An intelligent Chinese medicine ear diagnosis auxiliary system

Abstract

The present invention discloses an intelligent TCM ear diagnosis auxiliary system, comprising: an ear data acquisition module for collecting pathological image data, temperature percentage data, and humidity percentage data from the patient's ear and performing preprocessing; a pathological image data analysis module for matching and analyzing the pathological image data to generate disease diagnosis data; a temperature and humidity percentage data analysis module for integrating and analyzing the temperature percentage data and the humidity percentage data to generate disease severity data; an accuracy analysis module for integrating and analyzing the disease diagnosis data and disease severity data to generate accuracy data; and a disease output module for outputting a diagnosis result and error level based on the disease diagnosis data and accuracy data. The present invention improves the authenticity and accuracy of diagnosis and reduces the incidence of misdiagnosis.

Description

Technical Field

[0001] The present invention relates to the technical field of traditional Chinese medicine diagnosis and treatment, and in particular to an intelligent traditional Chinese medicine ear diagnosis auxiliary system. Background Art

[0002] The prior art with publication number CN115083588A proposes a traditional Chinese medicine ear diagnosis instrument and its diagnostic system. The prior art can obtain high-definition pictures of the ear through the ear diagnosis instrument body and automatically upload them to an artificial intelligence module in the cloud. The artificial intelligence module uses image recognition technology and follows the logic of traditional Chinese medicine ear diagnosis to obtain a preliminary diagnosis result. The diagnosis result is then confirmed online by a traditional Chinese medicine ear diagnosis expert. Finally, the medicine is prescribed online through an APP module and delivered to the home by express delivery. This allows users to diagnose ear problems at home, avoiding the tediousness and inconvenience of going to the hospital.

[0003] Combining the existing technology for analysis, we found that the following problems still exist:

[0004] 1. The existing ear diagnosis logic using image recognition technology does not provide detailed descriptions of the collected image data, resulting in low accuracy in the analysis of the input recognition model.

[0005] 2. The existing technology that uses only image recognition for detection cannot provide real-time feedback on the temperature and humidity conditions inside the ear canal. If secretions have just occurred in the ear canal, real-time diagnosis cannot be performed based on image analysis, resulting in a delayed diagnosis. Summary of the Invention

[0006] In view of this, the purpose of the present invention is to provide an intelligent Chinese medicine ear diagnosis auxiliary system.

[0007] In order to achieve the above technical objectives, the technical solution adopted by the present invention is:

[0008] The present invention provides an intelligent traditional Chinese medicine ear diagnosis auxiliary system, comprising:

[0009] The in-ear data acquisition module is used to collect pathological image data, temperature percentage data, and humidity percentage data of the patient's ear and perform preprocessing;

[0010] Pathological image data analysis module, used to perform matching analysis on pathological image data and generate disease diagnosis data;

[0011] The temperature and humidity ratio data analysis module is used to integrate and analyze the temperature ratio data and humidity ratio data to generate disease severity data;

[0012] The accuracy analysis module is used to integrate and analyze disease diagnosis data and disease severity data to generate accuracy data;

[0013] The condition output module is used to output the diagnosis results and error degree based on the condition diagnosis data and accuracy data.

[0014] Furthermore, the pathological image data includes the shape of the auricle, the lines of the auricle, the size of the ear canal, the color of the ear canal and the presence of foreign matter in the ear canal; the temperature ratio data is the ratio of the temperature inside the ear to the indoor temperature; and the humidity ratio data is the ratio of the humidity inside the ear to the indoor humidity.

[0015] Furthermore, the pathological image data analysis module uses computer vision technology for processing, and the computer vision technology requires the construction of a machine learning model. The construction process of the machine learning model includes:

[0016] Collecting m sets of pathological image data, including normal and abnormal auricle shapes;

[0017] Marking and annotating the m groups of pathological image data according to the disease conditions corresponding to the m groups of pathological image data, and classifying them into normal or abnormal groups;

[0018] Take m groups of pathological image data as the input of the machine learning model, the disease diagnosis data as the output of the machine learning model, and the prediction accuracy of the machine learning model as the training target;

[0019] The machine learning model is trained. When the prediction accuracy of the machine learning model reaches the prediction accuracy threshold Z, the training is stopped and a trained machine learning model Zn is generated.

[0020] Furthermore, the pathological image data analysis module compares the pathological image data through the machine learning model Zn, and matches the auricle shape, auricle lines, ear hole size, ear hole color with the foreign body situation in the ear hole to generate disease diagnosis data.

[0021] Furthermore, the temperature and humidity ratio data analysis module includes:

[0022] Set the temperature ratio threshold T1 and humidity ratio threshold H1;

[0023] Compare the temperature ratio data Tr with the temperature ratio threshold T1. If the temperature ratio data Tr is less than the temperature ratio threshold T1, generate a first normal label; if the temperature ratio data Tr is not less than the temperature ratio threshold T1, generate a first abnormal label;

[0024] Compare the humidity ratio data Hr with the humidity ratio threshold H1. If the humidity ratio data Hr is less than the temperature ratio threshold H1, generate a second normal label; if the humidity ratio data Hr is not less than the temperature ratio threshold H1, generate a second abnormal label;

[0025] The temperature-humidity ratio coefficient Jer is calculated based on the temperature ratio data Tr and the humidity ratio data Hr, and compared with the preset threshold value J1 and the preset threshold value J2, where the preset threshold value J1 is less than the preset threshold value J2; if the temperature-humidity ratio coefficient Jer is less than the preset threshold value J1, a low abnormality label is generated; if the temperature-humidity ratio coefficient Jer is not less than the preset threshold value J1 and is less than the preset threshold value J2, a moderate abnormality label is generated; if the temperature-humidity ratio coefficient Jer is not less than the preset threshold value J2, a high abnormality label is generated;

[0026] The generated first normal label, first abnormal label, second normal label, second abnormal label, low abnormal label, moderate abnormal label and high abnormal label are integrated to generate disease severity data.

[0027] Furthermore, the calculation formula of the temperature-humidity ratio coefficient Jer is:

[0028] Jer= (Jer>0);

[0029] Among them, K is the error correction constant, e is the natural constant, which is used to enhance the sensitivity of temperature ratio data Tr in pathological analysis; a is the humidity adjustment factor.

[0030] Furthermore, the disease severity data includes a high-risk disease indicator, a medium-high-risk disease indicator, a medium-risk disease indicator, and a low-risk disease indicator, and the generation process includes:

[0031] Classify the first abnormal label, the second abnormal label, and the highly abnormal label as high-risk labels; classify the moderately abnormal label as a medium-risk label; and classify the first normal label, the second normal label, and the low-abnormal label as low-risk labels.

[0032] The corresponding disease severity identification is generated according to the combination of high-risk, medium-risk and low-risk labels. Specifically: if there are two or more high-risk labels at the same time, a high-risk disease identification is generated; if there is one high-risk label, one medium-risk label and one low-risk label, or one high-risk label and two low-risk labels, a medium-high-risk disease identification is generated; if there is one medium-risk label and two low-risk labels, a medium-risk disease identification is generated; if there are three low-risk labels, a low-risk disease identification is generated.

[0033] Furthermore, the accuracy data includes high-precision objects, suspicious objects, and abnormal objects, and the generation process of the accuracy data includes:

[0034] Constructing a disease data matching model to generate simulated disease data based on disease diagnosis data, wherein the simulated disease data includes a high-risk simulation identifier, a medium-high-risk simulation identifier, a medium-risk simulation identifier, and a low-risk simulation identifier;

[0035] Classifying the simulated condition data and the condition severity data to generate level identifications, specifically: classifying the high-risk simulation identification and the high-risk condition identification as a first-level identification, classifying the medium-high risk simulation identification and the medium-high risk condition identification as a second-level identification, classifying the medium-risk simulation identification and the medium-risk condition identification as a third-level identification, and classifying the low-risk simulation identification and the low-risk condition identification as a fourth-level identification;

[0036] Generate label data for high-precision objects, suspicious objects or abnormal objects based on the difference in level identification. Specifically: if the level identification between the simulated medical condition data and the disease severity data differs by more than two levels, generate label data for abnormal objects; if the level identification between the simulated medical condition data and the disease severity data differs by only one level, generate label data for suspicious objects; if the simulated medical condition data and the disease severity data have the same level identification, generate label data for high-precision objects.

[0037] Furthermore, the process of establishing the disease data matching model includes:

[0038] Set up a disease data matching model, classify the disease severity by simulating disease data, and mark each group of simulated disease data as the corresponding disease severity, i.e. high risk, medium-high risk, medium risk or low risk;

[0039] Merge the actual disease severity data and simulated disease severity data into one dataset;

[0040] Use the merged dataset to train the disease data matching model based on disease severity, and use simulated disease data for verification, and calculate the accuracy of the disease data matching model on the simulated disease data;

[0041] Evaluate the matching performance of the disease data matching model for different disease severities;

[0042] Based on the performance evaluation results, analyze whether the disease data matching model has the matching ability to correctly distinguish different disease severities.

[0043] Furthermore, the condition output module displays the diagnosis result and error degree through an output panel, wherein the error degree is distinguished by color.

[0044] By adopting the above technical solution, the present invention has the following beneficial effects compared with the prior art:

[0045] 1. Through the integrated analysis of temperature and humidity ratio data and pathological image data, the machine learning model and the disease data matching model are used to classify the conditions detected by the images. The temperature and humidity ratio data is used to further improve the authenticity and accuracy of the diagnosis and reduce misdiagnosis.

[0046] 2. Through the two-way analysis of temperature and humidity data and pathological image data, while making a diagnosis of the disease, the severity of the disease is accurately detected, thereby improving the accuracy of medication dosage and improving the early warning of the risk of mild disease. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0048] Figure 1 This is a module architecture diagram of an intelligent TCM ear diagnosis auxiliary system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The present invention will be described in further detail below with reference to the accompanying drawings and examples. It is particularly noted that the following examples are intended only to illustrate the present invention and are not intended to limit the scope of the present invention. Similarly, the following examples are only some embodiments of the present invention and are not intended to be exhaustive. All other embodiments obtained by those of ordinary skill in the art without creative effort are intended to fall within the scope of protection of the present invention.

[0050] See Figure 1 The intelligent TCM ear diagnosis auxiliary system of the present invention includes: an ear data acquisition module, a pathological image data analysis module, a temperature and humidity ratio data analysis module, an accuracy analysis module and a condition output module;

[0051] The in-ear data acquisition module is used to collect pathological image data, temperature proportion data and humidity proportion data in the patient's ear and perform preprocessing; send the pathological image data to the pathological image data analysis module, and send the temperature proportion data and humidity proportion data to the temperature and humidity proportion data analysis module;

[0052] In this embodiment, the pathological image data includes the shape of the auricle, the lines of the auricle, the size of the ear canal, the color of the ear canal and the presence of foreign matter in the ear canal. The temperature ratio data is the ratio of the temperature inside the ear to the indoor temperature. The humidity ratio data is the ratio of the humidity inside the ear to the indoor humidity.

[0053] It should be noted that the collection of pathological image data inside the ear can be performed using an ear image detector, which is specifically acquired by a camera and other image acquisition equipment, while the temperature percentage data and humidity percentage data can be collected by the temperature sensor and humidity sensor of the inner ear monitoring instrument. A preliminary diagnosis of the disease is made by observing the shape of the auricle, the lines of the auricle, the size of the ear hole, the color of the ear hole, and the presence of foreign matter in the ear hole. The collection of the temperature percentage data and the humidity percentage data is to further confirm whether there are any errors in the diagnostic results obtained from the pathological image data.

[0054] The steps of the pretreatment are:

[0055] Resize the images to the same square size for input to the model. Generally speaking, images are usually scaled to the same size, such as 224x224 pixels or 256x256 pixels;

[0056] Use image enhancement technology to perform brightness adjustment, contrast enhancement, histogram equalization, sharpening, and denoising;

[0057] Convert the image from the original RGB color space to a grayscale image (single channel) or other color spaces (such as HSV, Lab, etc.) to reduce the dimension of the data and reduce the computational burden;

[0058] According to the task requirements, the image is cropped to retain the specific area that needs to be detected and remove irrelevant parts. For example, in ear pathology image processing, only the auricle and ear canal may need to be retained.

[0059] By performing random transformations on images, such as rotation, flipping, scaling, and translation, we can increase the diversity of data and help the generalization ability of the model;

[0060] Normalize the pixel values ​​of the image and scale them to a specific range, usually [0, 1] or [-1, 1], to match the input requirements of the model;

[0061] Convert image data into a format acceptable to the model, usually tensor format, including rearranging channels, converting data types, etc.

[0062] The pathological image data analysis module is used to perform matching analysis on the pathological image data and generate disease diagnosis data;

[0063] In this embodiment, the pathological image data analysis module uses computer vision technology for processing. The computer vision technology requires the construction of a machine learning model. The construction process of the machine learning model includes:

[0064] Collecting m sets of pathological image data, including normal and abnormal auricle shapes;

[0065] Marking and annotating the m groups of pathological image data according to the disease conditions corresponding to the m groups of pathological image data to indicate the shape characteristics of the auricle of each sample and classify them into normal or abnormal groups;

[0066] Take m groups of pathological image data (their influencing attribute values) as the input of the machine learning model, the disease diagnosis data as the output of the machine learning model, the disease diagnosis data as the prediction target, and the prediction accuracy of the machine learning model as the training target;

[0067] The machine learning model is trained, and a prediction accuracy threshold Z is pre-set. When the prediction accuracy of the machine learning model reaches the prediction accuracy threshold Z, the training is stopped. At this time, the group that has completed training is recorded as n; and the nth group of machine learning models that have completed training is marked as Zn, that is, a trained machine learning model Zn is generated, and the trained machine learning model Zn is sent to the pathology image data analysis module.

[0068] In this embodiment, the logic of the pathological image data analysis module for analyzing the pathological image data is:

[0069] The pathological image data analysis module compares the pathological image data through the machine learning model Zn, matches the influencing attribute values ​​such as the auricle shape, auricle texture, ear hole size, ear hole color and foreign body conditions in the ear hole, and generates disease diagnosis data.

[0070] It should be noted that the disease diagnosis data is specifically:

[0071] The shape of the auricle can be observed to detect kidney deficiency, spleen deficiency and other internal organ problems. Kidney deficiency may cause changes in the shape of the auricle, such as loose auricle; spleen deficiency may cause the auricle to enlarge or deform;

[0072] Changes in the lines on the auricle can reflect problems with the body's blood circulation or meridians;

[0073] Ear hole size may be influenced by genetic factors, but it may also be related to external ear disease or pain;

[0074] The color of your ear piercing may indicate circulation problems, such as anemia or poor circulation. Specific colors may be associated with ear infections or inflammation.

[0075] Foreign objects in the ear canal: Foreign objects in the ear canal may cause ear pain, infection or other ear problems.

[0076] The type and location of the foreign body may help determine the possible problem.

[0077] The pathological image data analysis module sends the disease diagnosis data to the disease output module and the accuracy analysis module;

[0078] The temperature and humidity ratio data analysis module is used to integrate and analyze the temperature ratio data and the humidity ratio data to generate disease severity data;

[0079] In this embodiment, the temperature and humidity ratio data analysis module includes:

[0080] The temperature ratio data collected from the patient's ear is set as Tr, and the humidity ratio data collected from the patient's ear is set as Hr; the temperature ratio threshold T1 and the humidity ratio threshold H1 are set;

[0081] The temperature ratio data Tr is compared and analyzed with the temperature ratio threshold value T1 to generate a first label. The first label includes a first normal label and a first abnormal label. The steps for generating the first label are:

[0082] If the temperature ratio data Tr is less than the temperature ratio threshold T1, a first normal label is generated; if the temperature ratio data Tr is not less than the temperature ratio threshold T1, a first abnormal label is generated;

[0083] The humidity ratio data Hr is compared and analyzed with the humidity ratio threshold H1 to generate a second label. The second label includes a second normal label and a second abnormal label. The steps of generating the second label are:

[0084] If the humidity ratio data Hr is less than the temperature ratio threshold H1, a second normal label is generated; if the humidity ratio data Hr is not less than the temperature ratio threshold H1, a second abnormal label is generated;

[0085] The temperature-humidity ratio coefficient Jer is calculated according to the temperature proportion data Tr and the humidity proportion data Hr. The calculation formula of the temperature-humidity ratio coefficient Jer is:

[0086] Jer= (Jer>0);

[0087] Among them, K is the error correction constant, e is the natural constant, which is used to enhance the sensitivity of temperature ratio data Tr in pathological analysis; a is the humidity adjustment factor.

[0088] The temperature-humidity ratio coefficient Jer is compared with the preset threshold value J1 and the preset threshold value J2, wherein the preset threshold value J1 is less than the preset threshold value J2; the temperature-humidity ratio coefficient Jer is substituted into the preset threshold value J1 and the preset threshold value J2 to generate an abnormality degree label, wherein the abnormality degree label includes a low abnormality label, a moderate abnormality label, and a high abnormality label. The steps for generating the abnormality degree label are as follows:

[0089] If the temperature-humidity ratio coefficient Jer is less than the preset threshold J1, a low abnormality label is generated; if the temperature-humidity ratio coefficient Jer is not less than the preset threshold J1 and less than the preset threshold J2, a moderate abnormality label is generated; if the temperature-humidity ratio coefficient Jer is not less than the preset threshold J2, a high abnormality label is generated;

[0090] The generated first normal label, first abnormal label, second normal label, second abnormal label, low abnormal label, moderate abnormal label and high abnormal label are integrated to generate disease severity data.

[0091] In this embodiment, the disease severity data includes a high-risk disease indicator, a medium-high-risk disease indicator, a medium-risk disease indicator, and a low-risk disease indicator. The process of generating the disease severity includes:

[0092] Classify the first abnormal label, the second abnormal label, and the highly abnormal label as high-risk labels; classify the moderately abnormal label as a medium-risk label; and classify the first normal label, the second normal label, and the low-abnormal label as low-risk labels.

[0093] The corresponding disease severity identification is generated according to the combination of high-risk, medium-risk and low-risk labels. Specifically: if there are two or more high-risk labels at the same time, a high-risk disease identification is generated; if there is one high-risk label, one medium-risk label and one low-risk label, or one high-risk label and two low-risk labels, a medium-high-risk disease identification is generated; if there is one medium-risk label and two low-risk labels, a medium-risk disease identification is generated; if there are three low-risk labels, a low-risk disease identification is generated.

[0094] It should be noted that: in the case of common ear canal diseases, the temperature of the auricle and ear canal increases, and at this time, secretions are produced, and the humidity inside the ear canal or the epidermis of the auricle increases. When the temperature and humidity of the auricle or ear canal increase, the temperature ratio data Tr and the humidity ratio data Hr will also increase. The temperature-humidity ratio coefficient Jer, which is positively correlated with the temperature ratio data Tr and the humidity ratio data Hr, will also increase at the same time, and further judgment will be made based on the degree of increase in the temperature-humidity ratio coefficient Jer.

[0095] That is, when the temperature and humidity do not increase, the temperature-humidity ratio coefficient Jer is in a normal state;

[0096] When only the temperature of the auricle and ear canal increases, but the humidity does not increase, or the humidity increases but the temperature does not increase, the growth rate of the temperature-humidity ratio coefficient Jer rises slowly.

[0097] When the temperature and humidity of the auricle and ear canal increase at the same time, the temperature-humidity ratio coefficient Jer increases rapidly, which can be used to determine the severity of the ear infection.

[0098] The abnormality degree label is used to further analyze and determine the first label and the second label, avoiding the inaccuracy caused by detecting based on simple temperature and humidity conditions.

[0099] The temperature and humidity ratio data analysis module sends the disease severity data to the accuracy analysis module;

[0100] The accuracy analysis module is used to integrate and analyze the disease diagnosis data and disease severity data to generate accuracy data;

[0101] In this embodiment, the accuracy rate data includes high-precision objects, suspicious objects, and abnormal objects. The generation process of the accuracy rate data includes:

[0102] Constructing a disease data matching model to generate simulated disease data based on disease diagnosis data, wherein the simulated disease data includes a high-risk simulation identifier, a medium-high-risk simulation identifier, a medium-risk simulation identifier, and a low-risk simulation identifier;

[0103] Classifying the simulated condition data and the condition severity data to generate level identifications, specifically: classifying the high-risk simulation identification and the high-risk condition identification as a first-level identification, classifying the medium-high risk simulation identification and the medium-high risk condition identification as a second-level identification, classifying the medium-risk simulation identification and the medium-risk condition identification as a third-level identification, and classifying the low-risk simulation identification and the low-risk condition identification as a fourth-level identification;

[0104] Generate label data for high-precision objects, suspicious objects or abnormal objects based on the difference in level identification. Specifically: if the level identification between the simulated medical condition data and the disease severity data differs by more than two levels, generate label data for abnormal objects; if the level identification between the simulated medical condition data and the disease severity data differs by only one level, generate label data for suspicious objects; if the simulated medical condition data and the disease severity data have the same level identification, generate label data for high-precision objects.

[0105] In this embodiment, the process of establishing the condition data matching model includes:

[0106] Set up a disease data matching model, classify the disease severity by simulating disease data, and mark each group of simulated disease data as the corresponding disease severity, i.e. high risk, medium-high risk, medium risk or low risk;

[0107] Merge the actual disease severity data and simulated disease severity data into one dataset;

[0108] Use the merged dataset to train the disease data matching model based on disease severity, and use simulated disease data for verification, and calculate the accuracy of the disease data matching model on the simulated disease data;

[0109] Evaluate the matching performance of the disease data matching model for different disease severities;

[0110] Based on the performance evaluation results, analyze whether the disease data matching model has the matching ability to correctly distinguish different disease severities.

[0111] The accuracy analysis module sends the accuracy data to the condition output module;

[0112] The condition output module is used to output the diagnosis result and error degree according to the condition diagnosis data and accuracy data.

[0113] In this embodiment, the condition output module displays the diagnosis result and the degree of error through an output panel, wherein the degree of error is distinguished by color, that is, different color images are used to indicate the degree of accuracy.

[0114] The above descriptions are only some embodiments of the present invention and do not limit the scope of protection of the present invention. Any equivalent device or equivalent process transformation made by using the contents of the description and drawings of the present invention, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An intelligent Chinese medicine ear diagnosis auxiliary system, characterized in that: include: An in-ear data acquisition module is used to collect and pre-process pathological image data, temperature percentage data, and humidity percentage data from the patient's ear; the pathological image data includes auricle shape, auricle lines, ear canal size, ear canal color, and the presence of foreign matter in the ear canal; the temperature percentage data is the ratio of the in-ear temperature to the indoor temperature; and the humidity percentage data is the ratio of the in-ear humidity to the indoor humidity; A pathological image data analysis module is used to perform matching analysis on the pathological image data to generate disease diagnosis data; the pathological image data analysis module compares the pathological image data using a machine learning model Zn, and generates disease diagnosis data by matching the auricle shape, auricle lines, ear hole size, ear hole color and the presence of foreign matter in the ear hole; A temperature and humidity ratio data analysis module is used to generate a first normal label or a first abnormal label, a second normal label or a second abnormal label, and a low abnormal label, a moderate abnormal label, or a high abnormal label based on the temperature ratio data and the humidity ratio data, and integrate the generated first normal label or the first abnormal label, the second normal label or the second abnormal label, and the low abnormal label, the moderate abnormal label, or the high abnormal label to generate a disease severity data label for integrated analysis to generate disease severity data; the disease severity data includes a high-risk disease mark, a medium-high-risk disease mark, a medium-risk disease mark, and a low-risk disease mark; The accuracy analysis module is used to integrate and analyze disease diagnosis data and disease severity data to generate accuracy data; The condition output module is used to output the diagnosis results and error degree based on the condition diagnosis data and accuracy data.

2. The intelligent TCM ear diagnosis auxiliary system according to claim 1, characterized in that: The pathological image data analysis module uses computer vision technology for processing. The computer vision technology requires the construction of a machine learning model. The construction process of the machine learning model includes: Collecting m sets of pathological image data, including normal and abnormal auricle shapes; Marking and annotating the m groups of pathological image data according to the disease conditions corresponding to the m groups of pathological image data, and classifying them into normal or abnormal groups; Take m groups of pathological image data as the input of the machine learning model, the disease diagnosis data as the output of the machine learning model, and the prediction accuracy of the machine learning model as the training target; The machine learning model is trained. When the prediction accuracy of the machine learning model reaches the prediction accuracy threshold Z, the training is stopped and a trained machine learning model Zn is generated.

3. The intelligent TCM ear diagnosis auxiliary system according to claim 1, characterized in that: The temperature and humidity ratio data analysis module includes: Set the temperature ratio threshold T1 and humidity ratio threshold H1; Compare the temperature ratio data Tr with the temperature ratio threshold T1. If the temperature ratio data Tr is less than the temperature ratio threshold T1, generate a first normal label; if the temperature ratio data Tr is not less than the temperature ratio threshold T1, generate a first abnormal label; Compare the humidity ratio data Hr with the humidity ratio threshold H1. If the humidity ratio data Hr is less than the temperature ratio threshold H1, generate a second normal label; if the humidity ratio data Hr is not less than the temperature ratio threshold H1, generate a second abnormal label; The temperature-humidity ratio coefficient Jer is calculated based on the temperature proportion data Tr and the humidity proportion data Hr, and compared with the preset threshold value J1 and the preset threshold value J2, where the preset threshold value J1 is less than the preset threshold value J2; if the temperature-humidity ratio coefficient Jer is less than the preset threshold value J1, a low abnormality label is generated; if the temperature-humidity ratio coefficient Jer is not less than the preset threshold value J1 and is less than the preset threshold value J2, a moderate abnormality label is generated; if the temperature-humidity ratio coefficient Jer is not less than the preset threshold value J2, a high abnormality label is generated.

4. The intelligent TCM ear diagnosis auxiliary system according to claim 3, characterized in that: The calculation formula of the temperature-humidity ratio coefficient Jer is: You= (You>0); Among them, K is the error correction constant, e is the natural constant, which is used to enhance the sensitivity of temperature ratio data Tr in pathological analysis; a is the humidity adjustment factor.

5. The intelligent TCM ear diagnosis auxiliary system according to claim 3, characterized in that: The process of generating the disease severity data includes: Classify the first abnormal label, the second abnormal label, and the highly abnormal label as high-risk labels; classify the moderately abnormal label as a medium-risk label; and classify the first normal label, the second normal label, and the low-abnormal label as low-risk labels. The corresponding disease severity identification is generated according to the combination of high-risk, medium-risk and low-risk labels. Specifically: if there are two or more high-risk labels at the same time, a high-risk disease identification is generated; if there is one high-risk label, one medium-risk label and one low-risk label, or one high-risk label and two low-risk labels, a medium-high-risk disease identification is generated; if there is one medium-risk label and two low-risk labels, a medium-risk disease identification is generated; if there are three low-risk labels, a low-risk disease identification is generated.

6. The intelligent TCM ear diagnosis auxiliary system according to claim 5, characterized in that: The accuracy rate data includes high-precision objects, suspicious objects, and abnormal objects. The generation process of the accuracy rate data includes: Constructing a disease data matching model to generate simulated disease data based on disease diagnosis data, wherein the simulated disease data includes a high-risk simulation identifier, a medium-high-risk simulation identifier, a medium-risk simulation identifier, and a low-risk simulation identifier; Classifying the simulated condition data and the condition severity data to generate level identifications, specifically: classifying the high-risk simulation identification and the high-risk condition identification as a first-level identification, classifying the medium-high risk simulation identification and the medium-high risk condition identification as a second-level identification, classifying the medium-risk simulation identification and the medium-risk condition identification as a third-level identification, and classifying the low-risk simulation identification and the low-risk condition identification as a fourth-level identification; Generate label data for high-precision objects, suspicious objects or abnormal objects based on the difference in level identification. Specifically: if the level identification between the simulated medical condition data and the disease severity data differs by more than two levels, generate label data for abnormal objects; if the level identification between the simulated medical condition data and the disease severity data differs by only one level, generate label data for suspicious objects; if the simulated medical condition data and the disease severity data have the same level identification, generate label data for high-precision objects.

7. The intelligent TCM ear diagnosis auxiliary system according to claim 6, characterized in that: The process of establishing the disease condition data matching model includes: Set up a disease data matching model, classify the disease severity by simulating disease data, and mark each group of simulated disease data as the corresponding disease severity, i.e. high risk, medium-high risk, medium risk or low risk; Merge the actual disease severity data and simulated disease severity data into one dataset; Use the merged dataset to train the disease data matching model based on disease severity, and use simulated disease data for verification, and calculate the accuracy of the disease data matching model on the simulated disease data; Evaluate the matching performance of the disease data matching model for different disease severities; Based on the performance evaluation results, analyze whether the disease data matching model has the matching ability to correctly distinguish different disease severities.

8. The intelligent TCM ear diagnosis auxiliary system according to claim 1, characterized in that: The condition output module displays the diagnosis result and error degree through an output panel, wherein the error degree is distinguished by color.

Images