Tinnitus personalized treatment method based on artificial intelligence

Abstract

The invention discloses a personalized tinnitus treatment method based on artificial intelligence, and particularly relates to the technical field of artificial intelligence, and the method comprises the steps: S1, user data collection: collecting user data, and transmitting the collected data to an edge computing node; s2, user data preprocessing: carrying out data preprocessing; s3, sending a medical diagnosis request: collecting medical data, and sending the medical data and the medical diagnosis request to a control center; s4, comprehensive analysis and scheme making, wherein comprehensive analysis is carried out, and a personalized tinnitus treatment scheme is made; s5, decision instruction transmission and scheme execution; and S6, dynamic optimization and feedback: collecting treatment feedback data in real time, obtaining a treatment effect evaluation index, and dynamically adjusting a treatment scheme. By collecting user data and medical data in real time, constructing an artificial intelligence model to judge the tinnitus probability and tinnitus subtype, formulating a treatment scheme, judging the treatment effect and dynamically adjusting according to the treatment effect, the treatment effect of the user is effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology. More specifically, the present invention relates to a personalized tinnitus treatment method based on artificial intelligence. Background Art

[0002] According to the "Global Hearing Health Report" released by the World Health Organization in 2025, there are more than 150 million tinnitus patients globally. The prevalence rate among Chinese adults is about 10% - 15%, and it shows a trend of getting younger. This trend of getting younger brings great pressure to the medical system, and traditional medical means cannot meet the increasingly serious treatment needs. The causes of tinnitus are relatively complex and may be caused by ear diseases, nervous system abnormalities, cardiovascular diseases or psychological factors. So far, there are more than 1 billion tinnitus patients globally, accumulating a large amount of clinical data. And the cross-research of neuroscience, acoustic engineering and AI has promoted the whole-chain innovation from mechanism research to clinical transformation. With the development of artificial intelligence technology, machine learning has begun to be used for tinnitus data analysis to predict tinnitus types.

[0003] However, when it is actually used, there are still some disadvantages. First, the data collection dimension is single. Traditional tinnitus diagnosis and treatment rely on single-dimensional data collection. There are problems of missing diagnostic information in traditional diagnosis and treatment, and it is easy to ignore the association between some autonomic functions and tinnitus, resulting in the inability to comprehensively evaluate the causes and a high misdiagnosis rate. Second, the data standardization is insufficient. The data storage of traditional tinnitus diagnosis and treatment methods is scattered and the formats are not unified, resulting in insufficient knowledge discovery ability and difficulty in mining the hidden associations of data, and thus unable to provide effective data support for the formulation of subsequent tinnitus treatment plans. Third, the lack of dynamic optimization ability. Traditional tinnitus diagnosis and treatment plans lack a dynamic adjustment mechanism and cannot adapt to the changes of patients, resulting in a decline in the treatment effect of tinnitus. And traditional plans rely on regular follow-up visits of patients and cannot capture treatment responses in time, resulting in the inability to adjust treatment plans in time, thus reducing the treatment effect of patients. Summary of the Invention

[0004] In view of this, the embodiments of the present invention provide a personalized tinnitus treatment method based on artificial intelligence. By collecting data from user terminals and hospital terminals for multi-source data fusion processing, and constructing an artificial intelligence model to judge the tinnitus probability and tinnitus subtypes, formulating a treatment plan, and dynamically adjusting the treatment strategy according to the treatment effect, it effectively solves the problems of single data, lack of data standardization and insufficient dynamic optimization ability proposed in the background art.

[0005] To achieve the above object, the present invention provides the following technical solutions: A personalized tinnitus treatment method based on artificial intelligence, including a user terminal, an edge computing node, a hospital terminal, a control center, and a cloud database, where the user terminal and the edge computing node are connected via Bluetooth, and the edge computing node, the hospital terminal, the control center, and the cloud database are connected via a 5G communication protocol, including the following steps: S1: User data collection: Real-time collection of user data through the user terminal, where the user data includes tinnitus data, physiological data, and environmental data, and the collected data is transmitted to the edge computing node; S2: User data preprocessing: The edge computing node performs data preprocessing on the received user data, transmits the processed data to the control center, and caches historical data locally; S3: Sending a medical diagnosis request: Collecting medical data through the hospital terminal and sending the medical data and the medical diagnosis request to the control center; S4: Comprehensive analysis and solution formulation: The control center comprehensively analyzes the processed user data and medical data, formulates a personalized tinnitus treatment plan, and generates a decision instruction according to the treatment plan; S5: Decision instruction transmission and solution execution: The control center sends the decision instruction to the edge computing node, which is parsed into real-time treatment parameters and transmitted to the user terminal, and the user terminal executes the treatment operation; S6: Dynamic optimization and feedback: During the user's treatment process, real-time collection of treatment feedback data, processing and analysis of the treatment feedback data to obtain a treatment effect evaluation index, and dynamically adjusting the treatment plan.

[0006] Technical effects and advantages of the present invention: 1. The present invention collects user data and medical data through the user terminal and the medical terminal, where the user data includes tinnitus data, physiological data, and environmental data, and performs multi-source data fusion on the preprocessed user data and medical data through the control center, which can more comprehensively reflect the patient's tinnitus situation, and the control center uses the historical data of the cloud database to construct an artificial intelligence model to judge the tinnitus probability and tinnitus subtype, comprehensively evaluate the cause of the disease, and reduce the misdiagnosis rate; 2. The present invention adopts the method of deploying an edge computing node to preprocess the user data collected at the user terminal, offline cache the historical data, and unify the data format, reducing the difficulty of calculation and providing effective data support for the subsequent judgment of tinnitus probability and tinnitus cause; 3. By collecting treatment feedback data in real time, the present invention determines the treatment effect of the personalized tinnitus treatment plan, calculates the treatment effect evaluation index based on the tinnitus reduction percentage, the improvement percentage of the HRV stress index, the reduction percentage of skin conductance, the treatment cycle, and the resource consumption coefficient, sets a preset TEI warning line, determines the treatment effect of tinnitus by comparing the treatment effect evaluation index with the TEI warning line, and dynamically adjusts the treatment strategy according to the treatment effect by taking corresponding solutions, effectively improving the treatment effect of patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 It is a schematic diagram of the method steps of the present invention.

[0008] Figure 2 It is a schematic diagram of the overall structure of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0009] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0010] As shown in the attached Figure 1 A personalized tinnitus treatment method based on artificial intelligence includes a user terminal, an edge computing node, a hospital terminal, a control center, and a cloud database. The user terminal and the edge computing node are connected by Bluetooth, and the edge computing node, the hospital terminal, the control center, and the cloud database are connected by a 5G communication protocol.

[0011] In a more specific application of the present invention, the user terminal consists of a wearable device and a mobile application, which can be a smart treatment earplug, a wrist physiological monitor, and an AR assistive glasses, and is used to collect physiological data, sense environmental data, and interact with the user in real time. The user terminal is connected to the edge computing node by wireless connection.

[0012] The edge computing node is deployed in a community health center or a home gateway and is used for real-time data processing, dynamic optimization, and edge caching. Real-time data processing refers to denoising and feature extraction of the original data collected by the user terminal, which helps to reduce the pressure on the cloud. Dynamic optimization refers to real-time adjustment of treatment parameters according to environmental noise or patient feedback. Edge storage refers to temporarily storing historical data for supporting offline analysis and synchronizing it to the cloud after the network is restored. The edge computing node is connected to the control center by a 5G communication protocol.

[0013] The hospital terminal includes a doctor workstation and a medical imaging system. The doctor workstation is specifically used to integrate the electronic medical record system, visually display the tinnitus pattern of patients, combine AI prescriptions with doctor experience, formulate personalized plans for patients, and directly control in-hospital treatment equipment through the IoT protocol. The hospital terminal is connected to the control center through a medical private network.

[0014] The control center includes an AI model server and a decision engine, which are used for global model training, multi-modal data fusion, and resource scheduling. Global model training refers to aggregating data from multiple edge computing nodes and using deep learning to predict the change trend of tinnitus. Multi-modal data fusion refers to integrating gene data, imaging materials, and patient feedback to improve the accuracy of the model. Resource scheduling refers to dynamically allocating cloud computing power to high-priority tasks. The control center is connected to the cloud database through an optical transmission network.

[0015] The cloud database is used for data storage, cross-institutional sharing, and research support. Data storage refers to storing multi-dimensional data using a distributed structure. Cross-institutional sharing refers to realizing data rights confirmation through blockchain technology. Research support refers to providing API interfaces for academic institutions to access de-identified data.

[0016] For the connection methods of the above user terminal, edge computing nodes, hospital terminal, control center, and cloud database, see Figure 2 。

[0017] The specific implementation of the present invention includes the following steps: S1: User data collection: Real-time collect user data through the user terminal. The user data includes tinnitus data, physiological data, and environmental data, and transmit the collected data to the edge computing node.

[0018] Furthermore, the tinnitus data specifically includes tinnitus frequency, tinnitus intensity, pitch, duration, and attack frequency. The physiological data specifically includes electroencephalogram signals, heart rate variability, and blood pressure. The environmental data specifically includes environmental noise intensity, noise frequency components, noise duration, temperature, humidity, and light intensity.

[0019] In this embodiment, it should be specifically noted that the intelligent earplug in the user terminal is used to collect the sound signals near the ear, analyze the spectrum through fast Fourier transform to obtain the dominant frequency and intensity of tinnitus, analyze the waveform and spectrum characteristics of the sound signals, judge the tinnitus tone according to the preset tone classification model, and use the timing module built in the user terminal to record the start and end times of each detected tinnitus signal and accumulate the number of attacks within a week to obtain the attack frequency; by wearing the head-mounted device in the user terminal, weak electrical signals on the scalp surface are collected and the electroencephalogram signals are obtained after amplification and filtering, the blood vessel pulsation at the wrist is monitored by the intelligent bracelet to obtain continuous heart rate data, the user terminal calculates the difference between adjacent cardiac cycle intervals to obtain heart rate variability, and the blood pressure of the patient is detected by the intelligent bracelet; the surrounding environmental sounds are collected by the microphone of the user terminal, the environmental noise intensity and frequency components are analyzed by using signal processing algorithms, and the start and end times of the environmental noise are recorded, the noise duration is counted, the environmental temperature and humidity are monitored in real time by the temperature and humidity sensors integrated in the user terminal, and the environmental light intensity is detected by the fiber optic sensor.

[0020] S2: User data preprocessing: The edge computing node performs data preprocessing on the received user data, transmits the processed data to the control center, and caches historical data locally.

[0021] Furthermore, the data preprocessing includes the following steps: A1: The edge computing node performs data cleaning and denoising on the received user data, including noise filtering and missing value filling; In this embodiment, it should be specifically noted that noise filtering specifically refers to removing tinnitus noise, removing environmental interference noise through an adaptive filtering algorithm, retaining the pure signal, detecting abnormal noise spikes by using the sliding window method, marking and removing sudden interferences, removing physiological noise by using independent component analysis, eliminating power frequency interference through a filter, detecting and correcting motion artifacts based on the moving average method, and removing environmental noise by performing a moving average smoothing process on the temperature and humidity sensors to eliminate the influence of instantaneous temperature and humidity fluctuations; missing value filling specifically refers to filling short-term data missing by using time series interpolation and marking long-term missing data as invalid.

[0022] A2: Perform feature extraction and dimensionality reduction processing on the data after cleaning and denoising; Specifically in this embodiment, feature extraction specifically includes extracting acoustic features and temporal features in tinnitus data features. The dominant frequency and sound pressure level are extracted through Fourier transform, the pitch feature is analyzed through Mel frequency cepstral coefficients, and the standard deviation of single-duration and attack interval time is calculated. The electroencephalogram signal, heart rate variability, and blood pressure in physiological data features are extracted. By extracting the power spectral density of each frequency band, the synchronization index of different brain regions is calculated. The autonomic nerve balance is reflected by calculating time-domain indexes and high-low frequency power ratios, and the mean value, maximum value, minimum value, and fluctuation range of blood pressure are statistically analyzed. The noise and physical environment in environmental data features are extracted. The equivalent continuous A-weighted sound level and noise impact index in the noise are extracted, and the 1 / 3 octave spectrum is analyzed. The original values of temperature, humidity, and light intensity in the physical environment are directly retained and normalized. Dimensionality reduction processing specifically refers to applying linear discriminant analysis to multi-dimensional features, compressing the feature dimension, and reducing the computational complexity.

[0023] A3: Finally, data compression and block transmission are performed on the data. Sensitive data is desensitized, and the local cache is managed.

[0024] Specifically in this embodiment, data compression specifically refers to using MP3 encoding compression for non-critical data to retain the main frequency components. For critical data, such as electroencephalogram signals, compressive sensing is used to reduce the data volume without losing critical information. Block transmission specifically refers to encapsulating the preprocessed data according to time windows and preferentially sending data blocks related to abnormal times, such as the tinnitus attack time period, and reducing the transmission priority of regular data. Desensitization processing specifically refers to anonymizing physiological signals and deleting user identification information to ensure that privacy data cannot be traced.

[0025] S3: Send a medical diagnosis request: Collect medical data through the hospital terminal and send the medical data and medical diagnosis request to the control center.

[0026] Furthermore, the medical data includes patient basic data, clinical examination data, and auxiliary examination data. Among them, the patient basic data specifically includes identity identification, medical record, and current symptom information. The clinical examination data specifically includes tinnitus-specific data, physiological monitoring data, and environmental data. The auxiliary examination data specifically includes imaging materials and special detections.

[0027] It should be specifically noted in this embodiment that the identity identifier specifically includes name, age, gender, medical record number, and medical insurance ID, and the medical record specifically includes historical diagnosis results, medication history, allergy history, and surgical records. The special tinnitus data specifically includes pure tone test results, tinnitus matching results, tinnitus disability inventory scores, and visual analogue scale scores. The physiological monitoring data specifically includes real-time electroencephalogram waveforms, heart rate variability, blood pressure trend charts, and laboratory test results. The imaging test materials specifically include MRI / CT scans and electrocochleograms, and the special tests specifically include vestibular function tests and acoustic impedance tests.

[0028] Furthermore, the medical diagnosis request needs to obtain diagnosis request parameters, and the setting of the diagnosis request parameters is determined according to the request type, priority, and resource requirements, and data formatting and data processing are performed on the medical parameters and diagnosis data, and a suitable transmission protocol is selected and transmitted to the control center.

[0029] It should be specifically noted in this embodiment that the request types include routine diagnosis, emergency green channel, multidisciplinary consultation request, and remote real-time diagnosis. The priority identifiers include the urgency level and the demand for the congestion response time. For example, the urgency level is represented by red for sudden hearing loss and yellow for continuous tinnitus aggravation. The resource requirements include the specified expert field, the required equipment support, and the data processing requirements. Data formatting specifically includes encapsulating structured data in the HL7 FHIR format, transmitting unstructured data through the DICOM standard, and attaching metadata tags. The transmission protocol specifically refers to using an encrypted channel, meeting the privacy protection requirements, and supporting resume from breakpoint and priority queue.

[0030] S4: Comprehensive analysis and plan formulation: The control center comprehensively analyzes the processed user data and medical data, formulates a personalized tinnitus treatment plan, and generates decision instructions according to the treatment plan.

[0031] Further, the comprehensive analysis specifically includes the following steps: B1: Perform feature engineering fusion on the received data. The feature engineering fusion specifically includes composite feature generation and multimodal fusion strategies; It should be specifically noted in this embodiment that the composite feature generation specifically refers to calculating the tinnitus trigger index and derivative dynamic features. Among them, the calculation of the tinnitus trigger index needs to be obtained according to the noise exposure duration, and the derivative dynamic feature can be the daily tinnitus frequency fluctuation amplitude or the circadian rhythm of physiological indicators. The multimodal fusion strategies specifically include early fusion and late fusion. Among them, early fusion refers to splicing the features of different data sources into a unified vector, and late fusion refers to independently training sub-models and then integrating the prediction results through weighted average, and selecting a suitable multimodal fusion strategy according to the requirements.

[0032] B2: Build and train an artificial intelligence model, select the algorithm of the model, construct the network structure, and formulate a training strategy; Specifically in this embodiment, the model algorithms include the Transformer architecture and the deep neural network. Among them, the Transformer architecture processes time series data, and the deep neural network fuses multi-modal features. The network structure includes an input layer, a feature extraction layer, and a fusion layer. The input layer is used to receive preprocessed physiological, environmental, and tinnitus data. The feature extraction layer is used to capture data patterns through a convolutional neural network (CNN) or an attention mechanism. The fusion layer is used to splice multi-modal features and output the tinnitus probability. Formulating a training strategy includes data augmentation, class balancing, and optimization objectives. Among them, data augmentation refers to injecting noise into time series data to improve the robustness of the model. Class balancing refers to using oversampling to process minority class data to avoid the model being biased. The optimization objective refers to minimizing the binary cross-entropy loss and maximizing the diagnostic accuracy.

[0033] B3: Perform real-time inference, dynamic decision-making, and causal inference through the artificial intelligence model to obtain the tinnitus probability and tinnitus subtypes.

[0034] Specifically in this embodiment, real-time inference specifically includes edge-side inference and control center refinement. Edge-side inference specifically means that the lightweight model quickly generates a preliminary diagnosis result on the local device. Control center refinement means that through in-depth analysis using the artificial intelligence model in the control center, the tinnitus probability is obtained. Dynamic decision-making includes threshold adjustment and multi-dimensional verification. Threshold adjustment refers to dynamically adjusting the judgment threshold according to the patient's age and medical history. Multi-dimensional verification refers to combining clinical guidelines with AI prediction results to reduce the risk of misdiagnosis. Causal inference specifically means using causal inference methods to analyze the direct association between tinnitus data, physiological data, and environmental data and tinnitus, and identifying the patient's tinnitus subtypes.

[0035] Furthermore, the formulation of a personalized tinnitus treatment plan requires subtype classification and probability threshold decision-making. Select a treatment plan according to the tinnitus subtype and tinnitus probability. The control center generates a decision instruction by parsing the parameters of the treatment plan, adapting the device, and converting the protocol, and performs security verification and conflict detection on the decision instruction.

[0036] It should be specifically noted in this embodiment that tinnitus subtypes include noise-induced tinnitus, Meniere's disease-related tinnitus, central tinnitus, and vascular tinnitus. The identification characteristics of noise-induced tinnitus are high-frequency hearing loss and a history of environmental noise exposure, and it is treated by sound therapy + noise protection. The identification characteristics of Meniere's disease-related tinnitus are fluctuating hearing loss, vertigo, and MRI endolymphatic hydrops, and it is treated by medication + vestibular rehabilitation training. The identification characteristics of central tinnitus are abnormal electroencephalogram gamma waves and no organic lesions, and it is treated by neuromodulation + cognitive behavioral therapy. The identification characteristics of vascular tinnitus are pulsatile tinnitus and MRA vascular malformations, and it is treated by vascular intervention + antihypertensive therapy. Tinnitus probabilities include high probability, medium probability, and low probability. For high probability, targeted treatment is immediately initiated. For medium probability, combined diagnosis and preventive intervention are adopted. For low probability, observation follow-up and environmental monitoring are adopted.

[0037] S5: Decision instruction transmission and plan execution: The control center sends decision instructions to the edge computing node, which are parsed into real-time treatment parameters and transmitted to the user terminal, and the user terminal executes treatment operations.

[0038] Furthermore, the parsing of real-time treatment parameters refers to parameter mapping. Parameter mapping means converting the received decision instructions into physical parameters that can be executed by the devices in the user terminal. The execution of treatment operations by the user terminal specifically refers to multi-device collaborative control.

[0039] It should be specifically noted in this embodiment that multi-device collaborative control can include intelligent earplugs, VR devices, and drug infusion systems. The intelligent earplugs generate masking sounds of matching frequencies and adjust the frequency every 10 minutes through an adaptive algorithm to prevent auditory adaptation. The VR device loads personalized scenarios and automatically switches to a relaxation mode according to the patient's real-time physiological data. The drug infusion system adjusts the drug dosage according to the gene detection results and ensures stable blood drug concentration through closed-loop control.

[0040] S6: Dynamic optimization and feedback: During the user's treatment process, treatment feedback data is collected in real time, processed and analyzed to obtain a treatment effect evaluation index, and the treatment plan is dynamically adjusted.

[0041] Furthermore, to obtain the treatment effect evaluation index, a time region needs to be set. This time region should be the time after the user executes the personalized diagnosis and treatment plan. The tinnitus handicap inventory (THI), heart rate variability (HRV), skin conductance (SC), treatment cycle (T), and resource consumption coefficient (C) within this time region are collected. The percentage of tinnitus reduction ΔTHI is obtained through the tinnitus handicap inventory using the formula ΔTHI = (THI i - THI) / THI i where THI iIndicates the tinnitus handicap index before treatment. The percentage improvement in the HRV stress index ΔHRV is calculated using heart rate variability. The percentage reduction in skin conductance ΔSC is obtained through skin conductance using the formula ΔSC = (SC i - SC) / SC i The percentage reduction in skin conductance is obtained. The percentage reduction in tinnitus, the percentage improvement in the HRV stress index, the percentage reduction in skin conductance, the treatment cycle, and the resource consumption coefficient are calculated through the formula: , The treatment effect evaluation index TEI is calculated. μ1, μ2, and μ3 represent the weight coefficients of the percentage reduction in tinnitus, the percentage improvement in the HRV stress index, and the percentage reduction in skin conductance respectively, and the sum is 1.

[0042] It should be specifically noted in this embodiment that THI is the core index to measure the impact of tinnitus on life, and its improvement directly reflects the treatment effectiveness. The weight should be the highest, reflecting the efficacy orientation centered on the patient's subjective experience. HRV reflects the autonomic balance, and its improvement indicates stress relief and physiological state optimization, which is an objective verification of the physiological mechanism effectiveness of the treatment plan. Skin conductance reflects the level of emotional arousal, and its reduction indicates reduced anxiety, which is a supplementary index for the psychological regulation effect of the treatment plan. The treatment cycle reflects the resource occupation time, and shortening the cycle can improve efficiency. Resource consumption includes drug dosage, equipment usage duration, etc., reflecting the treatment economy. The setting of the above weights should be determined according to the actual needs in clinical treatment.

[0043] Furthermore, for dynamically adjusting the treatment plan, a TEI warning line of TEI = 0.5 needs to be preset. When TEI < 0.5, it indicates that the treatment effect is unqualified, and strategies such as preferential adjustment of ΔTHI, collaborative intervention of ΔHRV, supplementary adjustment of ΔSC, shortening the treatment cycle T, and reducing resource consumption C are adopted to adjust the treatment plan.

[0044] It should be specifically noted in this embodiment that when ΔTHI does not meet the expectation, increase the sound treatment intensity or superimpose cognitive behavioral therapy. When ΔHRV stagnates, introduce biofeedback training or adjust the treatment method of drugs, such as deep breathing guidance or adding oryzanol to regulate the autonomic nerve. When ΔSC does not meet the standard, add a mindfulness meditation module or short-term anti-anxiety drugs. Identify patients with fast treatment response through an artificial intelligence prediction model, terminate the treatment in advance, and replace part of the drug treatment with VR cognitive training.

[0045] Secondly: In the accompanying drawings of the disclosed embodiments of the present invention, only the structures related to the disclosed embodiments are involved. For other structures, reference can be made to the general design. Without conflict, the same embodiment and different embodiments of the present invention can be combined with each other; Finally, the above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A personalized tinnitus treatment method based on artificial intelligence, characterized in that, It includes a user terminal, an edge computing node, a hospital terminal, a control center, and a cloud database. The user terminal and the edge computing node are connected via Bluetooth, and the edge computing node, the hospital terminal, the control center, and the cloud database are connected via the 5G communication protocol. It includes the following steps: S1: User data collection: The user terminal collects user data in real time. The user data includes tinnitus data, physiological data, and environmental data, and transmits the collected data to the edge computing node; S2: User data preprocessing: The edge computing node preprocesses the received user data, transmits the processed data to the control center, and caches historical data locally; S3: Sending a medical diagnosis request: The hospital terminal collects medical data and sends the medical data and the medical diagnosis request to the control center; S4: Comprehensive analysis and solution formulation: The control center comprehensively analyzes the processed user data and medical data, formulates a personalized tinnitus treatment plan, and generates a decision instruction according to the treatment plan; S5: Decision instruction transmission and solution execution: The control center sends the decision instruction to the edge computing node, which is parsed into real-time treatment parameters and transmitted to the user terminal, and the user terminal executes the treatment operation; S6: Dynamic optimization and feedback: During the user's treatment process, treatment feedback data is collected in real time, and the treatment feedback data is processed and analyzed to obtain a treatment effect evaluation index, and the treatment plan is dynamically adjusted.

2. The personalized tinnitus treatment method based on artificial intelligence according to claim 1, characterized in that: The tinnitus data specifically includes tinnitus frequency, tinnitus intensity, pitch, duration, and attack frequency. The physiological data specifically includes electroencephalogram signals, heart rate variability, and blood pressure. The environmental data specifically includes environmental noise intensity, noise frequency components, noise duration, temperature, humidity, and light intensity.

3. The personalized tinnitus treatment method based on artificial intelligence according to claim 1, characterized in that: The data preprocessing includes the following steps: A1: The edge computing node performs data cleaning and denoising on the received user data, including noise filtering and missing value filling; A2: Perform feature extraction and dimensionality reduction processing on the cleaned and denoised data; A3: Finally, perform data compression and block transmission on the data, desensitize sensitive data, and manage the local cache.

4. The personalized tinnitus treatment method based on artificial intelligence according to claim 1, characterized in that: The medical data includes patient basic data, clinical examination data, and auxiliary examination data. The patient basic data specifically includes identity identification, medical records, and current symptom information. The clinical examination data specifically includes tinnitus-specific data, physiological monitoring data, and environmental data. The auxiliary examination data specifically includes imaging materials and special detections.

5. The personalized tinnitus treatment method based on artificial intelligence according to claim 1, characterized in that: The medical diagnosis request needs to obtain diagnosis request parameters. The setting of the diagnosis request parameters is determined according to the request type, priority, and resource requirements, and the medical parameters and diagnostic data are formatted and processed, and a suitable transmission protocol is selected and transmitted to the control center.

6. The personalized tinnitus treatment method based on artificial intelligence according to claim 1, wherein: The comprehensive analysis specifically includes the following steps: B1: Perform feature engineering fusion on the received data. Feature engineering fusion specifically includes composite feature generation and multimodal fusion strategies; B2: Build and train an artificial intelligence model, select the algorithm of the model, build a network structure, and formulate a training strategy; B3: Through artificial intelligence models for real-time reasoning, dynamic decision-making, and causal reasoning, the tinnitus probability and tinnitus subtypes are obtained.

7. A personalized tinnitus treatment method based on artificial intelligence according to claim 1, characterized in that: The formulation of the personalized tinnitus treatment plan requires subtype classification and probability threshold decision-making. The treatment plan is selected based on the tinnitus subtype and tinnitus probability. The control center generates decision instructions by parsing the parameters of the treatment plan, adapting the device, and protocol conversion, and performs security verification and conflict detection on the decision instructions.

8. A personalized tinnitus treatment method based on artificial intelligence according to claim 1, characterized in that: The parsing of the real-time treatment parameters refers to parameter mapping. Parameter mapping means converting the received decision instructions into physical parameters executable by the devices in the user terminal. The execution of the treatment operation by the user terminal specifically refers to multi-device collaborative control.

9. The personalized tinnitus treatment method based on artificial intelligence according to claim 1, characterized in that: The acquisition of the treatment effect evaluation index requires setting a time period, which should be the time after the user executes the personalized diagnosis and treatment plan. Collect the tinnitus handicap inventory (THI), heart rate variability (HRV), skin conductance (SC), treatment cycle (T), and resource consumption coefficient (C) within this time period. By using the tinnitus handicap inventory, the formula ΔTHI = (THI i - THI) / THI i is used to obtain the tinnitus reduction percentage ΔTHI, where THI i represents the tinnitus handicap inventory before treatment. The improvement percentage of the HRV stress index ΔHRV is calculated using the heart rate variability. By using the skin conductance, the formula ΔSC = (SC i - SC) / SC i is used to obtain the skin conductance reduction percentage. The tinnitus reduction percentage, the improvement percentage of the HRV stress index, the skin conductance reduction percentage, the treatment cycle, and the resource consumption coefficient are combined through the formula: ， The treatment effect evaluation index TEI is calculated. μ1, μ2, and μ3 respectively represent the weight coefficients of the percentage of tinnitus reduction, the percentage of improvement in the HRV stress index, and the percentage of reduction in skin conductance, and the sum is 1.

10. A personalized tinnitus treatment method based on artificial intelligence according to claim 1, characterized in that: For the dynamic adjustment of the treatment plan, a TEI warning line of TEI = 0.5 is preset. When TEI < 0.5, it indicates that the treatment effect is unqualified, and strategies such as preferential adjustment of ΔTHI, collaborative intervention of ΔHRV, supplementary adjustment of ΔSC, shortening the treatment cycle T, and reducing resource consumption C are adopted to adjust the treatment plan.

Images