Large vestibular aqueduct syndrome prediction model construction method and prediction method

Abstract

The invention relates to the technical field of medical assistance, and discloses a large vestibular aqueduct syndrome prediction model construction method and a large vestibular aqueduct syndrome prediction method.The method comprises the steps that a medical history information data set, a hearing examination data set, an iconography examination data set and a genetic examination data set are obtained, based on the medical history information dataset, the hearing examination dataset, the iconography examination dataset and the genetic examination dataset, features related to the large vestibular aqueduct syndrome are extracted respectively, and feature datasets are obtained; and training a machine learning model based on the feature data set, and obtaining a large vestibular aqueduct syndrome prediction model after training is completed. The constructed large vestibular aqueduct syndrome prediction model can provide diagnosis help for subjects who have not clear diagnosis of the large vestibular aqueduct syndrome yet, reduce missed diagnosis and misdiagnosis of the large vestibular aqueduct syndrome caused by insufficient experience of clinicians, and can also provide diagnosis assistance for subjects who have already diagnosed as the large vestibular aqueduct syndrome. A prediction of hearing loss progression is provided.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method for constructing a prediction model for a large vestibular aqueduct syndrome and a prediction method. Background Art

[0002] Inner ear hearing diseases are a type of disease that affects human hearing function, mainly involving the loss or dysfunction of the inner ear structure. They can be caused by a variety of reasons, such as: genetic, congenital, traumatic, infection, etc. Hearing diseases can be mainly divided into conductive hearing loss, sensorineural hearing loss and mixed hearing loss. In addition, large vestibular aqueduct syndrome (LVAS) is an autosomal recessive genetic disease that belongs to the developmental malformation of the inner ear labyrinth. It is a more common type of sensorineural hearing loss. The clinical characteristics are progressive, fluctuating or delayed hearing loss, accounting for about 1% to 12% of the incidence of sensorineural hearing loss in children and adolescents. The disease can occur at any age from birth to adolescence, and it manifests suddenly or insidiously, and is prone to missed diagnosis and misdiagnosis in clinical practice.

[0003] However, it is still very challenging to detect large vestibular aqueduct syndrome in the early stages of the disease. The diagnosis requires the clinician's rich diagnostic and treatment experience. However, due to the clinician's lack of experience or limited examination methods, it is easy to miss or misdiagnose large vestibular aqueduct syndrome, delaying the timely treatment of children. Summary of the invention

[0004] In view of this, the present invention provides a method for constructing a prediction model for large vestibular aqueduct syndrome and a prediction method to solve the problem in the prior art that the diagnosis of large vestibular aqueduct syndrome is too dependent on the diagnostic experience of clinicians, which is prone to missed diagnosis or misdiagnosis.

[0005] In a first aspect, the present invention provides a method for constructing a prediction model for large vestibular aqueduct syndrome, the method comprising:

[0006] Acquire a medical history information data set, where the medical history information data set includes medical history information of several patients who have been ill;

[0007] Acquiring a hearing test data set, where the hearing test data set includes hearing data obtained after testing a patient according to a plurality of preset hearing test methods;

[0008] Acquiring an imaging examination data set, wherein the imaging examination data set includes imaging data obtained after detecting a sick patient according to a plurality of preset imaging detection methods;

[0009] Obtaining a genetic examination data set, where the genetic examination data set includes genetic data obtained after genetic testing of a patient with the disease;

[0010] Based on the medical history information dataset, the hearing test dataset, the imaging test dataset, and the genetic test dataset, features related to the large vestibular aqueduct syndrome are extracted to obtain feature datasets;

[0011] A machine learning model is trained based on the feature data set, and a large vestibular aqueduct syndrome prediction model is obtained after the training is completed. The large vestibular aqueduct syndrome prediction model is used to predict whether the subject to be diagnosed is likely to suffer from large vestibular aqueduct syndrome or predict the progression of hearing loss in the subject to be diagnosed based on at least one of the following of the subject to be diagnosed: medical history information, hearing data, imaging data, and genetic data.

[0012] The large vestibular aqueduct syndrome prediction model constructed in this embodiment can provide diagnostic assistance to subjects who have not yet been diagnosed with large vestibular aqueduct syndrome, reduce missed diagnosis and misdiagnosis of large vestibular aqueduct syndrome due to lack of experience of clinicians, and guide clinicians to provide a basis for further examination and a diagnostic plan. In addition, the large vestibular aqueduct syndrome prediction model constructed in this embodiment can also provide a prediction of the progression of hearing loss for subjects who have been diagnosed with large vestibular aqueduct syndrome, provide a guiding basis and treatment plan for clinical treatment and rehabilitation, and facilitate subject tracking services. Furthermore, the large vestibular aqueduct syndrome prediction model can also provide data guidance to help clinicians obtain more diagnosis, treatment and rehabilitation experience of large vestibular aqueduct syndrome from the analysis of big data, so that subjects can obtain more accurate medical services.

[0013] In an optional embodiment, the method further includes:

[0014] In the case where the patient already has a hearing aid, the hearing aid usage data of the patient already has a hearing aid is obtained, and the hearing aid usage data includes: hearing aid fitting time and hearing aid usage;

[0015] In the case of a patient who has a cochlear implant, obtaining the patient's cochlear implant usage data, the cochlear implant usage data including: cochlear implant time, cochlear implant usage effect;

[0016] Extracting hearing aid feature data based on hearing aid usage data, and using the hearing aid feature data to train a machine learning model;

[0017] Cochlear implant feature data is extracted based on the cochlear implant usage data, and the cochlear implant feature data is used to train a machine learning model.

[0018] Using the above multiple data types as datasets for training machine learning models can provide a more comprehensive understanding of the patient's condition, thereby improving the accuracy of large vestibular aqueduct syndrome prediction.

[0019] In an optional embodiment, the method further includes:

[0020] Obtain renal function test data from patients who are already ill;

[0021] Renal function feature data is extracted based on renal function test data, and the renal function feature data is used to train the machine learning model.

[0022] Examination of kidney function can help assess possible genetic tendencies in the family, and good or bad kidney function will affect the patient's overall health prognosis. Studying kidney function data can help doctors develop more effective long-term management plans.

[0023] In an optional implementation, after obtaining the hearing test data set, the method further includes:

[0024] Perform data cleaning on the hearing test data set, remove abnormal data, and obtain the cleaned hearing test data set;

[0025] Performing noise suppression and / or data segmentation on the cleaned hearing test data set to obtain a pre-processed hearing test data set;

[0026] The hearing test data set after the initial processing is subjected to data verification and standardization to obtain a unified hearing test data set, and the unified hearing test data set is used for feature extraction.

[0027] In this embodiment, preprocessing the hearing test data set can effectively improve the accuracy of building a large vestibular aqueduct syndrome prediction model.

[0028] In an optional embodiment, features related to large vestibular aqueduct syndrome are extracted based on an imaging examination dataset, including:

[0029] Identify imaging data of the vestibular aqueduct and endolymphatic sac included in the imaging study dataset;

[0030] Identify the vestibular aqueduct and determine the morphological parameters of the vestibular aqueduct;

[0031] Identify the endolymphatic sac and determine the morphological parameters of the endolymphatic sac;

[0032] The vestibular aqueduct morphological parameters and endolymphatic sac morphological parameters are used as features related to large vestibular aqueduct syndrome for training machine learning models to improve the comprehensiveness of machine learning model training.

[0033] In a second aspect, the present invention provides a method for predicting large vestibular aqueduct syndrome, the prediction method comprising:

[0034] Obtaining diagnostic information of the subject to be diagnosed, the diagnostic information including at least one of the following: medical history information, hearing data, imaging data, and genetic data;

[0035] The diagnostic information is input into a large vestibular aqueduct syndrome prediction model constructed according to any of the above-mentioned large vestibular aqueduct syndrome prediction model construction methods to predict whether the subject to be diagnosed is likely to suffer from large vestibular aqueduct syndrome or to predict the progression of hearing loss in the subject to be diagnosed.

[0036] In a third aspect, the present invention provides a device for constructing a prediction model for large vestibular aqueduct syndrome, the device comprising:

[0037] An acquisition module is used to acquire a medical history information data set, which includes medical history information of several patients who have already been ill; acquire a hearing test data set, which includes hearing data obtained after testing patients who have already been ill according to a plurality of preset hearing test methods; acquire an imaging test data set, which includes imaging data obtained after testing patients who have already been ill according to a plurality of preset imaging test methods; acquire a genetic test data set, which includes genetic data obtained after testing patients who have already been ill according to a plurality of preset genetic test methods;

[0038] An extraction module is used to extract features related to large vestibular aqueduct syndrome based on a medical history information data set, a hearing test data set, an imaging test data set, and a genetic test data set, respectively, to obtain a feature data set;

[0039] A model is constructed for training a machine learning model based on a feature data set. After the training is completed, a large vestibular aqueduct syndrome prediction model is obtained. The large vestibular aqueduct syndrome prediction model is used to predict whether the subject to be diagnosed may suffer from large vestibular aqueduct syndrome or predict the progression of hearing loss in the subject to be diagnosed based on at least one of the following: medical history information, hearing data, imaging data, and genetic data of the subject to be diagnosed.

[0040] In a fourth aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the method for constructing a prediction model for a large vestibular aqueduct syndrome according to the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0041] In a fifth aspect, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the method for constructing a prediction model for a large vestibular aqueduct syndrome according to the first aspect or any corresponding embodiment thereof.

[0042] In a sixth aspect, the present invention provides a computer program product, comprising computer instructions for causing a computer to execute the method for constructing a prediction model for a large vestibular aqueduct syndrome according to the first aspect or any corresponding embodiment thereof.

[0043] It should be noted that the large vestibular aqueduct syndrome prediction model construction device, computer device and computer readable storage medium provided by the present invention correspond to the large vestibular aqueduct syndrome prediction model construction method described above. Therefore, for the beneficial effects of the large vestibular aqueduct syndrome prediction model construction device, computer device and computer readable storage medium, please refer to the description of the corresponding beneficial effects of the large vestibular aqueduct syndrome prediction model construction method described above, which will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 is a flow chart of a method for constructing a prediction model for a large vestibular aqueduct syndrome according to an embodiment of the present invention;

[0046] Figure 2 is a schematic diagram of a process for constructing another vestibular aqueduct syndrome prediction model according to an embodiment of the present invention;

[0047] Figure 3 is a structural block diagram of a device for constructing a prediction model for a large vestibular aqueduct syndrome according to an embodiment of the present invention;

[0048] Figure 4 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0050] For patients with large vestibular aqueduct syndrome who have not yet developed the disease, the diagnosis needs to rely on the clinician's diagnostic and treatment experience. Due to the clinician's lack of experience or limited examination methods, it is easy to lead to missed diagnosis and misdiagnosis of large vestibular aqueduct syndrome.

[0051] In view of this, according to an embodiment of the present invention, an embodiment of a method for constructing a prediction model for a large vestibular aqueduct syndrome is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.

[0052] In this embodiment, a method for constructing a large vestibular aqueduct syndrome prediction model is provided, which can be executed by a server, a terminal, a mobile terminal, etc. Figure 1 is a flow chart of a method for constructing a large vestibular aqueduct syndrome prediction model according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0053] Step S101, obtaining a medical history information data set, the medical history information data set includes medical history information of several patients who have been ill. The medical history information may include the patient's name, gender, date of birth, perinatal medical history, family history, hearing screening results, first diagnosed age, deafness and dizziness history, growth and development, medication and other medical history, etc. Each of the several patients who have been ill corresponds to a medical history information subset, and several medical history information subsets constitute the medical history information data set.

[0054] Step S102, obtaining a hearing test data set, the hearing test data set includes hearing data obtained after testing the patient according to a plurality of preset hearing test methods. The hearing test methods may include pure tone audiometry, pediatric behavioral audiometry, acoustic impedance, electrophysiological testing (ABR, ASSR, CM, etc.), otoacoustic emissions, etc. Each patient may be tested using a plurality of hearing test methods, the hearing data detected by each hearing test method corresponds to a data set, each patient corresponds to a hearing subset composed of one or more data sets, and a plurality of hearing subsets constitute the hearing test data set.

[0055] Among them, the pure tone audiometry test method can play pure tones of different frequencies and intensities through headphones. The patient presses a button when hearing the sound, and the patient's reaction time range can be used as judgment information. The pediatric behavioral audiometry test method can use games or other stimuli (such as sound or visual cues) to guide the patient to respond, such as turning the head, looking or other behavioral responses, and record the time range of the response. The acoustic impedance test method can insert a probe through the ear canal to measure the mobility of the eardrum and the middle ear pressure. In electrophysiological examinations, the ABR (a method for testing auditory brainstem responses) test method can place electrodes on the scalp to record the electrical activity of the brainstem after auditory stimulation. The ASSR (a method for testing auditory steady-state evoked responses) test method is similar to the ABR, but uses continuous tone stimulation to produce stable electrophysiological responses. The CM (a method for testing otoacoustic emissions) test method can use weak audio stimulation to record the weak current response generated by the cochlea. The otoacoustic emission test method can send audio stimulation through the ear canal to detect whether the cochlea produces reflected sound waves.

[0056] Step S103, obtaining an imaging examination data set, the imaging examination data set includes imaging data obtained after detecting a sick patient according to a plurality of preset imaging detection methods. The imaging detection method may include computed tomography (CT), magnetic resonance imaging (MRI), etc. Each sick patient corresponds to an imaging data subset, and a plurality of imaging data subsets constitute the imaging examination data set.

[0057] Step S104, obtaining a genetic examination data set, the genetic examination data set includes genetic data obtained after genetic testing of patients with the disease. The genetic data may include: large vestibular aqueduct syndrome related genes, large vestibular aqueduct syndrome non-related genes, and normal genes.

[0058] Step S105 , based on the medical history information data set, the hearing test data set, the imaging test data set, and the genetic test data set, respectively extract features related to the large vestibular aqueduct syndrome to obtain a feature data set.

[0059] In this embodiment, the hearing test data set and the imaging test data set are mainly used as the basis, and other test data sets are used as references for prediction. It is necessary to select the features most relevant to the diagnosis of large vestibular aqueduct syndrome. For example, for the imaging test data set, these features may include the morphological parameters of the vestibular aqueduct, the degree and type of hearing loss, etc. As for feature extraction, for image data, computer vision technology can be used to extract key features in the image, such as the width and shape of the vestibular aqueduct. The image data features are combined with the hearing data features such as audiology pure tone audiometry to form a set of data factors, and then a certain weight is given to this factor, which can be used to predict whether the object to be diagnosed is normal.

[0060] More specifically, for the bone defect shadow in the CT image data features, that is, when the vestibular aqueduct is enlarged, at the level of the horizontal semicircular canal or the common bone crus, a trumpet-shaped or cone-shaped bone defect shadow can be seen at the posterior edge of the petrous cone, with sharp and clear edges, and its inner opening is often connected to the vestibule, which can be used as a key data factor; and possible special types, such as the vestibular aqueduct is enlarged and connected to the common crus; the vestibular aqueduct is enlarged in a fissure shape with bone defects; the enlarged aqueduct is in the shape of an "inverted fishhook"; the vestibular aqueduct is less than 1.5mm wide, but it can be observed that the vestibular aqueduct is connected to the common crus of the semicircular canals, and the vestibular aqueduct is enlarged with Mondini (congenital cochlear) malformation, etc., as another set of data factors. This type of data factor is integrated with the test subject's genetics, embryonic development, increased intracranial pressure, external inducements, and physiological variations. These factors may act alone or together, and can be called specific data factors here. This specific data factor can be a dynamically adjustable variable or a constant, and can be expressed as a two-dimensional array or a three-dimensional array. The two-dimensional array expresses a single-channel grayscale image, while the three-dimensional array expresses a multi-channel RGB color image. It can also be considered that the two-dimensional or three-dimensional array here can be used as a convolution kernel for subsequent information processing. The application of this factor is used to determine whether there is an abnormality in the shape and size of the vestibular aqueduct, thereby affecting the prediction of whether the object to be diagnosed is normal.

[0061] Step S106, training a machine learning model based on the feature data set, and obtaining a large vestibular aqueduct syndrome prediction model after the training is completed. The large vestibular aqueduct syndrome prediction model is used to predict whether the subject to be diagnosed may suffer from large vestibular aqueduct syndrome or predict the progression of hearing loss of the subject to be diagnosed based on at least one of the following of the subject to be diagnosed: medical history information, hearing data, imaging data, and genetic data.

[0062] Specifically, the data in the feature data set is organized into a data set suitable for machine learning algorithms, and the data can be divided into a training set, a validation set, and a test set. For the diagnosis of large vestibular aqueduct syndrome, a suitable machine learning algorithm can be selected. The algorithm may include logistic regression, support vector machine (SVM), random forest, etc. Considering that the diagnosis of large vestibular aqueduct syndrome may involve complex nonlinear relationships, deep learning algorithms (such as convolutional neural network CNN) are preferred, which are suitable for processing image data. Machine learning models can choose algorithms such as random forest, C5.0 decision tree, K nearest neighbor, BP neural network, etc. Specifically, a multi-layer CNN model can be designed, including convolutional layer, pooling layer, fully connected layer, etc. The convolutional layer is responsible for extracting image features, the pooling layer is used to reduce dimensions and prevent overfitting, and the fully connected layer is used for classification. Then select appropriate activation functions (such as ReLU) and loss functions (such as cross entropy loss function) to optimize the performance of the model.

[0063] When using the convolutional layer of CNN to extract features from the preprocessed image, local features related to vestibular aqueduct syndrome, such as the shape, size, and position of the vestibular aqueduct, can be extracted by designing appropriate convolution kernels and convolutional layer structures.

[0064] For the convolution kernel and convolution layer structure, it is a small weight matrix used to perform sliding convolution operations on the input image to extract features. The size of the convolution kernel is usually 3×3, 5×5 or 7×7, but it can also be other sizes. The weight of the convolution kernel is learned through the training process. For example, a 3×3 convolution kernel can be used for edge detection because it is sensitive to vertical and horizontal edges, especially the special images mentioned above (enlarged vestibular aqueduct, enlarged vestibular aqueduct in the shape of a fissure, with bone defects, enlarged aqueduct in the shape of an "inverted fishhook", vestibular aqueduct width is too small, vestibular aqueduct is connected to the common crus of the semicircular canals, vestibular aqueduct enlargement with Mondini deformity, etc.) and the corresponding specific data factors. When it slides on the input image, it will calculate the weighted sum of the pixels in the current window, thereby highlighting the edges in the image, and has a very good recognition effect for the above-mentioned special images.

[0065] For each convolution kernel, the convolution operation generates a feature map. In this embodiment, the input image size is 6×6, the convolution kernel size is 3×3, the step size is 1, and the padding is 0. Therefore, the size of the generated feature map is (6-3+1)×(6-3+1)=4×4. After applying the edge detection convolution kernel, a feature map is obtained, which highlights the edge features in the input image.

[0066] The convolution layer usually contains multiple convolution kernels, each of which generates a feature map. At least one convolution kernel is retained to operate with the specific data factor to obtain a specific feature map, which serves as key information or important reference information.

[0067] Feature analysis is to further analyze and process the extracted features, such as using a pooling layer for dimensionality reduction and a fully connected layer for classification or regression. Through these operations, useful information about vestibular aqueduct syndrome can be obtained, including specific feature map information obtained by combining specific data factors as key information or important reference information.

[0068] In this way, the convolution kernel is a small weight matrix used to perform sliding convolution operations on the image. The convolution layer contains multiple convolution kernels, each of which independently extracts features and generates feature maps. By stacking multiple convolution layers and other types of layers (such as pooling layers, fully connected layers, etc.), CNN can build the image processing model used.

[0069] Use the training set data to train the selected machine learning algorithm and optimize the performance of the model by adjusting the parameters and structure of the model. During the training process, techniques such as cross-validation can be used to evaluate the generalization ability of the model.

[0070] Use validation set data to evaluate the performance of the trained model. Common evaluation indicators include accuracy, recall, F1 score, etc. In addition, confusion matrix can be used to visualize the classification performance of the model. Optimize the model based on the evaluation results, such as adjusting hyperparameters, improving feature extraction methods, adjusting network structure, adding data enhancement technology, etc. Data enhancement can increase the diversity of data through rotation, translation, scaling, etc., thereby improving the generalization ability of the model.

[0071] The trained CNN model is then applied to the test set to evaluate the performance of the model on unknown data. If the performance is good, the model can be considered for application in actual clinical diagnosis. The prediction results of the model are analyzed, including the accuracy, false positive rate, and false negative rate of identifying vestibular aqueduct enlargement. These analysis results can provide doctors with valuable auxiliary diagnosis information.

[0072] If the trained and optimized CNN model achieves an accuracy rate of more than 90% on the test set, in practical applications, the model can successfully identify the enlarged vestibular aqueduct features of most patients with large vestibular aqueduct syndrome, providing doctors with timely diagnostic support. By building and training a suitable large vestibular aqueduct syndrome model, optimizing and evaluating it, doctors can be provided with accurate and efficient auxiliary diagnostic tools.

[0073] Finally, the trained multivariate diagnostic prediction model for large vestibular aqueduct syndrome is deployed in practical applications to assist doctors in diagnosing large vestibular aqueduct syndrome. In addition, new data can be collected regularly and the model can be updated to maintain the accuracy and timeliness of the model.

[0074] The large vestibular aqueduct syndrome prediction model constructed in this embodiment can provide diagnostic assistance to subjects who have not yet been diagnosed with large vestibular aqueduct syndrome, reduce missed diagnosis and misdiagnosis of large vestibular aqueduct syndrome due to lack of experience of clinicians, and guide clinicians to provide a basis for further examination and a diagnostic plan. In addition, the large vestibular aqueduct syndrome prediction model constructed in this embodiment can also provide a prediction of the progression of hearing loss for subjects who have been diagnosed with large vestibular aqueduct syndrome, provide a guiding basis and treatment plan for clinical treatment and rehabilitation, and facilitate subject tracking services. Furthermore, the large vestibular aqueduct syndrome prediction model can also provide data guidance to help clinicians obtain more diagnosis, treatment and rehabilitation experience of large vestibular aqueduct syndrome from the analysis of big data, so that subjects can obtain more accurate medical services.

[0075] In some optional embodiments, the method for constructing a prediction model for large vestibular aqueduct syndrome further comprises:

[0076] In the case where the ill patient wears a hearing aid, the hearing aid usage data of the ill patient is obtained, and the hearing aid usage data includes: hearing aid fitting time and hearing aid usage status.

[0077] In the case of a cochlear implanted patient, the cochlear implant usage data of the patient is obtained, and the cochlear implant usage data includes: cochlear implant implant time and cochlear implant usage effect.

[0078] Hearing aid feature data is extracted based on the hearing aid usage data, and the hearing aid feature data is used to train the machine learning model.

[0079] Cochlear implant feature data is extracted based on the cochlear implant usage data, and the cochlear implant feature data is used to train a machine learning model.

[0080] In some optional embodiments, the method for constructing a prediction model for large vestibular aqueduct syndrome further comprises:

[0081] Obtain renal function test data from patients who are already ill;

[0082] Renal function feature data is extracted based on renal function test data, and the renal function feature data is used to train the machine learning model.

[0083] In addition to medical history information, audiological examination results, imaging examination results, and genetic examination results, data collection can also include other examination data, such as rehabilitation strategy and rehabilitation status, hearing aid use data, and cochlear implant use data. In addition, it can also include renal function test data and other department examination results data.

[0084] Using the above data types as data sets for training machine learning models can provide a more comprehensive understanding of the patient's condition, thereby improving the accuracy of large vestibular aqueduct syndrome prediction. In addition, large vestibular aqueduct syndrome is an autosomal recessive genetic disease. Examination of renal function can help assess possible genetic tendencies in the family. Whether renal function is good or not will affect the patient's overall health prognosis. Learning renal function data can help doctors develop more effective long-term management plans.

[0085] In some optional implementations, after obtaining the hearing test data set, the method further includes:

[0086] The hearing test data set is cleaned to remove abnormal data and obtain a cleaned hearing test data set. Data cleaning of the hearing test data set can check and remove abnormal data points caused by equipment failure, operating errors or abnormal reactions of subjects.

[0087] The cleaned hearing test data set is subjected to noise suppression and / or data segmentation to obtain the initially processed hearing test data set. For data containing noise, an appropriate noise suppression algorithm (such as wavelet transform, adaptive filtering, etc.) can be used to reduce the impact of noise on data analysis. For continuous data streams (such as electrophysiological test data), they can be divided into smaller data segments or windows as needed for more detailed analysis.

[0088] The hearing test data set after the initial processing is subjected to data verification and standardization to obtain a unified hearing test data set, which is used for feature extraction. Data verification can effectively check the accuracy and consistency of data records. Data from different sources and dimensions are converted into a unified standard format for comparison and analysis. For example, all hearing thresholds can be converted into decibel (dB) units.

[0089] In addition, if the amount of data is small, you can consider increasing the amount of data through data enhancement techniques (such as resampling, noise addition, etc.) to improve the generalization ability of the model.

[0090] Regarding feature extraction of hearing test data sets, meaningful features can be extracted from the original data, such as hearing threshold, reaction time, waveform characteristics, etc. These features will be used for subsequent data analysis and model training.

[0091] In this embodiment, preprocessing the hearing test data set can effectively improve the accuracy of building a large vestibular aqueduct syndrome prediction model.

[0092] In some optional embodiments, features related to large vestibular aqueduct syndrome are extracted based on an imaging examination dataset, including:

[0093] The imaging data sets included the vestibular aqueduct and endolymphatic sac.

[0094] Identify the vestibular aqueduct and determine the morphological parameters of the vestibular aqueduct.

[0095] Identify the endolymphatic sac and determine the morphological parameters of the endolymphatic sac.

[0096] Vestibular aqueduct morphological parameters and endolymphatic sac morphological parameters were considered as features associated with large vestibular aqueduct syndrome.

[0097] In this embodiment, high-resolution image data including the vestibular aqueduct and the endolymphatic sac, such as CT or MRI scan images, can be obtained, and then the clearest image showing the vestibular aqueduct and the endolymphatic sac can be selected, and then image recognition can be performed.

[0098] Before identification, the image can be pre-processed, such as denoising, standardization and cropping, to ensure image quality and highlight the key features of the vestibular aqueduct. Then in the pre-processed image, the position of the vestibular aqueduct is identified, and the measured values ​​of the vestibular aqueduct and the endolymphatic sac are extracted, which are the vestibular aqueduct morphological parameters in this embodiment. In this embodiment, professional image analysis software can be used to measure the width of the vestibular aqueduct, especially the inner diameter width (MDVA) of the aqueduct at the midpoint between the outer opening of the vestibular aqueduct and the common bone foot, and the outer opening width (ODVA) of the vestibular aqueduct. The normal vestibular aqueduct distal end inner diameter width is 0.4mm-1.0mm. If MDVA>1.5mm or ODVA>2.0mm, it can be diagnosed as vestibular aqueduct enlargement. At the same time, the endolymphatic sac is identified, which is located outside the vestibular aqueduct and is connected to the vestibular aqueduct. On the image, it can be presented as a specific shape and density. Image analysis software can also be used to measure the endolymphatic sac, such as measuring its size, morphology and other parameters. The measured parameters of the vestibular aqueduct and endolymphatic sac are recorded in detail, including the measurement location, value, etc. The measurement data can be statistically analyzed to compare the data differences between different individuals or at different time points to assist in the diagnosis and treatment of the disease. Ultimately, the vestibular aqueduct morphological parameters and endolymphatic sac morphological parameters are used as features related to large vestibular aqueduct syndrome for the training of machine learning models to improve the comprehensiveness of machine learning model training.

[0099] Reference Figure 2 As shown in the figure, in addition to medical history information, audiological examination results, imaging examination results, and genetic examination results, data collection can also include other examination data. After storing these data in the server, the data are preprocessed, including analyzing the reliability of the data and eliminating erroneous data. It mainly includes image enhancement, denoising, and standardization and normalization of the data, as well as outlier detection and missing value processing. The preprocessed vestibular aqueduct-related data and other related information are then feature extracted to form a complete feature data set, and the data is stored. The data storage can choose a suitable data storage format and method, such as CSV files, databases, etc., for subsequent model training and statistical analysis. The finally trained large vestibular aqueduct syndrome prediction model can output the large vestibular aqueduct syndrome prediction results, hearing loss progression, and large vestibular aqueduct syndrome big data statistical analysis results based on the input medical history information, hearing data, imaging data, genetic data, etc. of the patient to be diagnosed, which can provide more accurate auxiliary medical services for the subjects.

[0100] In this embodiment, a method for predicting large vestibular aqueduct syndrome is also provided, which can be executed by a server, a terminal, a mobile terminal and other devices. The prediction method includes:

[0101] Obtaining diagnostic information of the subject to be diagnosed, the diagnostic information including at least one of the following: medical history information, hearing data, imaging data, and genetic data;

[0102] The diagnostic information is input into the large vestibular aqueduct syndrome prediction model constructed according to the large vestibular aqueduct syndrome prediction model construction method described in any of the above embodiments to predict whether the subject to be diagnosed is likely to suffer from large vestibular aqueduct syndrome or to predict the progression of hearing loss in the subject to be diagnosed.

[0103] For details on the construction of the large vestibular aqueduct syndrome prediction model, please refer to any of the above implementation methods, which will not be repeated here.

[0104] The large vestibular aqueduct syndrome prediction model constructed in this embodiment can provide diagnostic assistance to subjects who have not yet been diagnosed with large vestibular aqueduct syndrome, reduce missed diagnosis and misdiagnosis of large vestibular aqueduct syndrome due to lack of experience of clinicians, and guide clinicians to provide a basis for further examination and a diagnostic plan. In addition, the constructed large vestibular aqueduct syndrome prediction model can also provide a prediction of the progression of hearing loss for subjects who have been diagnosed with large vestibular aqueduct syndrome, provide a guiding basis and treatment plan for clinical treatment and rehabilitation, and facilitate subject tracking services. Furthermore, the large vestibular aqueduct syndrome prediction model can also provide data guidance to help clinicians obtain more diagnosis, treatment and rehabilitation experience of large vestibular aqueduct syndrome from the analysis of big data, so that subjects can obtain more accurate medical services.

[0105] In this embodiment, a device for constructing a prediction model for a large vestibular aqueduct syndrome is also provided, and the device is used to implement the above-mentioned embodiments and preferred embodiments, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0106] This embodiment provides a device for constructing a prediction model for large vestibular aqueduct syndrome. Figure 3 As shown, the device comprises:

[0107] The acquisition module 201 is used to acquire a medical history information data set, which includes the medical history information of several existing patients; acquire a hearing test data set, which includes hearing data obtained after detecting existing patients according to multiple preset hearing test methods; acquire an imaging test data set, which includes imaging data obtained after detecting existing patients according to multiple preset imaging test methods; acquire a genetic test data set, which includes genetic data obtained after detecting existing patients according to multiple preset genetic test methods; in the case where the existing patient wears a hearing aid, acquire the hearing aid usage data of the existing patient, the hearing aid usage data includes: hearing aid fitting time, hearing aid usage; in the case where the existing patient has a cochlear implant, acquire the cochlear implant usage data of the existing patient, the cochlear implant usage data includes: cochlear implant implant time, cochlear implant usage effect; acquire renal function test data of the existing patient.

[0108] The extraction module 202 is used to extract features related to the large vestibular aqueduct syndrome based on the medical history information data set, the hearing test data set, the imaging test data set, and the genetic test data set to obtain a feature data set; it is also used to extract hearing aid feature data based on the hearing aid usage data, and the hearing aid feature data is used to train the machine learning model; it is also used to extract cochlear implant feature data based on the cochlear implant usage data, and the cochlear implant feature data is used to train the machine learning model. It is also used to extract renal function feature data based on the renal function test data, and the renal function feature data is used to train the machine learning model.

[0109] A model 203 is constructed for training a machine learning model based on a feature data set. After the training is completed, a large vestibular aqueduct syndrome prediction model is obtained. The large vestibular aqueduct syndrome prediction model is used to predict whether the subject to be diagnosed is likely to suffer from large vestibular aqueduct syndrome or predict the progression of hearing loss in the subject to be diagnosed based on at least one of the following of the subject to be diagnosed: medical history information, hearing data, imaging data, and genetic data.

[0110] In some optional embodiments, the device further comprises:

[0111] The preprocessing module is used to clean the hearing test data set, remove abnormal data, and obtain a cleaned hearing test data set; perform noise suppression and / or data segmentation on the cleaned hearing test data set to obtain a pre-processed hearing test data set; perform data verification and standardization on the pre-processed hearing test data set to obtain a unified hearing test data set, and the unified hearing test data set is used for feature extraction.

[0112] In some optional implementations, the extraction module 202 includes:

[0113] The extraction unit is used to determine the image data of the vestibular aqueduct and the endolymphatic sac contained in the imaging examination data set; identify the vestibular aqueduct and determine the morphological parameters of the vestibular aqueduct; identify the endolymphatic sac and determine the morphological parameters of the endolymphatic sac; and use the morphological parameters of the vestibular aqueduct and the morphological parameters of the endolymphatic sac as features related to large vestibular aqueduct syndrome.

[0114] The device for constructing a prediction model for greater vestibular aqueduct syndrome in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0115] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0116] The embodiment of the present invention also provides a computer device having the above Figure 3 The apparatus for building a predictive model for large vestibular aqueduct syndrome is shown.

[0117] See also Figure 4 , Figure 4 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 4 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 4 A processor 10 is taken as an example.

[0118] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0119] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0120] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0121] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0122] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0123] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0124] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0125] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for constructing a prediction model for large vestibular aqueduct syndrome, characterized in that: The method comprises: Acquire a medical history information data set, wherein the medical history information data set includes medical history information of a number of patients who have been ill; Acquiring a hearing test data set, wherein the hearing test data set includes hearing data obtained after testing the patient according to a plurality of preset hearing test methods; Acquiring an imaging examination data set, wherein the imaging examination data set includes imaging data obtained after detecting the sick patient according to a plurality of preset imaging detection methods; Acquiring a genetic examination data set, wherein the genetic examination data set includes genetic data obtained after genetic testing of the patient; Based on the medical history information dataset, the hearing test dataset, the imaging test dataset, and the genetic test dataset, respectively extracting features related to large vestibular aqueduct syndrome to obtain a feature dataset; A machine learning model is trained based on the feature data set, and a large vestibular aqueduct syndrome prediction model is obtained after the training is completed. The large vestibular aqueduct syndrome prediction model is used to predict whether the subject to be diagnosed is likely to suffer from large vestibular aqueduct syndrome or predict the progression of hearing loss in the subject to be diagnosed based on at least one of the following of the subject to be diagnosed: medical history information, hearing data, imaging data, and genetic data.

2. The method according to claim 1, characterized in that The method further comprises: In the case where the ill patient wears a hearing aid, obtaining hearing aid usage data of the ill patient, the hearing aid usage data including: hearing aid fitting time and hearing aid usage; In the case where the patient has a cochlear implant, obtaining the cochlear implant usage data of the patient, wherein the cochlear implant usage data includes: cochlear implant time and cochlear implant usage effect; extracting hearing aid feature data based on the hearing aid usage data, wherein the hearing aid feature data is used to train the machine learning model; Cochlear implant feature data is extracted based on the cochlear implant usage data, and the cochlear implant feature data is used to train the machine learning model.

3. The method according to claim 1, characterized in that: The method further comprises: Obtaining renal function test data of the ill patient; Renal function characteristic data is extracted based on the renal function test data, and the renal function characteristic data is used to train the machine learning model.

4. The method according to claim 1, characterized in that After obtaining the hearing test data set, the method further includes: Performing data cleaning on the hearing test data set to remove abnormal data and obtain a cleaned hearing test data set; performing noise suppression and / or data segmentation on the cleaned hearing test data set to obtain a pre-processed hearing test data set; The initially processed hearing test data set is subjected to data verification and standardization to obtain a unified hearing test data set, and the unified hearing test data set is used for feature extraction.

5. The method according to claim 1, characterized in that: The method of extracting features related to large vestibular aqueduct syndrome based on the imaging examination dataset includes: determining the image data of the vestibular aqueduct and the endolymphatic sac included in the imaging examination data set; Identify the vestibular aqueduct and determine the morphological parameters of the vestibular aqueduct; Identify the endolymphatic sac and determine the morphological parameters of the endolymphatic sac; The vestibular aqueduct morphological parameters and the endolymphatic sac morphological parameters are used as features associated with large vestibular aqueduct syndrome.

6. A method for predicting large vestibular aqueduct syndrome, characterized in that: The method comprises: Acquiring diagnostic information of the subject to be diagnosed, wherein the diagnostic information includes at least one of the following: medical history information, hearing data, imaging data, and genetic data; The diagnostic information is input into a large vestibular aqueduct syndrome prediction model constructed according to the large vestibular aqueduct syndrome prediction model construction method according to any one of claims 1 to 5 to predict whether the subject to be diagnosed is likely to suffer from large vestibular aqueduct syndrome or to predict the progression of hearing loss in the subject to be diagnosed.

7. A device for constructing a prediction model for large vestibular aqueduct syndrome, characterized in that: The device comprises: an acquisition module, for acquiring a medical history information data set, wherein the medical history information data set includes medical history information of a number of patients who have already been ill; acquiring a hearing test data set, wherein the hearing test data set includes hearing data obtained after testing the patients who have already been ill according to a plurality of preset hearing test methods; acquiring an imaging test data set, wherein the imaging test data set includes imaging data obtained after testing the patients who have already been ill according to a plurality of preset imaging test methods; acquiring a genetic test data set, wherein the genetic test data set includes genetic data obtained after testing the patients who have already been ill according to a plurality of preset genetic test methods; An extraction module, used to extract features related to large vestibular aqueduct syndrome based on the medical history information data set, the hearing test data set, the imaging test data set, and the genetic test data set, respectively, to obtain a feature data set; A model is constructed for training a machine learning model based on the feature data set, and a large vestibular aqueduct syndrome prediction model is obtained after the training is completed. The large vestibular aqueduct syndrome prediction model is used to predict whether the subject to be diagnosed is likely to suffer from large vestibular aqueduct syndrome or predict the progression of hearing loss of the subject to be diagnosed based on at least one of the following of the subject to be diagnosed: medical history information, hearing data, imaging data, and genetic data.

8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for constructing a large vestibular aqueduct syndrome prediction model as described in any one of claims 1 to 5 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the method for constructing a large vestibular aqueduct syndrome prediction model as described in any one of claims 1-5.

10. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the method for constructing a large vestibular aqueduct syndrome prediction model according to any one of claims 1 to 5.

Images