Broadband middle ear impedance and otoacoustic emission combined hearing threshold prediction method and system

Abstract

The invention discloses a broadband middle ear impedance and otoacoustic emission combined hearing threshold prediction method and a broadband middle ear impedance and otoacoustic emission combined hearing threshold prediction system. The method comprises the following steps: determining a detection mode and initializing detection parameters; auditory meatus sealing and environmental noise self-inspection, broadband middle ear impedance detection and otoacoustic emission detection are sequentially executed, and broadband middle ear impedance related features and otoacoustic emission related features are extracted respectively; fusing the two types of features to construct a joint feature, inputting the joint feature into a hearing prediction model, outputting a hearing threshold prediction value or a hearing condition evaluation result, and generating a detection report; according to the system, a modular architecture of a core control and acquisition unit and an air pressure adjusting unit is adopted, and core hardware is reused to realize one-stop detection of middle ear and inner ear functions; according to the method, the middle ear conduction characteristic and the inner ear active function are jointly modeled, key links of an auditory pathway are comprehensively covered, the accuracy of hearing threshold prediction is improved, the detection process is automatic, subjective cooperation of a subject is not needed, and the clinical adaptability is high.

Description

Technical Field

[0001] This invention relates to the field of objective detection of the auditory system and hearing assessment technology, specifically to a detection system and method that combines broadband middle ear impedance detection technology with otoacoustic emission detection technology, and predicts hearing thresholds and assesses hearing status based on the combined features. Background Technology

[0002] Pure-tone audiometry is currently the most widely used hearing test method in clinical practice. It determines hearing threshold levels by having the subject provide subjective feedback to pure-tone stimuli of different frequencies and intensities. However, this method is highly dependent on the subject's subjective cooperation and is easily affected by factors such as attention, comprehension ability, reaction time, and psychological state. Therefore, it is not suitable for infants, the elderly, patients with impaired consciousness, or those who cannot effectively cooperate with the test.

[0003] To overcome the limitations of subjective hearing testing, objective hearing testing techniques have been gradually developed. Among them, otoacoustic emission (OAE) testing technology can objectively reflect the functional status of the outer hair cells of the inner ear by recording acoustic signals generated by the active activity of the outer hair cells of the cochlea in the external auditory canal. It has been widely used in newborn hearing screening and some clinical applications. However, the otoacoustic emission signal needs to pass through the middle ear conduction system to travel from the inner ear to the external auditory canal, and its amplitude and signal-to-noise ratio are easily affected by changes in the middle ear condition. For example, when there is decreased tympanic membrane compliance, restricted ossicular chain movement, negative pressure in the ear canal, or middle ear effusion, even if the inner ear function is normal, the otoacoustic emission signal may be significantly attenuated or unable to be elicited, resulting in false negative results.

[0004] On the other hand, middle ear function testing technologies, especially wideband middle ear impedance testing (WBT or Wideband Admittance / Absorbance, WBA), can assess the middle ear's sound energy transmission characteristics over a wide frequency range, reflecting the energy absorption, reflection, and resonance characteristics of the middle ear system, which is of great significance for identifying middle ear conduction abnormalities. However, wideband middle ear impedance testing mainly reflects the mechanical conduction state of the middle ear and cannot directly characterize the active nonlinear function of the outer hair cells of the inner ear. Therefore, there is no simple one-to-one correspondence between its test results and hearing thresholds.

[0005] While existing technologies include hearing assessment methods based on otoacoustic emissions and middle ear function testing methods based on acoustic impedance or broadband middle ear impedance, most solutions utilize only single-source detection information. They fail to jointly model middle ear conduction characteristics and inner ear active functions within the same testing process, making it difficult to simultaneously resolve the contradiction between abnormal middle ear interference and the accuracy of hearing threshold prediction. Therefore, there is an urgent need for a hearing testing system capable of integrating multi-source information from the middle and inner ear to improve the accuracy, robustness, and clinical applicability of objective hearing threshold prediction. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a hearing threshold prediction method that combines broadband middle ear impedance and otoacoustic emissions. This method overcomes the deficiencies of existing technologies by simultaneously acquiring middle ear conduction characteristics and inner ear outer hair cell function information in the same detection process, and constructing a joint feature input model for analysis. This overcomes the problems of insufficient accuracy and insensitivity to middle ear abnormalities caused by relying on a single detection method in existing technologies.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] This invention provides a method for predicting hearing thresholds by combining broadband middle ear impedance and otoacoustic emissions, comprising:

[0009] The detection mode is determined according to the user selection or the system preset rules, and the detection parameters corresponding to the detection mode are initialized.

[0010] After completing the selection of the detection mode and parameter initialization steps, perform ear canal seal detection and environmental noise self-check.

[0011] After meeting the preset sealing and noise conditions, wideband middle ear impedance detection is performed to obtain the middle ear acoustic response as a function of frequency and pressure, and wideband middle ear impedance-related features are extracted from it.

[0012] After completing the broadband middle ear impedance detection, otoacoustic emission detection is performed to collect otoacoustic emission signals and extract otoacoustic emission related features from them;

[0013] The broadband middle ear impedance-related features and the otoacoustic emission-related features are fused together to construct a joint feature that includes information on middle ear conduction characteristics and information on inner ear active functions.

[0014] The combined features are input into the hearing prediction model to output a hearing threshold prediction or a hearing status assessment result.

[0015] A test report is generated and output based on the predicted hearing threshold or the hearing condition assessment results.

[0016] In one embodiment, the selection of the detection mode includes the following steps:

[0017] Based on the selected detection mode, the system automatically loads the preset parameters corresponding to that detection mode; wherein, the detection modes include hearing threshold prediction mode, hearing condition screening mode, and middle ear conduction abnormality assessment mode;

[0018] The preset parameters include general data quality control parameters, wideband middle ear impedance detection parameters, and otoacoustic emission detection parameters, and all parameters are adjustable.

[0019] In one embodiment, performing ear canal seal detection and environmental noise self-check includes the following steps:

[0020] The air pressure regulation unit is used to check the seal of the ear canal, collect ear canal pressure data and analyze pressure stability.

[0021] Collect and evaluate the background noise in the ear canal to determine whether it meets the preset noise threshold conditions;

[0022] If the seal test fails or the background noise does not meet the preset conditions, a prompt message is generated; if the conditions are met, the subsequent test steps continue.

[0023] In one embodiment, performing broadband middle ear impedance detection includes the following steps:

[0024] According to the preset air pressure scanning range and step, the air pressure points are stabilized and controlled sequentially.

[0025] After each pressure point stabilizes, a wideband detection signal is output, and reflected acoustic signals are acquired simultaneously.

[0026] Based on the collected reflected acoustic signals, broadband middle ear impedance-related characteristics are calculated, including broadband absorption rate, broadband reflectivity, and amplitude and phase characteristics of admittance or impedance.

[0027] The collected reflected acoustic signals are evaluated for data quality, and measurement stability and reliability indicators are recorded.

[0028] In one embodiment, the otoacoustic emission detection includes the following steps:

[0029] Otoacoustic emission evoked stimulation signals are generated, output to the subject's ear canal, and the transmitted otoacoustic emission signals are collected simultaneously.

[0030] The acquired otoacoustic emission signals are preprocessed to extract otoacoustic emission related features, including otoacoustic emission amplitude spectrum, signal-to-noise ratio spectrum, and fine structure features;

[0031] Data quality control was performed on the extracted otoacoustic emission-related features, and the signal-to-noise ratio and repeatability consistency index were recorded. Low-quality data were also marked.

[0032] In one embodiment, the joint feature construction includes the following steps:

[0033] The broadband middle ear impedance-related features and the otoacoustic emission-related features are frequency axis aligned so that the two types of features are represented at the same or mappable frequency sampling points.

[0034] The aligned feature data is then interpolated or resampled.

[0035] The processed feature data is then normalized.

[0036] Remove marked outliers or unreliable data, and complete any missing information;

[0037] Construct a joint feature matrix or tensor that integrates middle ear conduction characteristics information and inner ear active function information.

[0038] In one embodiment, the hearing assessment includes the following steps:

[0039] The joint features are input into the dual-branch fusion hearing prediction model;

[0040] The hearing prediction model uses two branches to independently extract high-order features from the middle ear impedance-related features and otoacoustic emission-related features.

[0041] The extracted features from the two branches are weighted and fused through a fusion layer, and then combined with a multi-task learning structure for joint inference.

[0042] The corresponding evaluation results are output according to the detection mode. The evaluation results include the predicted hearing threshold value, hearing loss probability, hearing classification result or middle ear conduction abnormality risk score for each target frequency point.

[0043] In one embodiment, the result output includes the following steps:

[0044] The broadband middle ear impedance-related characteristics, otoacoustic emission-related characteristics, and evaluation results during the detection process are visualized.

[0045] By integrating the detection parameters, the broadband middle ear impedance-related characteristics, the otoacoustic emission-related characteristics, the measurement stability and reliability indicators, the signal-to-noise ratio and repeatability consistency indicators, and the evaluation results, a standardized test report is generated.

[0046] The test report can be stored on a local or remote server and can be exported or printed.

[0047] The present invention also provides a wideband middle ear impedance and otoacoustic emission combined hearing threshold prediction system for implementing the wideband middle ear impedance and otoacoustic emission combined hearing threshold prediction method described in any of the above claims. The system includes a core control and acquisition unit and a pressure regulation unit.

[0048] The core control and acquisition unit consists of a computer, a general-purpose acquisition sound card, and an integrated acoustic probe.

[0049] The computer has built-in signal analysis program, feature fusion program, and hearing prediction model;

[0050] The air pressure regulation unit is connected to the core control and acquisition unit and is used to adjust the ear canal air pressure during the detection process to provide a controllable air pressure environment for broadband middle ear impedance detection.

[0051] The system is configured to perform wideband middle ear impedance detection and extract wideband middle ear impedance-related features through the cooperation of the air pressure regulation unit and the core control and acquisition unit in wideband middle ear impedance detection mode; and to directly perform otoacoustic emission detection and extract otoacoustic emission-related features by the core control and acquisition unit in otoacoustic emission detection mode.

[0052] The system uses a feature fusion program built into the computer to fuse the broadband middle ear impedance-related features and the otoacoustic emission-related features to construct a joint feature. The system then analyzes the joint feature using the hearing prediction model and outputs a hearing threshold prediction value or a hearing status assessment result.

[0053] The present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the hearing threshold prediction method according to any one of the preceding claims.

[0054] This invention provides a hearing threshold prediction method combining broadband middle ear impedance and otoacoustic emissions. It offers the following advantages: by sequentially performing broadband middle ear impedance detection and otoacoustic emission detection within the same detection process, broadband middle ear impedance-related features and otoacoustic emission-related features are extracted separately. These two types of features are then fused to construct a joint feature, which is input into a hearing prediction model for analysis. This technical solution can compensate for and correct otoacoustic emission-related features using broadband middle ear impedance-related features in the presence of middle ear conduction abnormalities, effectively counteracting the interference of middle ear state changes on otoacoustic emission signals, improving the accuracy and robustness of hearing threshold prediction, and reducing the false negative rate of single detection methods.

[0055] This invention employs a modular design of "core reuse + dedicated expansion." The core control and acquisition unit (including a computer, a general-purpose acquisition sound card, and an integrated acoustic probe) serves as the system's basic architecture, providing shared hardware support for broadband middle ear impedance testing and otoacoustic emission testing. During broadband middle ear impedance testing, the air pressure regulation unit works in conjunction with the core control and acquisition unit to perform the test; during otoacoustic emission testing, the core control and acquisition unit is directly reused, eliminating the need to activate the air pressure regulation unit. Both testing modules reuse the core acquisition and transmission hardware within the same testing process, eliminating the need to replace testing equipment or hardware, achieving one-stop objective hearing testing, significantly improving testing efficiency, and simplifying clinical procedures.

[0056] By employing preset detection parameters, an automated data quality control mechanism, and a standardized feature fusion and model inference process, the entire detection process is automated. No subjective cooperation from the subject is required during the test. Ear canal sealing checks and environmental noise self-checks ensure that the testing conditions meet the requirements. Low-quality data is marked and processed through data quality assessment, ultimately generating a standardized test report. This technical solution is suitable for infants, the elderly, patients with impaired consciousness, and individuals who cannot effectively cooperate with pure-tone audiometry, demonstrating strong clinical applicability.

[0057] This invention employs a dual-branch fusion hearing prediction model. Two branches independently extract high-order features from middle ear impedance-related features and otoacoustic emission-related features. A fusion layer weights and fuses the features from both branches, and a multi-task learning structure is used for joint inference. This model can simultaneously output multiple assessment results, including multi-frequency hearing threshold predictions, hearing loss probability, hearing classification results, and middle ear conduction abnormality risk scores, providing rich auxiliary reference information for clinical judgment.

[0058] In summary, this invention can be widely applied in the fields of hearing testing and clinical hearing assessment, and has good clinical application prospects. Attached Figure Description

[0059] Figure 1 This is a flowchart illustrating the hearing threshold detection and hearing condition screening process according to an embodiment of the present invention.

[0060] Figure 2 This is a schematic diagram of the model architecture in an example of the present invention;

[0061] Figure 3 This is a schematic diagram of the structure of an embodiment of the hardware system of the present invention;

[0062] The attached figures are labeled as follows: 1. Computer; 2. Universal sound card; 3. Integrated acoustic probe; 31. Miniature speaker; 32. Miniature microphone; 33. Airway interface; 4. Air pressure regulation unit; 41. Sub-main control unit; 42. Motor driver; 43. Miniature DC motor; 44. Airbag; 45. Air pressure sensor. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] like Figure 1 As shown, in one embodiment, a method for predicting hearing thresholds by combining broadband middle ear impedance and otoacoustic emissions includes steps S1 to S7:

[0065] Step S1: Detection mode selection and parameter initialization: Determine the detection mode according to user selection or system preset rules, and initialize the detection parameters corresponding to the detection mode;

[0066] The selection of the detection mode includes the following steps:

[0067] Based on the selected detection mode, the system automatically loads the preset parameters corresponding to that detection mode; wherein, the detection modes include hearing threshold prediction mode, hearing condition screening mode, and middle ear conduction abnormality assessment mode;

[0068] The preset parameters include general data quality control parameters, wideband middle ear impedance detection parameters, and otoacoustic emission detection parameters, and all parameters are adjustable.

[0069] Specifically, the operator selects the detection mode through the human-computer interaction terminal of computer 1; the hearing threshold prediction mode is used to output the predicted hearing threshold values ​​at multiple target frequency points; the hearing condition screening mode is used to output the probability or / and grading results of hearing loss; and the middle ear conduction abnormality assessment mode is used to assess the middle ear function status and provide a basis for conduction correction for hearing prediction.

[0070] Based on the detection mode, the system automatically sets or loads the corresponding detection parameters, including:

[0071] Wideband middle ear impedance detection parameters: detection signal type, frequency range, frequency resolution, ear canal pressure scan range and step size;

[0072] Otoacoustic emission detection parameters: detection type (DPOAE, SFOAE or TEOAE), stimulation frequency combination, stimulation intensity, stimulation duration, number of repetitions and stopping criteria;

[0073] Data quality control parameters include background noise threshold, minimum signal-to-noise ratio threshold, and repeatability threshold. These parameters can be flexibly adjusted according to clinical testing needs. Once the parameters are set, they are stored by computer 1 and matched with the corresponding testing program.

[0074] Step S2 Ear canal seal test and environmental noise self-test: After completing the selection of the detection mode and parameter initialization steps, perform ear canal seal test and environmental noise self-test;

[0075] Performing an ear canal seal test and environmental noise self-test includes the following steps:

[0076] The air pressure regulation unit is used to check the seal of the ear canal, collect ear canal pressure data and analyze pressure stability to determine the sealing effect;

[0077] Collect and evaluate the background noise in the ear canal to determine whether it meets the preset noise threshold conditions;

[0078] If the seal test fails or the background noise does not meet the preset conditions, a prompt message is generated; if the conditions are met, the subsequent test steps continue.

[0079] Specifically, the subject assumes a comfortable position and keeps their head still. The operator selects an integrated acoustic probe 3 of the appropriate specification and gently inserts it into the subject's external auditory canal, ensuring a proper fit. Computer 1 issues a seal detection command, controlling the air pressure regulation unit to perform a seal check on the ear canal. The air pressure sensor 45 collects real-time ear canal pressure data and analyzes pressure stability to determine the seal effect. Simultaneously, the system collects and evaluates background noise within the ear canal to determine if it meets the preset noise threshold. If the seal is substandard or the noise exceeds the limit, the system automatically issues a prompt, guiding the operator to re-wear the probe or improve the testing environment. If the test results meet the requirements, the next step of the formal testing proceeds.

[0080] Step S3 Wideband Middle Ear Impedance Detection: After meeting the preset sealing and noise conditions, perform wideband middle ear impedance detection to obtain the middle ear acoustic response as a function of frequency and pressure, and extract wideband middle ear impedance-related features from it.

[0081] The process of performing wideband middle ear impedance detection includes the following steps:

[0082] According to the preset air pressure scanning range and step, the air pressure points are stabilized and controlled sequentially.

[0083] After each pressure point stabilizes, a wideband detection signal is output, and reflected acoustic signals are acquired simultaneously.

[0084] Based on the collected reflected acoustic signals, broadband middle ear impedance-related characteristics are calculated, including broadband absorption rate, broadband reflectivity, and amplitude and phase characteristics of admittance or impedance.

[0085] The collected reflected acoustic signals are evaluated for data quality, and measurement stability and reliability indicators are recorded.

[0086] Specifically, the wideband middle ear impedance detection computer 1 initiates the middle ear detection program, sequentially performing pressure stabilization and wideband signal acquisition at each pressure point according to the preset pressure scanning range and step size. For each target pressure value, the computer 1 controls the pressure regulation unit to perform closed-loop adjustment, rapidly stabilizing the ear canal pressure to the target range. After pressure stabilization, the computer 1 generates a preset type of wideband detection signal, which is transmitted into the ear canal via the speaker of the universal acquisition sound card 2 and the integrated acoustic probe 3, while simultaneously acquiring the reflected acoustic signal from the ear canal through the microphone. The computer 1 performs real-time data quality assessment on the acquired signal. If the signal quality is acceptable, frequency domain analysis is performed to calculate middle ear impedance characteristic parameters such as wideband absorption rate, reflectivity, and admittance impedance. If the signal quality is poor, the system automatically increases the number of signal acquisitions and performs averaging; if it still does not meet the standard, a quality warning is issued. After completing the detection at a single pressure point, the system switches to the next pressure point according to the preset sequence and repeats the above operations, ultimately generating a frequency-pressure related multi-dimensional middle ear impedance characteristic dataset and recording measurement stability and reliability indicators.

[0087] Step S4 Otoacoustic emission detection: After completing the broadband middle ear impedance detection, otoacoustic emission detection is performed to collect otoacoustic emission signals and extract otoacoustic emission related features from them;

[0088] The otoacoustic emission detection includes the following steps:

[0089] Otoacoustic emission evoked stimulation signals are generated, output to the subject's ear canal, and the transmitted otoacoustic emission signals are collected simultaneously.

[0090] The acquired otoacoustic emission signals are preprocessed to extract otoacoustic emission related features, including otoacoustic emission amplitude spectrum, signal-to-noise ratio spectrum, and fine structure features;

[0091] Data quality control was performed on the extracted otoacoustic emission-related features, and the signal-to-noise ratio and repeatability consistency index were recorded. Low-quality data were also marked.

[0092] Specifically, after the broadband middle ear impedance detection of otoacoustic emissions (OAEs) is completed, computer 1 can directly switch to the OAE detection process without switching any hardware, reusing the original general-purpose acquisition sound card 2, integrated acoustic probe 3, and other core acquisition and transmission hardware. Computer 1 generates corresponding OAE evoked stimulation signals (SFOAE single-frequency stimulation + inhibition sound or DPOAE dual-tone stimulation, etc.) according to the preset detection type and parameters, and sends them into the subject's external auditory canal via the transmission link, simultaneously acquiring the OAE feedback signals generated by the cochlea. After preprocessing the acquired signals such as coherent averaging, noise floor estimation, and denoising, the system extracts core features and fine structural information such as the OAE amplitude spectrum and signal-to-noise ratio spectrum; simultaneously, it performs data quality control on the OAE measurement results, recording indicators such as signal-to-noise ratio and repeatability consistency, and marking frequency points with poor signal quality for subsequent feature processing.

[0093] Step S5 Joint Feature Construction: The broadband middle ear impedance-related features and the otoacoustic emission-related features are fused to construct a joint feature that includes middle ear conduction characteristics information and inner ear active function information;

[0094] The joint feature construction includes the following steps:

[0095] The broadband middle ear impedance-related features and the otoacoustic emission-related features are frequency axis aligned so that the two types of features are represented at the same or mappable frequency sampling points.

[0096] The aligned feature data is then interpolated or resampled to unify the feature dimensions and resolution.

[0097] The processed feature data is then normalized.

[0098] Remove marked outliers or unreliable data, and complete any missing information;

[0099] Construct a joint feature matrix or tensor that integrates middle ear conduction characteristics information and inner ear active function information.

[0100] Specifically, computer 1 initiates a feature fusion processing program to uniformly standardize the broadband middle ear impedance features and otoacoustic emission features: First, the two types of features are aligned on the frequency axis so that the features are represented at the same frequency sampling points, and non-overlapping frequency points are interpolated to complete the feature dimension and resolution; then, the aligned feature data are normalized to eliminate differences in dimensions and amplitudes; finally, outliers or unreliable data marked in the previous stage are removed, and missing information is weighted and completed in a reasonable way, ultimately constructing a joint feature matrix that integrates middle ear conduction characteristics and inner ear functional information, which serves as the input data for the deep learning prediction model.

[0101] Step S6 Hearing condition assessment: Input the combined features into the hearing prediction model to output the hearing threshold prediction value or hearing condition assessment result;

[0102] The hearing assessment includes the following steps:

[0103] The joint features are input into the dual-branch fusion hearing prediction model;

[0104] The hearing prediction model uses two branches to independently extract high-order features from the middle ear impedance-related features and otoacoustic emission-related features.

[0105] The extracted features from the two branches are weighted and fused through a fusion layer, and then combined with a multi-task learning structure for joint inference.

[0106] The corresponding evaluation results are output according to the detection mode. The evaluation results include the predicted hearing threshold value, hearing loss probability, hearing classification result or middle ear conduction abnormality risk score for each target frequency point.

[0107] Specifically, the dual-branch fusion hearing prediction model is used to jointly analyze broadband middle ear impedance characteristics and otoacoustic emission characteristics, outputting hearing threshold prediction results and hearing status assessment results.

[0108] like Figure 2 As shown, the model consists of an input preprocessing layer, a middle ear feature branch, an OAE feature branch, a feature fusion layer, and an inference output layer. The input preprocessing layer performs normalization and noise reduction on the joint feature matrix to eliminate dimensional differences and abnormal interference. The middle ear feature branch specifically extracts high-order features from broadband middle ear impedance characteristics to mine information related to middle ear conduction properties. The OAE feature branch independently extracts features from otoacoustic emissions to characterize the functional state of outer hair cells in the inner ear. The feature fusion layer weighted and fused the high-order features extracted from the two branches, integrating information from both the middle ear and inner ear. The inference output layer performs regression or classification inference based on the fused features, outputting predicted hearing thresholds, hearing status classification results, or middle ear abnormality risk scores.

[0109] The model is trained using supervised learning, with clinical pure-tone audiometry results as the gold standard. Sufficient multi-type hearing samples are input to complete training and optimization. It has strong generalization and anti-interference capabilities, and can effectively correct the interference of middle ear abnormalities on OAE signals, thereby improving the reliability of prediction results.

[0110] Step S7 Result Output: Generate and output a test report based on the predicted hearing threshold or hearing status assessment result;

[0111] The output of the result includes the following steps:

[0112] The broadband middle ear impedance-related characteristics, otoacoustic emission-related characteristics, and evaluation results during the detection process are visualized.

[0113] By integrating the detection parameters, the broadband middle ear impedance-related characteristics, the otoacoustic emission-related characteristics, the measurement stability and reliability indicators, the signal-to-noise ratio and repeatability consistency indicators, and the evaluation results, a standardized test report is generated.

[0114] The test report can be stored on a local or remote server and can be exported or printed.

[0115] Specifically, Computer 1 displays the detection data, feature maps, and prediction results in real time on the human-computer interaction terminal, including a three-dimensional broadband middle ear impedance map, OAE amplitude spectrum signal-to-noise ratio spectrum, hearing threshold prediction curve, hearing loss probability distribution, and middle ear conduction abnormality risk score, supporting interactive viewing of the maps. Simultaneously, Computer 1 automatically integrates the detection parameters, raw feature data, data quality control indicators, and prediction results from the entire process to generate a standardized detection report. The report covers the subject's basic information, detection mode and parameters, various feature maps, hearing threshold prediction hearing status screening results, measurement quality and reliability assessment, etc., supporting local storage, printing, and export in common formats. Results can also be stored on a remote server for subsequent analysis or follow-up.

[0116] The following describes a broadband middle ear impedance and otoacoustic emission combined hearing threshold prediction system provided by the present invention. The broadband middle ear impedance and otoacoustic emission combined hearing threshold prediction system described below corresponds to the method described above.

[0117] like Figure 3 As shown, in one embodiment, a wideband middle ear impedance and otoacoustic emission combined hearing threshold prediction system includes a core control and acquisition unit and a pressure regulation unit.

[0118] The core control and acquisition unit consists of a computer 1, a general-purpose acquisition sound card 2, and an integrated acoustic probe 3;

[0119] Specifically, Computer 1 employs a multi-core, low-power processor with an embedded operating system, featuring a built-in floating-point arithmetic unit and acceleration module, sufficient storage capacity, and communication interfaces including a USB interface and a UART serial port. Computer 1 integrates acoustic calibration, wideband middle ear impedance analysis, otoacoustic emission fine structure extraction, joint feature fusion, deep learning prediction models, and report generation programs. As the core of the system for command issuance, data processing, and flow control, it achieves bidirectional data communication and power supply with the general-purpose acquisition sound card 2 via the USB interface, and real-time command interaction and data feedback with the air pressure regulation unit via the UART serial port. Simultaneously, Computer 1 serves as the system's operation input and result display terminal, supporting parameter setting, real-time monitoring of detection progress, visualization of detection results, and quick operation. It can display information such as test progress, current air pressure value, signal-to-noise ratio, module operating status, three-dimensional wideband middle ear impedance spectrum, otoacoustic emission amplitude spectrum signal-to-noise ratio spectrum, and hearing threshold prediction curve.

[0120] Universal Acquisition Sound Card 2: This is a high-precision USB external sound card with multiple DAC digital-to-analog conversion channels and ADC analog-to-digital conversion channels to realize digital-to-analog conversion of excitation signals and analog-to-digital conversion of acquisition signals. The sound card is powered and transmits data bidirectionally to the computer 1 via a USB interface. Its DAC channel output is connected to the miniature speaker 31 of the integrated acoustic probe 3 to output the detection excitation signal. Its ADC channel input is connected to the miniature microphone 32 of the integrated acoustic probe 3 to acquire the acoustic signal transmitted back from the ear canal, ensuring the synchronization and stability of signal conversion.

[0121] The integrated acoustic probe 3 features an arc-shaped ergonomic design to fit the physiological structure of the human external auditory canal. The tip is equipped with medical-grade silicone earplugs in multiple sizes to ensure a tight seal. Internally, the probe encapsulates a miniature speaker 31 and a high-sensitivity miniature microphone 32. Both the speaker and microphone signal lines are independently shielded to prevent electromagnetic interference. The probe also includes an airway interface 33, which connects to the airbag 44 of the air pressure regulation unit via a silicone hose, ensuring airtightness. The miniature speaker 31 is connected to the DAC channel output of the universal acquisition sound card 2, and the miniature microphone 32 is connected to the ADC channel input, enabling the transmission of acoustic excitation signals and the acquisition of return signals.

[0122] The core control and acquisition unit generates, acquires, and extracts fine structures from the evoked signals of SFOAE and DPOAE. The middle ear detection module consists of a computer (1), a universal acquisition sound card (2), an integrated acoustic probe (3), and a pressure regulation unit, enabling high-precision closed-loop control of ear canal pressure and wideband middle ear impedance detection and analysis. Both modules reuse the core acquisition and transmission hardware, allowing for function switching solely through the computer (1) program, eliminating the need for hardware replacement, significantly improving detection efficiency and reducing clinical operational complexity.

[0123] The computer 1 has built-in signal analysis program, feature fusion program and hearing prediction model; as the system's command issuance and data processing center, it realizes data interaction and command transmission with the general acquisition sound card 2 and dedicated unit through the communication interface, and also has human-computer interaction function, supporting detection parameter setting, detection progress monitoring and result visualization display.

[0124] Specifically, the universal acquisition sound card 2 serves as the signal conversion core, possessing multiple DAC digital-to-analog conversion channels and ADC analog-to-digital conversion channels to realize the digital-to-analog conversion of the excitation signal and the analog-to-digital conversion of the acquired signal, ensuring high precision and synchronization of the signal conversion. Its output and input ends are respectively connected to the signal transmission and acquisition components of the integrated acoustic probe 3.

[0125] The integrated acoustic probe 3 adopts an ergonomic design, with multi-size medical silicone earplugs at the tip to ensure a good seal in the ear canal. It internally encapsulates a miniature speaker 31 and a high-sensitivity miniature microphone 32. The signal line adopts an independent shielding design to avoid electromagnetic interference. At the same time, it has a reserved air passage interface 33 to provide a physical channel for air pressure regulation, realizing the integration of acoustic signal transmission and air pressure conduction.

[0126] The integrated acoustic probe 3 is used to output acoustic excitation signals and simultaneously acquire acoustic signals transmitted back from the ear canal.

[0127] Throughout the testing process, computer 1 monitors the operational status of each module and the quality of the collected data in real time. If any anomaly is detected, the testing process is immediately paused, a clear prompt is issued via the human-computer interaction terminal, and the anomaly information is recorded. Specific anomaly handling rules are as follows:

[0128] Abnormal air pressure regulation: If the ear canal pressure cannot stabilize to the target range within a reasonable time, the system will immediately terminate the air pressure regulation and issue a fault prompt, guiding the operator to check the relevant components of the air pressure module and the air circuit sealing.

[0129] Probe detachment or seal failure: During non-pressure regulation phases, if there are significant abnormal fluctuations in ear canal pressure that persist for a certain period of time, the system will determine that the probe has detached or is leaking, suspend the test, and prompt the user to put the probe back on. After resetting, the current test process can be resumed.

[0130] The quality of the acquired signal is seriously substandard: In the middle ear or OAE detection process, the signal cannot meet the quality control requirements after multiple consecutive acquisitions. The system indicates environmental interference or probe wearing problems, and guides you to improve the detection environment or retest after repeating the operation.

[0131] Hardware communication failure: Communication between computer 1 and peripherals such as the sound card and air pressure regulating unit 4 is interrupted. Immediately pause the entire process detection and prompt a communication abnormality. Guide the inspection of hardware interfaces and connection status. After troubleshooting, restart the system to restore.

[0132] The air pressure regulation unit, connected to the core control and acquisition unit, is used to adjust the ear canal air pressure during the detection process, providing a controllable air pressure environment for broadband middle ear impedance detection.

[0133] Specifically, the air pressure regulating unit 4 includes a sub-main control unit 41, a motor driver 42, a micro DC motor 43, an airbag 44 (medical silicone airbag) and an air pressure sensor 45 (piezoresistive air pressure sensor). Each component is connected in sequence to form a closed-loop control.

[0134] The sub-main control unit 41 communicates with the computer 1 via a UART serial port, receives the target air pressure sequence command issued by the computer 1, and runs a PID control algorithm to generate a PWM control signal. The motor driver 42 receives the PWM signal from the sub-main control unit 41, amplifies it, and drives the micro DC motor 43 to rotate forward and backward. The output of the micro DC motor 43 is connected to the airbag 44, and the air pressure in the ear canal is changed by adjusting the volume of the airbag 44. The airbag 44 is connected to the air path interface 33 of the integrated acoustic probe 3 through a silicone tube to achieve precise adjustment of the ear canal air pressure. The air pressure sensor 45 is installed in the air path near the probe to collect the air pressure value in the ear canal in real time. Its analog output signal is connected to the built-in ADC of the sub-main control unit 41 to provide real-time feedback for the air pressure closed-loop control and achieve high-precision and stable control of the ear canal air pressure.

[0135] The system is configured to perform wideband middle ear impedance detection and extract wideband middle ear impedance-related features through the cooperation of the air pressure regulation unit and the core control and acquisition unit in wideband middle ear impedance detection mode; and to directly perform otoacoustic emission detection and extract otoacoustic emission-related features by the core control and acquisition unit in otoacoustic emission detection mode.

[0136] The system uses the feature fusion program built into the computer 1 to fuse the broadband middle ear impedance-related features and the otoacoustic emission-related features to construct joint features. The system then analyzes the joint features using the hearing prediction model and outputs a hearing threshold prediction value or a hearing status assessment result.

[0137] In one embodiment, a broadband middle ear impedance and otoacoustic emission (OAE) combined hearing threshold prediction system further includes a computer 1 that directly reuses the core control and acquisition unit, a universal acquisition sound card 2, and an integrated acoustic probe 3, without the need for additional dedicated hardware. Functional adaptation is achieved solely through program switching on the computer 1, completing the induction, acquisition, and feature extraction of OAE signals. It supports stimulation frequency otoacoustic emission (SFOAE), distortion product otoacoustic emission (DPOAE), or transient evoked otoacoustic emission (TEOAE) detection. The computer 1 generates corresponding evoked stimulus signals, which are transmitted to the ear canal via the universal acquisition sound card 2 and the integrated acoustic probe 3. Simultaneously, the acquired OAE signals are preprocessed with coherent averaging, noise suppression, and frequency domain analysis to extract the corresponding OAE amplitude spectrum, signal-to-noise ratio spectrum, and optional fine structural features, providing data support for subsequent joint feature construction.

[0138] The feature preprocessing module is used to perform unified preprocessing on the middle ear features (including broadband absorption rate, reflectivity, admittance impedance amplitude and phase, etc.) obtained by broadband middle ear impedance detection and the inner ear features (including amplitude spectrum, signal-to-noise ratio spectrum, fine structure, etc.) obtained by otoacoustic emission detection.

[0139] The preprocessing includes steps such as frequency axis alignment, interpolation or resampling, normalization, and outlier removal to eliminate the frequency dimension and dimensional differences between the two types of features, remove low-quality and unreliable data, and finally construct a joint feature matrix containing information on middle ear conduction characteristics and inner ear active function, and organize frequency-specific multimodal input features according to the target predicted frequency.

[0140] The dual-branch feature extraction module includes a middle ear feature branch and an OAE feature branch, which independently extract high-order features from the preprocessed middle ear features and inner ear features, respectively. The middle ear feature branch is specifically designed to mine core information related to the conduction characteristics of the middle ear; the OAE feature branch is specifically designed to characterize key features related to the functional state of the outer hair cells of the inner ear. By extracting features in a targeted manner, the limitations of a single feature dimension are avoided, ensuring the comprehensiveness and effectiveness of the feature information.

[0141] The feature fusion module performs weighted fusion of the high-order features extracted from the two branches, integrates information from both the middle ear and inner ear, and enhances the model's ability to characterize hearing status through feature complementarity.

[0142] The inference output module performs regression and classification inference based on fused features, and outputs multi-frequency hearing threshold prediction values, hearing loss probability or grading results. It can also output middle ear conduction abnormality risk warnings or otoacoustic emission reliability scores to provide auxiliary reference for clinical judgment. The joint modeling method significantly improves the stability and reliability of hearing threshold prediction.

[0143] An electronic device includes a memory and a processor, the memory storing a computer program, characterized in that, when the processor executes the computer program, it implements the wideband middle ear impedance and otoacoustic emission combined hearing threshold prediction method described in any one of the preceding claims.

[0144] Specifically, the electronic device can be a smart terminal, comprising a processor, internal memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a wideband hearing threshold prediction method combining middle ear impedance and otoacoustic emissions.

[0145] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting hearing thresholds by combining broadband middle ear impedance and otoacoustic emissions, characterized in that, include: The detection mode is determined according to the user selection or the system preset rules, and the detection parameters corresponding to the detection mode are initialized. After completing the selection of the detection mode and parameter initialization steps, perform ear canal seal detection and environmental noise self-check. After meeting the preset sealing and noise conditions, wideband middle ear impedance detection is performed to obtain the middle ear acoustic response as a function of frequency and pressure, and wideband middle ear impedance-related features are extracted from it. After completing the broadband middle ear impedance detection, otoacoustic emission detection is performed to collect otoacoustic emission signals and extract otoacoustic emission related features from them; The broadband middle ear impedance-related features and the otoacoustic emission-related features are fused together to construct a joint feature that includes information on middle ear conduction characteristics and information on inner ear active functions. The combined features are input into the hearing prediction model to output a hearing threshold prediction or a hearing status assessment result. A test report is generated and output based on the predicted hearing threshold or the hearing condition assessment results.

2. The method for predicting hearing thresholds by combining broadband middle ear impedance and otoacoustic emissions according to claim 1, characterized in that, The selection of the detection mode includes the following steps: Based on the selected detection mode, the system automatically loads the preset parameters corresponding to that detection mode; wherein, the detection modes include hearing threshold prediction mode, hearing condition screening mode, and middle ear conduction abnormality assessment mode; The preset parameters include general data quality control parameters, wideband middle ear impedance detection parameters, and otoacoustic emission detection parameters, and all parameters are adjustable.

3. The method for predicting hearing thresholds by combining broadband middle ear impedance and otoacoustic emissions according to claim 1, characterized in that, Performing an ear canal seal test and environmental noise self-test includes the following steps: The air pressure regulation unit is used to check the seal of the ear canal, collect ear canal pressure data and analyze pressure stability. Collect and evaluate the background noise in the ear canal to determine whether it meets the preset noise threshold conditions; If the seal test fails or the background noise does not meet the preset conditions, a prompt message is generated; if the conditions are met, the subsequent test steps continue.

4. The method for predicting hearing thresholds by combining broadband middle ear impedance and otoacoustic emissions according to claim 1, characterized in that, The process of performing wideband middle ear impedance detection includes the following steps: According to the preset air pressure scanning range and step, the air pressure points are stabilized and controlled sequentially. After each pressure point stabilizes, a wideband detection signal is output, and reflected acoustic signals are acquired simultaneously. Based on the collected reflected acoustic signals, broadband middle ear impedance-related characteristics are calculated, including broadband absorption rate, broadband reflectivity, and amplitude and phase characteristics of admittance or impedance. The collected reflected acoustic signals are evaluated for data quality, and measurement stability and reliability indicators are recorded.

5. The method for predicting hearing thresholds by combining broadband middle ear impedance and otoacoustic emissions according to claim 1, characterized in that, The otoacoustic emission detection includes the following steps: Otoacoustic emission evoked stimulation signals are generated, output to the subject's ear canal, and the transmitted otoacoustic emission signals are collected simultaneously. The acquired otoacoustic emission signals are preprocessed to extract otoacoustic emission related features, including otoacoustic emission amplitude spectrum, signal-to-noise ratio spectrum, and fine structure features; Data quality control was performed on the extracted otoacoustic emission-related features, and the signal-to-noise ratio and repeatability consistency index were recorded. Low-quality data were also marked.

6. The method for predicting hearing thresholds by combining broadband middle ear impedance and otoacoustic emissions according to claim 1, characterized in that, The joint feature construction includes the following steps: The broadband middle ear impedance-related features and the otoacoustic emission-related features are frequency axis aligned so that the two types of features are represented at the same or mappable frequency sampling points. The aligned feature data is then interpolated or resampled. The processed feature data is then normalized. Remove marked outliers or unreliable data, and complete any missing information; Construct a joint feature matrix or tensor that integrates middle ear conduction characteristics information and inner ear active function information.

7. The method for predicting hearing thresholds by combining broadband middle ear impedance and otoacoustic emissions according to claim 1, characterized in that, The hearing assessment includes the following steps: The joint features are input into the dual-branch fusion hearing prediction model; The hearing prediction model uses two branches to independently extract high-order features from the middle ear impedance-related features and otoacoustic emission-related features. The extracted features from the two branches are weighted and fused through a fusion layer, and then combined with a multi-task learning structure for joint inference. The corresponding evaluation results are output according to the detection mode. The evaluation results include the predicted hearing threshold value, hearing loss probability, hearing classification result or middle ear conduction abnormality risk score for each target frequency point.

8. The method for predicting hearing thresholds by combining broadband middle ear impedance and otoacoustic emissions according to claim 1, characterized in that, The output of the result includes the following steps: The broadband middle ear impedance-related characteristics, otoacoustic emission-related characteristics, and evaluation results during the detection process are visualized. By integrating the detection parameters, the broadband middle ear impedance-related characteristics, the otoacoustic emission-related characteristics, the measurement stability and reliability indicators, the signal-to-noise ratio and repeatability consistency indicators, and the evaluation results, a standardized test report is generated. The test report can be stored on a local or remote server and can be exported or printed.

9. A broadband middle ear impedance and otoacoustic emission combined hearing threshold prediction system, used to implement the hearing threshold prediction method according to any one of claims 1 to 8, characterized in that, Includes a core control and data acquisition unit and a pressure regulation unit; The core control and acquisition unit consists of a computer, a general-purpose acquisition sound card, and an integrated acoustic probe. The computer has built-in signal analysis program, feature fusion program, and hearing prediction model; The air pressure regulation unit is connected to the core control and acquisition unit and is used to adjust the ear canal air pressure during the detection process to provide a controllable air pressure environment for broadband middle ear impedance detection. The system is configured to perform wideband middle ear impedance detection and extract wideband middle ear impedance-related features through the cooperation of the air pressure regulation unit and the core control and acquisition unit in wideband middle ear impedance detection mode; and to directly perform otoacoustic emission detection and extract otoacoustic emission-related features by the core control and acquisition unit in otoacoustic emission detection mode. The system uses a feature fusion program built into the computer to fuse the broadband middle ear impedance-related features and the otoacoustic emission-related features to construct a joint feature. The system then analyzes the joint feature using the hearing prediction model and outputs a hearing threshold prediction value or a hearing status assessment result.

10. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the hearing threshold prediction method according to any one of claims 1 to 8.

Images