An ear-related vertigo diagnosis and treatment device, system and big data analysis platform

Through the diagnosis and treatment device and big data analysis platform of ear-borne vertigo, the diagnosis and treatment plan generation model is used to solve the complex problems of the diagnosis and treatment process of ear-borne vertigo, and the automatic diagnosis and treatment plan generation is realized. It is suitable for grassroots hospitals and simplified The diagnosis and treatment process was completed.

CN112037908BActive Publication Date: 2024-11-12EYE & ENT HOSPITAL SHANGHAI MEDICAL SCHOOL FUDAN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202010778178.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-08-05
Publication Date
2024-11-12
Estimated Expiration
2040-08-05

AI Technical Summary

Technical Problem

The diagnosis and treatment process of ear-borne vertigo is complex and requires the collaboration of multiple specialists. Especially in primary hospitals, it is difficult to conduct reasonable diagnosis and treatment due to medical level and equipment limitations.

Method used

Design an ear-borne vertigo diagnosis and treatment device and big data analysis platform, use diagnostic models and treatment plan generation models to process patients' vertigo information, and obtain the probability of illness and treatment plan, including the vertigo information acquisition module and the probability of illness acquisition module , treatment plan generation module and model construction module, to achieve automated diagnosis and treatment plan generation through speech recognition, natural language processing and big data analysis.

Benefits of technology

This device can basically eliminate the need for manual participation, liberate the diagnosis and treatment process of medical staff, simplify the diagnosis and treatment process of ear-borne vertigo, reduce the requirements for the level of medical staff, and is suitable for grassroots hospitals.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112037908B_ABST
    Figure CN112037908B_ABST
Patent Text Reader

Abstract

The present invention provides an ear-related vertigo diagnosis and treatment device, system and big data analysis platform. The ear-related vertigo diagnosis and treatment device includes: a dizziness information acquisition module, which is used to obtain the patient's dizziness information; a disease probability acquisition module, which is connected to the dizziness information acquisition module and is used to process the patient's dizziness information using a diagnosis model to obtain the patient's disease probability; a treatment plan generation module, which is connected to the dizziness information acquisition module and the disease probability acquisition module, and is used to process the patient's dizziness information and the patient's disease probability using a treatment plan generation model to generate a treatment plan for the patient; the diagnosis model and the treatment plan generation model are big data models constructed based on the historical diagnosis and treatment data of multiple patients. The ear-related vertigo diagnosis and treatment device can obtain the patient's dizziness information and automatically generate a treatment plan based on the dizziness information.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of computer-aided diagnosis, and relates to a diagnostic and treatment device, and in particular to an otopathic vertigo diagnostic and treatment device, system and big data analysis platform. Background Art

[0002] Otogenic vertigo is one of the common diseases in the otolaryngology department. It has a high incidence rate and great harm. It is mainly characterized by rotational vertigo and involves multiple organs and systems such as the central nervous system, sensory system, and motor system. Otogenic vertigo is often accompanied by functional damage to the vestibular organs of the ear. Even if the correct medication is used in time, some functions will not be fully restored. The patient's balance dysfunction is the biggest obstacle to the quality of life of vertigo patients. Scientific rehabilitation training can effectively enable patients to fully recover to normal life through functional compensation, and is an extremely effective auxiliary treatment for the sequelae of vertigo.

[0003] However, the inventors found in actual application that the clinical diagnosis of otogenic vertigo often requires imaging and neuroelectrophysiological examinations to make a clear diagnosis, and the treatment often requires the collaboration of multiple specialties such as otolaryngology, neurology, geriatrics, rehabilitation, and nursing. The treatment process is complicated and requires a high level of medical staff. Especially for grassroots hospitals, due to the limitations of medical level and medical equipment, it is difficult for grassroots hospitals to conduct reasonable diagnosis and treatment of patients' otogenic vertigo. Summary of the invention

[0004] In view of the shortcomings of the prior art mentioned above, the purpose of the present invention is to provide an ear-related vertigo diagnosis and treatment device, system and big data analysis platform to solve the problem that the diagnosis and treatment process is complicated and requires a high level of medical personnel in the existing treatment schemes for ear-related vertigo.

[0005] To achieve the above-mentioned purpose and other related purposes, the first aspect of the present invention provides an ear-induced vertigo diagnosis and treatment device; the ear-induced vertigo diagnosis and treatment device includes: a vertigo information acquisition module, used to obtain the patient's vertigo information; an illness probability acquisition module, connected to the vertigo information acquisition module, used to use a diagnostic model to process the patient's vertigo information to obtain the patient's illness probability; a treatment plan generation module, connected to the vertigo information acquisition module and the illness probability acquisition module, used to use a treatment plan generation model to process the patient's vertigo information and the patient's illness probability to generate a treatment plan for the patient; wherein the diagnostic model and the treatment plan generation model are big data models constructed based on historical diagnosis and treatment data of multiple patients.

[0006] In an embodiment of the first aspect, the ear-induced vertigo diagnosis and treatment device also includes: a dizziness questionnaire generation module, used to generate an ear-induced vertigo questionnaire; the ear-induced vertigo questionnaire is used to assist the patient in inputting questionnaire information; and the dizziness information acquisition module acquires the patient's dizziness information based on the questionnaire information.

[0007] In an embodiment of the first aspect, the questionnaire information is questionnaire information in voice form; the dizziness information acquisition module includes: a voice recognition unit, used to convert the questionnaire information in voice form into a text sequence; a natural language processing unit, connected to the voice recognition unit, used to perform natural language processing on the text sequence to obtain key text in the text sequence; a questionnaire matching unit, connected to the natural language processing unit and the dizziness questionnaire generation module, used to match the key text with the ear-related dizziness questionnaire to obtain the patient's dizziness information.

[0008] In an embodiment of the first aspect, the ear-induced vertigo diagnosis and treatment device also includes a vertigo detection module; wherein the vertigo detection module is connected to the vertigo information acquisition module, and is used to perform vertigo detection on the patient to obtain the patient's vertigo detection results; the patient's vertigo detection results include vestibular evoked myogenic potentials, acoustic impedance test results, pure tone audiometry results, video nystagmus, otolith examination results, head thrust test results and / or dynamic and static balance platform test results; the vertigo information acquisition module obtains the patient's vertigo information based on the patient's vertigo detection results.

[0009] In an embodiment of the first aspect, the diagnosis model and / or the treatment plan generation model is generated by a model construction module, and the model construction module includes: a data acquisition unit, used to acquire the historical diagnosis and treatment data of the multiple patients; a feature extraction unit, connected to the data acquisition unit, used to extract feature data from the historical diagnosis and treatment data of the multiple patients; a model training unit, connected to the feature extraction unit, used to train a big data model according to the feature data to obtain the diagnosis model and / or the treatment plan generation model.

[0010] In an embodiment of the first aspect, the model training unit includes: a model training subunit, connected to the feature extraction unit, and used to train at least two big data models based on the feature data to obtain at least two alternative models; a model selection subunit, connected to the model training subunit, and used to select the diagnostic model from the alternative models based on the performance of each of the alternative models.

[0011] In an embodiment of the first aspect, the model building module is included in the ear-induced vertigo diagnosis and treatment device, and is connected to the disease probability acquisition module and the treatment plan generation module.

[0012] In an embodiment of the first aspect, the ear-related vertigo diagnosis and treatment device also includes: a medical advice generation module, which is connected to the treatment plan generation module and is used to generate a medical advice based on the patient's treatment plan.

[0013] The second aspect of the present invention provides a big data analysis platform; the big data analysis platform includes: a storage module, which stores historical diagnosis and treatment data of multiple patients; a model building module, which is connected to the storage module and is used to perform big data analysis on the historical diagnosis and treatment data of the multiple patients to obtain a diagnosis model and / or a treatment plan generation model.

[0014] The third aspect of the present invention provides an ear-induced vertigo diagnosis and treatment system; the ear-induced vertigo diagnosis and treatment system includes: the big data analysis platform described in the second aspect of the present invention, which is used to obtain a diagnostic model and / or a treatment plan generation model; the ear-induced vertigo diagnosis and treatment device described in the first aspect of the present invention, which is communicatively connected to the big data analysis platform and is used to diagnose and treat the patient's ear-induced vertigo.

[0015] As described above, a technical solution of the ototoxic vertigo diagnosis and treatment device, system and big data analysis platform of the present invention has the following beneficial effects:

[0016] The ear-related vertigo diagnosis and treatment device can process the patient's vertigo information using two big data models, namely the diagnosis model and the treatment plan generation model, so as to obtain the patient's probability of illness and treatment plan. This process basically does not require human participation, so the diagnosis of ear-related vertigo is not limited by the level of medical personnel and the diagnosis and treatment process is simple. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 Shown is a schematic diagram of the structure of the ear-induced vertigo diagnosis and treatment device in a specific embodiment of the present invention.

[0018] Figure 2 Shown is a schematic diagram of the structure of the ear-induced vertigo diagnosis and treatment device in another specific embodiment of the present invention.

[0019] Figure 3 Shown is a schematic diagram of the structure of a vertigo information acquisition module in a specific embodiment of the ear-related vertigo diagnosis and treatment device described in the present invention.

[0020] Figure 4 Shown is a schematic diagram of the structure of the model building module of the ear-related vertigo diagnosis and treatment device in a specific embodiment of the present invention.

[0021] Figure 5A Shown is a flow chart of training a big data model in a specific embodiment of the ear-related vertigo diagnosis and treatment device described in the present invention.

[0022] Figure 5B Shown is a flow chart of a weight update subroutine in a specific embodiment of the ear-related vertigo diagnosis and treatment device described in the present invention.

[0023] Figure 6 Shown is a schematic diagram of the structure of the ear-induced vertigo diagnosis and treatment system in a specific embodiment of the present invention.

[0024] Component number description

[0025] 1 Otogenic vertigo diagnosis and treatment device

[0026] 11 Dizziness Information Acquisition Module

[0027] 111 Speech Recognition Unit

[0028] 112 Natural Language Processing Unit

[0029] 113 Questionnaire Matching Unit

[0030] 12 Disease Probability Acquisition Module

[0031] 13 Treatment plan generation module

[0032] 14 Model building module

[0033] 141 Data acquisition unit

[0034] 142 Feature Extraction Unit

[0035] 143 Model training unit

[0036] 15 Dizziness Questionnaire Generation Module

[0037] Steps S41 to 43

[0038] Steps S421 to S424 DETAILED DESCRIPTION

[0039] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0040] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and the illustrations only show the components related to the present invention rather than the number, shape and size of the components in actual implementation. The type, quantity and proportion of each component in actual implementation can be changed at will, and the layout of the components may be more complicated. In addition, in this article, relational terms such as "first", "second", etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0041] Clinically, the diagnosis of otogenic vertigo often requires imaging and neuroelectrophysiological examinations to make a clear diagnosis. Treatment often requires the cooperation of multiple specialists such as otolaryngology, neurology, geriatrics, rehabilitation, and nursing. The treatment process is complicated and requires a high level of medical staff. Especially for grassroots hospitals, due to the limitations of medical level and medical equipment, it is difficult for grassroots hospitals to reasonably diagnose and treat patients' otogenic vertigo. In response to this problem, the present invention provides an otogenic vertigo diagnosis and treatment device, which can process the patient's vertigo information using two big data models, a diagnosis model and a treatment plan generation model, so as to obtain the patient's probability of illness and treatment plan. This process basically does not require human participation. Therefore, the diagnosis of otogenic vertigo is not limited by the level of medical staff and the diagnosis and treatment process is simple.

[0042] See also Figure 1 In one embodiment of the present invention, the ototoxic vertigo diagnosis and treatment device 1 comprises:

[0043] The vertigo information acquisition module 11 is used to acquire the patient's vertigo information. The patient's vertigo information includes the patient's symptom information, the patient's basic information, the patient's laboratory test data, etc. Specifically, the symptom information is, for example, visual rotation, nausea, tinnitus, etc., the basic information is, for example, the patient's gender, age, place of residence, etc., and the laboratory test data is, for example, blood test data.

[0044] The disease probability acquisition module 12 is connected to the dizziness information acquisition module 11, and is used to process the patient's dizziness information using a diagnostic model to obtain the patient's disease probability. The diagnostic model is a big data model constructed based on the historical diagnosis and treatment data of multiple patients, and the big data model is, for example, a classification model, a prediction model, a clustering model, etc. The patient refers to a patient who has the same or similar dizziness information as the patient, and the multiple patients may or may not include the patient. The patient's disease probability may be the disease probability of otopathic vertigo, or it may be the disease probability of two or more diseases including otopathic vertigo. In this embodiment, the disease probability acquisition module 12 uses the patient's dizziness information as the input of the diagnostic model, and the output of the diagnostic model is the patient's disease probability.

[0045] In particular, for any disease, the patient's probability of suffering from the disease can be any value between 0 and 1, which is used to quantitatively indicate the possibility that the patient suffers from the disease; the patient's probability of suffering from the disease can also be any one of the two values ​​0 or 1, which is used to qualitatively indicate whether the patient is likely to suffer from the disease.

[0046] The treatment plan generation module 13 is connected to the dizziness information acquisition module 11 and the disease probability acquisition module 12, and is used to process the patient's dizziness information and the patient's disease probability using a treatment plan generation model to generate a treatment plan for the patient. The treatment plan generation model is a big data model constructed based on the historical diagnosis and treatment data of multiple patients, and the big data model is, for example, a correlation model, a regression model, etc. In this embodiment, the treatment plan generation module 13 uses the patient's dizziness information and the patient's disease probability as inputs to the treatment plan generation model, and the output of the treatment plan generation model is the patient's treatment plan. The treatment plan includes a medication plan, a medical treatment plan and / or lifestyle recommendations.

[0047] Among them, the diagnosis model and the treatment plan generation model are big data models constructed based on the historical diagnosis and treatment data of multiple patients.

[0048] Specifically, when constructing the diagnostic model, the patient's historical diagnosis and treatment data includes the patient's dizziness information and the patient's illness. The patient's dizziness information includes symptom information, basic information and / or laboratory test data. The patient's illness includes the disease the patient actually suffers from, which can be the disease that the patient is finally diagnosed with after visiting the hospital, which can be obtained from the patient's medical records.

[0049] When constructing the diagnosis plan generation model, the patient's historical diagnosis and treatment data includes the patient's dizziness information, the patient's illness and the patient's treatment plan. The patient's treatment plan refers to the treatment plan adopted by the patient for the disease he suffers from. The treatment plan can be the treatment plan given by the doctor after the patient's consultation, which can be obtained from the patient's medical records.

[0050] According to the above description, the otogenic vertigo diagnosis and treatment device described in this embodiment can process the patient's vertigo information using two big data models, the diagnosis model and the treatment plan generation model, so as to obtain the patient's disease probability and treatment plan. This process basically does not require human participation, so the diagnosis of otogenic vertigo is not limited by the level of medical personnel and the diagnosis and treatment process is simple.

[0051] Unlike general diseases, the symptoms of otogenic vertigo are often highly subjective. If only objective examinations are performed on the patient, the vertigo information obtained is often not comprehensive enough, which in turn affects the diagnosis of otogenic vertigo. For this issue, please refer to Figure 2 In one embodiment of the present invention, the ear-related vertigo diagnosis and treatment device 1 further includes a vertigo questionnaire generation module 15. The vertigo questionnaire generation module 15 is used to generate an ear-related vertigo questionnaire; the ear-related vertigo questionnaire is used to assist the patient in inputting questionnaire information; the vertigo information acquisition module acquires the patient's vertigo information according to the questionnaire information.

[0052] Specifically, the patient can obtain the otopathic vertigo questionnaire generated by the vertigo questionnaire generation module 15 through a web page, a mini-program, etc., and input questionnaire information according to the otopathic vertigo questionnaire. The questionnaire information can be questionnaire information in text form or in voice form. The vertigo information acquisition module 11 acquires the questionnaire information input by the patient, and acquires the patient's vertigo information according to the questionnaire information.

[0053] Preferably, the otogenic vertigo questionnaire includes a vertigo symptom sub-questionnaire, an onset timing sub-questionnaire, an accompanying symptom sub-questionnaire, a disease history sub-questionnaire and / or a medication history sub-questionnaire.

[0054] Among them, the vertigo symptom sub-questionnaire is used to assist patients in inputting symptoms during vertigo attacks, such as: whether the world is spinning or objects are rotating during an attack, how long the attack lasts, and whether it recurs; the onset time sub-questionnaire is used to assist patients in inputting the onset time, such as: whether the attack occurs when lying down or getting up quickly from a sitting or lying position, whether the attack occurs after insomnia or poor rest, etc.; the accompanying symptoms sub-questionnaire is used to assist patients in inputting accompanying symptoms during vertigo attacks, such as: whether the attack is accompanied by a headache, whether there is fluctuating hearing loss during the course of the disease, etc.; the medical history sub-questionnaire is used to assist patients in inputting medical history and / or family genetic history, such as: whether there is a history of eye disease, whether immediate relatives have a history of similar diseases, etc.; the medication history sub-questionnaire is used to assist patients in inputting medication information, such as: recent use of ototoxic drugs, recent use of sedatives, etc.

[0055] According to the above description, the vertigo questionnaire generation module in this embodiment can generate an otopathic vertigo questionnaire, and the patient can input questionnaire information according to the otopathic vertigo questionnaire. The questionnaire information is subjective information provided by the patient, so the symptoms of otopathic vertigo disease with strong subjectivity can be obtained.

[0056] See also Figure 3 In one embodiment of the present invention, the dizziness information acquisition module 11 includes:

[0057] The speech recognition unit 111 is used to convert the questionnaire information in speech form into a text sequence. Specifically, the speech recognition unit 111 preprocesses the questionnaire information in speech form, and converts each frame waveform into a multi-dimensional vector using the cepstral coefficients (LPCC); then, the speech recognition unit 111 can convert the multi-dimensional vector into phoneme information using a deep learning acoustic model, and match the factor information with a dictionary and convert it into a text sequence. Among them, the preprocessing is, for example, noise reduction, silence removal, sound framing, etc. For example, the speech recognition unit 111 can convert the multi-dimensional vector into phoneme information using a deep learning acoustic model based on a long short-term memory artificial neural network (LSTM). LSTM is composed of a forget gate, an input gate, and an output gate to control the implicit output state of the cell state. When the multi-dimensional vector is processed using LSTM, the following steps are mainly included:

[0058] Step 1: The forget gate forgets some cell states. Specifically, at time t, the forget gate receives a long-term memory C t-1 (The output from the previous unit module, i.e., the cell state at the previous moment) and decide whether to keep or forget C t-1 Which part of the process can be achieved by C t-1 Multiply by a forgetting factor f t To achieve this, that is: ft ×C t-1 =σ(W f ,[h t-1 ,x t ])×C t-1 . Among them, σ is the Sigmoid function; W f is the weight matrix of the forget gate, h t-1 represents the output of the network at the previous moment, x t represents the output of the network at the current moment, [h t-1 ,x t ] means to set h t-1 and x t Concatenate into a vector.

[0059] Step 2: The input gate inputs the current information and updates the cell state. The updated cell state is Among them, i t =σ(W t ,[h t-1 ,x t ]), Among them, tanh W C and W t It is also the weight matrix of the forget gate.

[0060] Step 3: Determine the hidden variable output. The output of the hidden variable h t =o t ×tanh(C t )=σ(W o ,[h t-1 ,x t ])×tanh(C t ), W o is another weight matrix. The above weight matrix W f , W C , W t and / or W o Both can be set according to experience.

[0061] Preferably, the dictionary is a dictionary generated according to the otogenic vertigo questionnaire, and the dictionary contains words and phrases related to the questionnaire content and answers of the otogenic vertigo questionnaire. For example, if the questionnaire information contains "whether the world is spinning or the visual objects are spinning during the attack", the dictionary may contain "the world is spinning", "visual objects are spinning" and related words.

[0062] The natural language processing unit 112 is connected to the speech recognition unit 111 and is used to perform natural language processing on the text sequence to obtain key text in the text sequence. The processing performed by the natural language processing unit 112 on the text sequence includes word segmentation, part-of-speech tagging, relationship analysis, and key text recognition, and the specific solution can be implemented by existing natural language processing technology.

[0063] The questionnaire matching unit 113 is connected to the natural language processing unit 112 and the vertigo questionnaire generation module 15, and is used to match the key text with the ear-induced vertigo questionnaire to obtain the patient's vertigo information. The matching includes: matching the key text with the question, and verifying and modifying the matching result by using relationship analysis, and the patient's answer result to the question in the questionnaire can be obtained through the matching. For example, if the key text includes "spinning", and the ear-induced vertigo questionnaire also includes "spinning", the judgment result obtained after matching the key text with the question is defaulted to "yes"; thereafter, the relationship analysis process will determine whether the text sequence contains "no spinning" and similar words, if so, the judgment result will be changed to "no", otherwise, the judgment result will remain "yes".

[0064] This embodiment provides a method for obtaining the patient's vertigo information based on the questionnaire information in voice form, thereby allowing the patient to complete the otopathic vertigo questionnaire by oral narration or other methods.

[0065] In one embodiment of the present invention, the ear-induced vertigo diagnosis and treatment device also includes a vertigo detection module; wherein the vertigo detection module is connected to the vertigo information acquisition module, and is used to perform vertigo detection on the patient to obtain the patient's vertigo detection results; the patient's vertigo detection results include vestibular evoked myogenic potentials, acoustic impedance test results, pure tone audiometry results, video nystagmus, otolith examination results, head thrust test results and / or dynamic and static balance platform test results; the vertigo information acquisition module obtains the patient's vertigo information based on the patient's vertigo detection results.

[0066] Specifically, the vestibular evoked myogenic potential is an evoked potential generated by bone conduction vibration or air conduction sound stimulation of the otoliths, and the functional state of the otolith organ and its nerve conduction can be explored through the relative amplitude and latency of the evoked potential. The vestibular evoked myogenic potential can be recorded on the patient's sternocleidomastoid muscle, at which point the vestibular evoked myogenic potential can detect the functional state of the saccule and the anterior stop nerve. In addition, the vestibular evoked myogenic potential can also be recorded on the eye muscles, at which point the vestibular evoked myogenic potential can detect the functional state of the utricle and the superior vestibular nerve. Auxiliary diagnosis of otopathic vertigo can be achieved according to the state of the vestibular evoked myogenic potential of the patient. For example, if the vestibular evoked myogenic potential recorded on the patient's sternocleidomastoid muscle is normal, and the vestibular evoked myogenic potential recorded on the eye muscles is abnormal, and the patient's anterior semicircular canal and lateral semicircular canal are normal, it means that the patient's damage may be mainly in the superior vestibular nerve.

[0067] The acoustic immittance test results include the patient's tympanogram, through which the relevant information of the patient's middle ear and Eustachian tube function and the relevant information of the patient's acoustic reflex path can be obtained. The pure tone audiometry results can be obtained by performing pure tone audiometry on the patient, and are used to reflect the patient's hearing level. The video nystagmus can reflect central or vestibular lesions through scanning, gaze, optokinetic tracking and / or hot and cold water tests. The otolith examination results are used to reflect the patient's otolith condition, wherein the otoliths are important organs for controlling the balance of the human body, and normal otoliths are in the three semicircular canals. The dynamic and static balance table test results can be used to evaluate the patient's vestibular system, proprioceptive system and / or visual system.

[0068] The head-thrust test result is used to evaluate whether the vestibular-ocular reflexes on both sides of the subject are symmetrical, and then to determine whether there is unilateral vestibular function decline. Specifically, the patient is required to fix his eyes forward and use the patient's nose as a visual target. When the patient cannot predict the direction of the head shaking and the start time of the test, the patient's head is shaken to both sides in a continuous and sudden manner, with an angle of about 15° to 30°. After the shaking stops, the head-thrust test result can be obtained by observing the subject's nystagmus. Preferably, the vertigo detection module can be implemented by one or more intelligent diagnosis and treatment devices. The intelligent diagnosis and treatment equipment can complete the items that originally required manual inspection by doctors through artificial intelligence-assisted equipment, which is conducive to reducing the workload of doctors, thereby improving inspection efficiency and accuracy. The detection results of the intelligent diagnosis and treatment equipment can be in the form of tables, numbers, images and / or text.

[0069] The dizziness information acquisition module can acquire the dizziness information of the patient according to the dizziness detection result of the patient. At this time, the dizziness information includes the patient's acoustic impedance test result, pure tone audiometry result, video nystagmus, otolith examination result, head thrust test result and / or dynamic and static balance platform test result. In addition, the dizziness information acquisition module can also acquire the patient's dizziness information according to the patient's questionnaire information and the patient's dizziness detection result.

[0070] See also Figure 4 In one embodiment of the present invention, the diagnosis model and / or the treatment plan generation model is generated by a model construction module 14. It should be noted that the model construction module 14 can be included in the ear-induced vertigo diagnosis and treatment device 1 and used as a functional module of the ear-induced vertigo diagnosis and treatment device 1; the model construction module 14 can also be a functional module in other devices or equipment outside the ear-induced vertigo diagnosis and treatment device 1. The model construction module 14 includes:

[0071] The data acquisition unit 141 is used to acquire the historical diagnosis and treatment data of the multiple patients, wherein the historical diagnosis and treatment data of the patients include the dizziness information of the patients and the illness conditions of the patients.

[0072] The feature extraction unit 142 is connected to the data acquisition unit 141, and is used to extract feature data from the historical diagnosis and treatment data of the multiple patients. In specific applications, the feature vectors obtained based on the historical diagnosis and treatment data of the multiple patients usually belong to a high-dimensional space. The feature data can be extracted by mapping the feature vectors of the high-dimensional space to a low-dimensional space. The mapping process can be regarded as a mapping or transformation from the measurement space to the feature space. The implementation method of mapping the feature vector from the high-dimensional space to the low-dimensional space can be implemented using existing solutions, which will not be described here. In addition, the feature extraction unit 142 can also select one or more representative features from the feature vectors of the low-dimensional space as the feature data. In specific applications, the existing Wilcoxon test, Mann-Whitney U test, Shapiro-Wilk W test, etc. can be used to implement the selection process.

[0073] The model training unit 143 is connected to the feature extraction unit 142, and is used to train the big data model according to the feature data to obtain the diagnosis model. The big data model is, for example, an Adaboost model, a decision tree model, a neural network model or a classifier model. Specifically, the model training unit 143 uses the feature data as training data to train the big data model, and the stable big data model obtained after the training is the diagnosis model. After the training is completed, the patient's vertigo information is used as the input of the diagnosis model, and the output of the diagnosis model is the patient's probability of illness. The model training unit 143 can use existing technology to implement the training of the big data model, which will not be repeated here.

[0074] Next, we will take the Adaboost model as an example to introduce the training process of the big data model. The basic principle of the Adaboost model is to reasonably combine multiple weak classifiers to form a strong classifier, which is the diagnostic model. Figure 5A The training process includes: S41, initializing the weight distribution of training data. The training data can be expressed as (x 1 ,y 1 ),(x 2 ,y 2 ),...,(x N ,y N ), where x i represents the i-th training sample, y i ∈{-1,1} is used to represent the category label of the i-th training sample, i=1,2,...,N; the initial weight distribution of the data can be expressed as D 1 =(w 1,1 ,w 1,2 ,...,w 1,N ), where w 1,i Indicates the initial weight value of the i-th training sample. In the initial state, the weight of each training sample can be set to the same value, for example: 1 / N. At this time, the initial weight distribution D of the data 1 =(1 / N,1 / N,...,1 / N).

[0075] S42, repeatedly executing a weight update sub-process until the termination condition is met. Figure 5B , the weight updating sub-process includes:

[0076] S421, select a weak classifier as the tth basic classifier H according to the error rate of the weak classifier t For example, a weak classifier with the lowest error rate can be selected as the basic classifier H t. Wherein, t is a positive integer, and its initial value is 1.

[0077] S422, calculate H t In the weight distribution D t The specific calculation formula of the error is: Among them, H t (x i ) is to use the basic classifier H t x i The classification result of t,i Indicates that when executing the weight update sub-process for the tth time, x i The weight value of I(H t (x i )≠y i ) is the indicator function, when H t (x i )≠y i When its value is 1; when H t (x i )=y i When its value is 0.

[0078] S423, calculate H t The weight in the final classifier is calculated as:

[0079] S424, update the weight distribution D of the training samples t+1 =(w t+1,1 ,w t+1 , 2,...,w t+1,N );in, and Represents a normalization constant. Let the value of t increase by 1. At this time, the termination condition is, for example, that the value of t reaches a threshold value T, that is, when t≤T, jump to step S421; when t>T, execute step S43.

[0080] S43, combining each weak classifier according to its weight to obtain the diagnosis model. Specifically, the classifier H(x) obtained after the combination is the diagnosis model, and the combination method is Wherein, sign represents a sign function; T is the threshold in the termination condition, which is also the number of the weak classifiers.

[0081] In this embodiment, the model building module 14 can also be used to build a treatment plan generation model. Specifically, the patient's historical diagnosis and treatment data acquired by the data acquisition unit 141 includes the patient's dizziness information, the patient's illness condition and the patient's treatment plan. At this time, the feature extraction unit 142 acquires the feature data based on the historical diagnosis and treatment data of the multiple patients, and the model training unit 143 trains the big data model based on the feature data to acquire the treatment plan generation model. In essence, the treatment plan generation module processes the patient's dizziness information and illness probability to generate a corresponding treatment plan based on the big data of a large number of patients' dizziness information, illness condition and adopted treatment plans. The construction process of the treatment plan generation model is similar to that of the diagnosis model, which will not be repeated here.

[0082] In one embodiment of the present invention, the model training unit includes a model training subunit and a model selection subunit. The model training subunit is connected to the feature extraction unit and is used to train at least two big data models according to the feature data to obtain at least two candidate models. The model selection subunit is connected to the model training subunit and is used to select the diagnostic model from the candidate models according to the performance of each of the candidate models.

[0083] In this embodiment, the two candidate models obtained by the model training subunit can be classifiers. At this time, the model selection subunit selects one classifier from the candidate models as the diagnostic model according to the classification performance of each classifier. Among them, the classification performance can be represented by the diagnostic rate and misdiagnosis rate of the classifier; the diagnostic rate refers to the probability that a patient with ear-related vertigo is diagnosed as ear-related vertigo, and the misdiagnosis rate refers to the probability that a patient with non-ear-related vertigo is misdiagnosed as ear-related vertigo. Specifically, for any classifier in the candidate model, the model selection subunit obtains the diagnostic rate and misdiagnosis rate of the classifier under multiple classification thresholds, and obtains multiple points corresponding to different classification thresholds with the misdiagnosis rate as the horizontal coordinate and the diagnostic rate as the vertical coordinate. By connecting multiple points, a classification performance curve can be obtained, and then the area of ​​the area surrounded by the classification performance curve and the horizontal coordinate can be obtained. The model selection subunit obtains the classification performance curves corresponding to each classifier respectively, and selects the diagnostic model according to the area of ​​the area surrounded by the classification performance curve and the horizontal coordinate. For example, the classifier corresponding to the classification performance curve with the largest area may be selected as the diagnosis model. The classification threshold refers to the threshold used by the classifier when completing the classification.

[0084] In one embodiment of the present invention, the model construction module is included in the ear-induced vertigo diagnosis and treatment device, and is connected to the disease probability acquisition module and the treatment plan generation module. Preferably, the ear-induced vertigo diagnosis and treatment device also includes a big data storage module, which stores a vertigo-specific disease database and other disease databases. Among them, the vertigo-specific disease database includes a large number of historical diagnosis and treatment data of patients; the construction module can realize the construction and / or training of the diagnosis model and the treatment plan generation model based on the data in the vertigo-specific disease database.

[0085] In one embodiment of the present invention, the otogenic vertigo diagnosis and treatment device further includes a consultation recommendation generation module. The consultation recommendation generation module is connected to the treatment plan generation module, and is used to generate consultation recommendations based on the patient's treatment plan. Specifically, the consultation recommendation generation module can obtain the hospital's department selection recommendations for different diseases and the doctor's specific consultation situation, and give consultation recommendations based on the patient's illness information and illness probability; wherein the consultation recommendations include department and doctor recommendations.

[0086] The present invention also provides a big data analysis platform. Figure 6 In one embodiment of the present invention, the big data analysis platform includes a storage module and a model building module. The storage module stores historical diagnosis and treatment data of multiple patients; the model building module is connected to the storage module and is used to perform big data analysis on the historical diagnosis and treatment data of the multiple patients to obtain a diagnosis model and / or a treatment plan generation model.

[0087] Based on the above description of the ear-induced vertigo diagnosis and treatment device and the big data analysis platform, the present invention also provides an ear-induced vertigo diagnosis and treatment system. Wherein, the ear-induced vertigo diagnosis and treatment system includes a big data analysis platform and an ear-induced vertigo diagnosis and treatment device communicatively connected to the big data analysis platform. The big data analysis platform is used to obtain a diagnostic model and / or a treatment plan generation model; the ear-induced vertigo diagnosis and treatment device is used to diagnose and treat the patient's ear-induced vertigo according to the diagnostic model and / or the treatment plan generation model. It should be noted that the diagnostic model and / or the treatment plan generation model used by the ear-induced vertigo diagnosis and treatment device can be constructed on the server by the big data analysis platform, or can be constructed locally by the ear-induced vertigo diagnosis and treatment device.

[0088] The ear-related vertigo diagnosis and treatment device of the present invention can process the patient's vertigo information using two big data models, namely the diagnosis model and the treatment plan generation model, so as to obtain the patient's probability of illness and treatment plan. This process basically does not require human participation, so the diagnosis of ear-related vertigo is not limited by the level of medical personnel and the diagnosis and treatment process is simple.

[0089] In summary, the present invention effectively overcomes various shortcomings of the prior art and has high industrial utilization value.

[0090] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A device for diagnosing and treating otopathic vertigo, It is characterized in that The otopathic vertigo diagnosis and treatment device comprises: A vertigo questionnaire generation module is used to generate an otopathic vertigo questionnaire; the otopathic vertigo questionnaire is used to assist patients in inputting questionnaire information, and the otopathic vertigo questionnaire includes a vertigo symptom sub-questionnaire, an onset time sub-questionnaire, an accompanying symptom sub-questionnaire, a disease history sub-questionnaire and / or a medication history sub-questionnaire, the vertigo symptom sub-questionnaire is used to assist patients in inputting symptoms during vertigo attacks, the onset time sub-questionnaire is used to assist patients in inputting onset times, the accompanying symptom sub-questionnaire is used to assist patients in inputting accompanying symptoms during vertigo attacks, the disease history sub-questionnaire is used to assist patients in inputting medical history and / or family genetic history, and the medication history sub-questionnaire is used to assist patients in inputting medication information; A dizziness information acquisition module, used to acquire the dizziness information of the patient; a disease probability acquisition module, connected to the vertigo information acquisition module, for processing the patient's vertigo information using a diagnostic model to obtain the patient's disease probability; the patient's disease probability is any value between 0 and 1 to quantitatively indicate the possibility that the patient suffers from otopathic vertigo, or is one of the two values ​​of 0 or 1 to qualitatively indicate whether the patient suffers from otopathic vertigo; A treatment plan generation module, connected to the dizziness information acquisition module and the disease probability acquisition module, for processing the dizziness information of the patient and the disease probability of the patient using a treatment plan generation model to generate a treatment plan for the patient; a vertigo detection module, connected to the vertigo information acquisition module, and used for performing vertigo detection on the patient to obtain the patient's vertigo detection result, wherein the patient's vertigo detection result includes vestibular evoked myogenic potential, acoustic impedance test result, pure tone audiometry result, video nystagmus, otolith examination result, head thrust test result and / or dynamic and static balance platform test result, and the vertigo information acquisition module obtains the patient's vertigo information according to the patient's vertigo detection result; Wherein, the diagnosis model and the treatment plan generation model are big data models constructed based on historical diagnosis and treatment data of multiple patients; The vestibular evoked myogenic potential is an evoked potential generated by stimulating the otoliths through bone conduction vibration or air conduction sound, and the vertigo detection module detects the functional state of the otolith organ and its nerve conduction according to the relative amplitude and latency of the evoked potential; the acoustic immittance test result includes the patient's tympanogram, and the vertigo detection module obtains the patient's middle ear and Eustachian tube function information and the patient's acoustic reflex path information through the acoustic immittance test result; the pure tone audiometry result is used to reflect the patient's hearing level; the video nystagmus reflects central or vestibular lesions through saccades, gaze, optokinetic tracking and / or hot and cold water tests; the otolith examination result is used to reflect the patient's otolith condition; the dynamic and static balance platform test result is used to evaluate the patient's vestibular system, proprioceptive system and / or visual system; the head-thrust test result is used to evaluate whether the vestibular-ocular reflex on both sides of the subject is symmetrical, and the vertigo detection module determines whether the patient has unilateral vestibular function decline according to the head-thrust test result; The questionnaire information is in the form of voice; the dizziness information acquisition module includes: A speech recognition unit, used for preprocessing the questionnaire information in speech form and converting each frame waveform into a multidimensional vector using cepstral coefficients, converting the multidimensional vector into phoneme information using a deep learning acoustic model, and matching the phoneme information with a dictionary and converting it into a text sequence; the dictionary is a dictionary generated based on the otopathic vertigo questionnaire, and contains text and words related to the questionnaire content and answers of the otopathic vertigo questionnaire; A natural language processing unit, connected to the speech recognition unit, for performing natural language processing on the text sequence to obtain key text in the text sequence; A questionnaire matching unit is connected to the natural language processing unit and the vertigo questionnaire generation module, and is used to match the key text with the otopathic vertigo questionnaire to obtain the patient's vertigo information. The matching includes matching the key text with the question and using relationship analysis to verify and modify the matching result to obtain the patient's answer to the questions in the questionnaire.

2. The otopathic vertigo diagnosis and treatment device according to claim 1, It is characterized in that The diagnosis model and / or the treatment plan generation model is generated by a model building module, and the model building module includes: A data acquisition unit, used to acquire historical diagnosis and treatment data of the multiple patients; a feature extraction unit, connected to the data acquisition unit, and configured to extract feature data from the historical diagnosis and treatment data of the plurality of patients; A model training unit is connected to the feature extraction unit and is used to train the big data model according to the feature data to obtain the diagnosis model and / or the treatment plan generation model.

3. The otopathic vertigo diagnosis and treatment device according to claim 2, It is characterized in that The model training unit comprises: A model training subunit, connected to the feature extraction unit, for training at least two big data models according to the feature data to obtain at least two candidate models; The model selection subunit is connected to the model training subunit and is used to select the diagnostic model from the candidate models according to the performance of each candidate model.

4. The otopathic vertigo diagnosis and treatment device according to claim 2, Features: The model building module is included in the ear-related vertigo diagnosis and treatment device, and is connected to the disease probability acquisition module and the treatment plan generation module.

5. The otopathic vertigo diagnosis and treatment device according to claim 1, It is characterized in that The otopathic vertigo diagnosis and treatment device further comprises: The medical advice generation module is connected to the treatment plan generation module and is used to generate medical advice according to the patient's treatment plan.

6. A big data analysis platform, It is characterized in that The big data analysis platform includes: A storage module stores historical diagnosis and treatment data of multiple patients; A model building module is connected to the storage module and is used to perform big data analysis on the historical diagnosis and treatment data of the multiple patients to obtain a diagnosis model and / or a treatment plan generation model, and the diagnosis model and / or treatment plan generation model is applied to the ear-related vertigo diagnosis and treatment device described in any one of claims 1 to 5.

7. A system for diagnosing and treating ototoxic vertigo. It is characterized in that The otopathic vertigo diagnosis and treatment system comprises: The big data analysis platform of claim 6 is used to obtain a diagnostic model and / or a treatment plan generation model; The ear-related vertigo diagnosis and treatment device described in any one of claims 1-5 is communicatively connected to the big data analysis platform, and is used to diagnose and treat the patient's ear-related vertigo.