation to realize the quantitative work of airflow pulse.

[0135] The experimental results showed that the instantaneous peak impact air pressure in the nasal cavity of the experimental group was reduced by about 24 percent compared with that of the control group. While ensuring effective physical penetration of tissues (significantly higher overall efficacy score), it avoided the potential risk of damage to the tympanic membrane caused by abnormal high pressure, thereby greatly reducing the subjective discomfort of users during the intervention process.

Claims

1. A swallowing-triggered Eustachian tube inflation control system based on signal recognition, characterized in that the system include: An electromyography (EMG) acquisition module, placed on the user's skin surface, is used to acquire electrophysiological signals from the swallowing muscles. The nasal end execution module, the outer surface of which is used to physically fit with the user's anterior nasal nostrils to form an airtight seal, and to introduce positive pressure airflow into the user's nasal cavity; A pressure sensing module is installed in the air passage inside the nasal end execution module to collect the air pressure fluctuation signal inside the nasal cavity and the actual pressure data inside the nasal cavity in real time. Gas storage module, used to store compressed gas; An air pump module, the output end of which is connected to the input end of the air storage module, is used to inject air into the air storage module; A solenoid valve module, wherein the input end of the solenoid valve module is connected to the output end of the gas storage module, and the output end is connected to the airflow input end of the nose actuator module; The main control module receives the electrophysiological signals of the swallowing muscles and the air pressure fluctuation signals inside the nasal cavity, confirms whether it is an effective swallowing cycle, calculates the optimal ventilation timing, and then calculates the target conduction pulse width time to match the optimal ventilation timing. The main control module outputs a level signal to control the solenoid valve module to open according to the target conduction pulse width time, and releases the compressed gas in the gas storage module into the user's nasal cavity in a quantitative manner.

2. The swallowing-triggered Eustachian tube inflation control system based on signal recognition according to claim 1, characterized in that, The main control module is configured to execute a baseline self-calibration program and a logic determination program: During the baseline self-calibration procedure, the main control module extracts the envelope features of the calibration period reference signal obtained during a single empty swallowing action of the user, calculates the reference peak value and the reference integral area, and generates a dynamic trigger threshold based on the reference peak value. When the system enters the real-time monitoring phase, the main control module performs rectification and filtering on the real-time raw electromyography signal to generate an envelope signal, and performs differentiation on the nasal cavity internal air pressure fluctuation signal to generate a first derivative sequence. When the main control module determines that the envelope signal is greater than or equal to the dynamic trigger threshold, and the nasal cavity internal air pressure fluctuation signal and the first derivative sequence are both greater than or equal to zero, the main control module confirms that the current period is an effective swallowing cycle.

3. The swallowing-triggered Eustachian tube inflation control system based on signal recognition according to claim 2, characterized in that, The main control module also deploys a deep prediction model based on a fusion of one-dimensional convolution and long short-term memory structures. The main control module will use the time window that confirms the effective swallowing cycle as a candidate time window to wake up the depth prediction model; when the main control module determines that the probability of the prediction output by the depth prediction model is greater than the preset action trigger threshold, it will extract the start and end time nodes of the effective swallowing cycle. The main control module extracts the real-time action integral area within the start and end time nodes, calculates the force ratio parameter based on the ratio of the real-time action integral area to the reference integral area, and converts the force ratio parameter into the target work amount to calculate the target conduction pulse width time corresponding to the target work amount.

4. The swallowing-triggered Eustachian tube inflation control system based on signal recognition according to claim 1, characterized in that, The nasal end actuator module is equipped with a housing unit, a sealing adapter unit, and an internal air guiding unit: The outer contour of the outer shell unit is a radially tapered frustum, and the outer surface of the outer shell unit is fixed with the elastic sealing adapter unit. The outer shell unit has an internal air guiding unit that runs through both ends. The hollow part of the side wall of the internal air guiding unit extends to form a bypass blind cavity that communicates with the main air passage. The other end of the bypass blind cavity is closed and isolated. The pressure sensing module is independently set in the bypass blind cavity to collect the air pressure fluctuation signal inside the nasal cavity without direct physical contact with the user.

5. The swallowing-triggered Eustachian tube inflation control system based on signal recognition according to claim 1, characterized in that, The electromyography (EMG) acquisition module is equipped with a surface electrode unit, a signal conditioning unit, an analog-to-digital conversion unit, and a data transmission unit. The surface electrode unit includes a measuring electrode and a reference electrode, and the electrophysiological signal of the swallowing muscle group is obtained by extracting the potential difference between the two. The signal conditioning unit is equipped with a differential amplifier circuit and a cascaded filter network, which are used to perform differential amplification, fixed-band filtering and notch filtering on the acquired signal in sequence to generate a continuous analog signal. The analog-to-digital conversion unit is used to discretize the continuous analog signal into a digital sequence, which is then packaged by the data transmission unit and sent to the main control module via a serial communication bus.

6. The swallowing-triggered Eustachian tube inflation control system based on signal recognition according to claim 1, characterized in that, The system incorporates a security blocking mechanism in its execution architecture: During the process of the solenoid valve module opening to release gas, the main control module is configured to use time-division multiplexing technology to instantly read the actual intranasal pressure data collected and generated by the pressure sensing module. When the main control module determines that the target conduction pulse width time has ended, or detects that the actual pressure data in the nasal cavity triggers the preset safety upper limit, the main control module outputs an interrupt command through hardware interrupt logic to pull down the drive level and control the solenoid valve module to shut down and close.

7. The swallowing-triggered Eustachian tube inflation control system based on signal recognition according to claim 1, characterized in that, A closed-loop regulation mechanism and a passive safety structure are configured between the air pump module and the air storage module. The main control module monitors the feedback gas pressure data in the gas storage module in real time. Based on the difference between the real-time pressure and the target pressure, it dynamically adjusts the pulse width modulation duty cycle of the motor in the gas pump module drive unit in combination with the proportional-integral-derivative-increment algorithm to maintain the basic gas storage pressure value. The gas storage module is equipped with a mechanical pressure exhaust unit on its outer shell. When the internal gas pressure of the gas storage module exceeds the preset physical elastic deformation limit resistance point, the mechanical pressure exhaust unit is passively opened to perform pure physical-level pressure relief and exhaust.

8. The swallowing-triggered Eustachian tube inflation control system based on signal recognition according to claim 1, characterized in that, The main control module is externally connected to a communication and storage module, and the communication and storage module is internally configured with a treatment efficacy evaluation unit. After the system performs a single positive pressure deflation operation, the efficacy assessment unit reads the dynamic decay sequence of nasal pressure within the ventilation time window continuously collected by the pressure sensing module. The efficacy assessment unit extracts the characteristics of the trough of the sudden drop in air pressure, and calculates a comprehensive scoring parameter by weighting the integral value of the pressure waveform in the time domain with the extreme pressure drop rate signal according to a preset weighting coefficient. If the comprehensive scoring parameter is greater than the lower limit of the verification threshold, an effectiveness label indicating physical connection is added to the physiological waveform data of this intervention.

9. The swallowing-triggered Eustachian tube inflation control system based on signal recognition according to claim 1, characterized in that, The system is equipped with a power management module that provides independent power supply, and the power management module contains a power status monitoring unit: The power status monitoring unit calculates the remaining number of complete ventilation intervention cycles that can be safely maintained based on the integral of the load change current in the battery discharge circuit and the ambient temperature data fed back by the external thermistor. When the number of remaining ventilation intervention cycles is lower than the preset minimum threshold, the power management module triggers a hard interrupt interception command to actively lock the opening permission of the solenoid valve module.

10. The swallowing-triggered Eustachian tube inflation control system based on signal recognition according to claim 1, characterized in that, The main control module is externally connected to a human-machine interaction module, which is equipped with a status alarm unit. The status alarm unit integrates adaptive volume adjustment control logic based on audio feedback. The status alarm unit uses a built-in sound acquisition circuit to continuously extract the background noise level in the current space. When the system is in a warning state and drives an external buzzer to output a warning audio, the audio gain of the warning audio is controlled to follow the background noise level and perform dynamic sound pressure compensation adjustment.

Images