Method and apparatus for processing medical consultation information, storage medium, and computer device

Abstract

A method and apparatus for processing medical consultation information, a storage medium, and a computer device, relating to the technical field of information. The method comprises: in response to a medical consultation signal of a subject under test, acquiring a facial image of the subject under test at any medical consultation moment; determining the facial image to be a frontal image, inputting the facial image into a preset feature extraction model for feature extraction, and obtaining sub-micro feature data to be tested of various facial tissues and organs to be tested of the subject under test; acquiring diseased sub-micro feature data of various diseased facial tissues and organs of a patient suffering from Alzheimer's disease, calculating, on the basis of the sub-micro feature data to be tested and the diseased sub-micro feature data, the similarity between the various facial tissues and organs to be tested and corresponding diseased facial tissues and organs, respectively, and sending the similarity to a doctor terminal; and, on the basis of the similarity, a doctor at the doctor terminal determines a medical consultation result of the subject under test.

Description

Medical information processing method, device, storage medium and computer equipment Technical Field

[0001] The present application relates to the field of information technology, and in particular to a method, apparatus, storage medium and computer equipment for processing medical consultation information. Background Art

[0002] Alzheimer's disease (AD) is a serious neurodegenerative disease that is clinically characterized by comprehensive dementia manifestations such as memory impairment, aphasia, disability, agnosia, visual impairment, behavioral and cognitive dysfunction, and personality changes. Globally, according to a survey by the International Alzheimer's Association, there are currently at least 50 million dementia patients in the world, and the number is expected to reach 152 million by 2050, of which about 60%-70% are Alzheimer's patients. With the accelerating pace of population aging, Alzheimer's disease in the elderly has become a serious global public health issue. China, as a country with a large population base in the world, has now become one of the countries with the largest number of Alzheimer's patients. Since there is currently no effective treatment for Alzheimer's disease, early diagnosis of Alzheimer's disease has become particularly important in order to take intervention measures to prevent Alzheimer's disease as early as possible.

[0003] Currently, the diagnosis results are typically obtained through verbal communication between the doctor and the patient, with the doctor manually processing the information based on the communication. However, this method of processing information is significantly affected by human subjective factors, resulting in low accuracy. Furthermore, this process requires the doctor and patient to communicate for a long time to obtain the information, resulting in low efficiency.

[0004] Application Contents

[0005] The present application provides a method, apparatus, storage medium and computer equipment for processing medical consultation information, which are mainly capable of improving medical consultation efficiency and extraction accuracy.

[0006] According to a first aspect of the present application, a method for processing medical inquiry information is provided, comprising:

[0007] In response to a medical inquiry signal from a subject to be detected, acquiring a facial image of the subject to be detected at any medical inquiry time;

[0008] Determining that the facial image is a frontal image, inputting the facial image into a preset feature extraction model to perform feature extraction, and obtaining submicron feature data of each facial tissue and organ to be detected of the object to be detected;

[0009] Obtain diseased submicron feature data of each diseased facial tissue and organ of an Alzheimer's patient, and based on the submicron feature data to be detected and the diseased submicron feature data, respectively calculate the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ, and send the similarity to the doctor terminal. The doctor at the doctor terminal determines the consultation result of the object to be detected based on the similarity.

[0010] According to a second aspect of the present application, there is provided a medical inquiry information processing device, comprising:

[0011] an acquisition unit configured to acquire a facial image of the subject to be detected at any time of the inquiry in response to an inquiry signal of the subject to be detected;

[0012] An extraction unit is configured to determine that the facial image is a frontal image, input the facial image into a preset feature extraction model to perform feature extraction, and obtain submicron feature data to be detected of each facial tissue organ to be detected of the object to be detected;

[0013] The calculation unit is configured to obtain diseased submicron feature data of each diseased facial tissue and organ of an Alzheimer's patient, and based on the submicron feature data to be detected and the diseased submicron feature data, respectively calculate the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ, and send the similarity to the doctor terminal. The doctor at the doctor terminal determines the consultation result of the object to be detected based on the similarity.

[0014] According to a third aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the following steps are implemented:

[0015] In response to a medical inquiry signal from a subject to be detected, acquiring a facial image of the subject to be detected at any medical inquiry time;

[0016] Determining that the facial image is a frontal image, inputting the facial image into a preset feature extraction model to perform feature extraction, and obtaining submicron feature data of each facial tissue and organ to be detected of the object to be detected;

[0017] Obtain diseased submicron feature data of each diseased facial tissue and organ of an Alzheimer's patient, and based on the submicron feature data to be detected and the diseased submicron feature data, respectively calculate the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ, and send the similarity to the doctor terminal. The doctor at the doctor terminal determines the consultation result of the object to be detected based on the similarity.

[0018] According to a fourth aspect of the present application, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the following steps are implemented:

[0019] In response to a medical inquiry signal from a subject to be detected, acquiring a facial image of the subject to be detected at any medical inquiry time;

[0020] Determining that the facial image is a frontal image, inputting the facial image into a preset feature extraction model to perform feature extraction, and obtaining submicron feature data of each facial tissue and organ to be detected of the object to be detected;

[0021] Obtain diseased submicron feature data of each diseased facial tissue and organ of an Alzheimer's patient, and based on the submicron feature data to be detected and the diseased submicron feature data, respectively calculate the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ, and send the similarity to the doctor terminal. The doctor at the doctor terminal determines the consultation result of the object to be detected based on the similarity.

[0022] According to a medical inquiry information processing method, device, storage medium and computer equipment provided by the present application, compared with the current method of determining the medical inquiry results through oral communication between the doctor and the examinee, the present application obtains the facial image of the examinee at any medical inquiry time by responding to the medical inquiry signal of the examinee; then determines that the facial image is a frontal image, inputs the facial image into a preset feature extraction model for feature extraction, and obtains the submicron feature data of each facial tissue and organ to be detected of the examinee; finally obtains the diseased submicron feature data of each diseased facial tissue and organ of the Alzheimer's patient, and based on the submicron feature data to be detected and the diseased submicron feature data, respectively calculates the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ, and sends the similarity to the doctor Terminal, the doctor of the doctor terminal determines the consultation result of the object to be detected based on the similarity, thereby obtaining the facial image of the object to be detected at any consultation time, and using the feature extraction model to extract submicro feature data of the facial image, and then calculating the similarity between the submicro feature data and the facial submicro feature data of the patient with Alzheimer's disease, and sending the similarity to the doctor. The doctor finally determines the consultation result of the object to be detected based on the similarity, which can avoid long-term consultation communication between the doctor and the patient, thereby improving the consultation efficiency. At the same time, the present application only uses ordinary optical facial images at any consultation time for analysis, so that the consultation information can be processed quickly, and the consultation cost can also be reduced. In addition, the present application uses the feature extraction model to analyze the facial image, which can improve the processing accuracy of the consultation information. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0024] FIG1 shows a flow chart of a method for processing medical inquiry information provided by an embodiment of the present application;

[0025] FIG2 shows a flow chart of another method for processing medical inquiry information provided by an embodiment of the present application;

[0026] FIG3 shows a schematic diagram of a training and verification process of a preset feature extraction model provided in an embodiment of the present application;

[0027] FIG4 shows a schematic structural diagram of a medical inquiry information processing device provided in an embodiment of the present application;

[0028] FIG5 shows a schematic structural diagram of another medical inquiry information processing device provided in an embodiment of the present application;

[0029] FIG6 shows a schematic diagram of the physical structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0030] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that, unless there is a conflict, the embodiments and features in the embodiments of the present application can be combined with each other.

[0031] At present, the method of determining the results of medical consultations through verbal communication between doctors and examinees is greatly affected by human subjective factors, which will lead to low accuracy in the processing of medical consultation information. At the same time, this manual determination method requires long-term communication between doctors and patients, resulting in low efficiency in the processing of medical consultation information.

[0032] In order to solve the above problems, an embodiment of the present application provides a method for processing medical inquiry information, as shown in FIG1 , the method comprising:

[0033] 101. In response to a medical inquiry signal from a subject to be detected, obtain a facial image of the subject to be detected at any medical inquiry time.

[0034] The facial image may be a frontal facial image of the object to be detected, or a side facial image of the object to be detected.

[0035] For the embodiment of the present application, a medical consultation system is used to implement medical consultation of the object to be detected. The medical consultation system includes an image acquisition module, an image transmission module, a deep learning algorithm module, and a detection result output module. The working process of the image transmission module is: use one or more ordinary optical high-definition cameras or video cameras and other devices to collect facial images of the object to be detected. The collected facial image can be a frontal image of the object to be detected or a side image. The facial image needs to include high-definition eyes, nose, mouth, forehead, chin and other facial tissue organs of the object to be detected. Then, the facial image is passed through the image transmission module to transmit the collected facial image in real time through the network or to the storage device to the deep learning algorithm module. The function of the deep learning algorithm module is to use a preset feature extraction model to analyze the facial image and extract sub-micro feature data of each facial tissue organ in the facial image. The preset feature extraction model can be a deep learning classification network model, for example, an improved multi-layer deep convolutional residual network The core part of the model is the residual network module. Then, based on the submicro feature data of each facial tissue and organ in the facial image, the purpose of using the residual module is to ensure that the system obtains enough submicro features while retaining the original input image information. This technology has a good effect on the accurate recognition and resolution of images. Then, the similarity between the facial tissue and organs of the object to be detected and the facial tissue and organs of the Alzheimer's patient is determined, and the similarity is output through the detection result output module, and the similarity is displayed to the doctor so that the doctor can determine the consultation result based on the similarity. Therefore, the consultation result of the object to be detected can be determined based on the ordinary optical facial image of the object to be detected at any consultation time, which can reduce the consultation cost and improve the consultation information processing efficiency. At the same time, the consultation result is determined by an ordinary optical image of the face of the object to be detected, which will not add any psychological and physiological burden to the object to be detected, and the ordinary optical image is simple to obtain and the cost is low. The embodiment of the present application is mainly set as a consultation scenario. The execution subject of the embodiment of the present application is a device or equipment that can conduct a consultation, which can be specifically set on the client or server side.

[0036] 102. Determine that the facial image is a frontal image, input the facial image into a preset feature extraction model for feature extraction, and obtain submicron feature data of each facial tissue and organ to be detected of the object to be detected.

[0037] The facial tissues and organs to be detected refer to the high-definition eyes, nose, mouth, forehead, chin and other facial tissues and organs of the object to be detected.

[0038] For the embodiment of the present application, after obtaining a facial image of the object to be detected at any time of consultation, if the facial image is a frontal image, the facial image is input into a preset feature extraction model for feature extraction to obtain submicro feature data of each facial tissue and organ to be detected of the object to be detected. Then, based on the submicro feature data to be detected, the similarity between each facial tissue and organ to be detected of the object to be detected and each diseased facial tissue and organ of an Alzheimer's patient is calculated. Finally, the doctor determines the consultation result of the object to be detected based on the similarity. Therefore, by analyzing the facial image at any time through the preset feature extraction model, the processing efficiency and accuracy of the consultation information can be improved.

[0039] 103. Obtain diseased submicron feature data of each diseased facial tissue and organ of an Alzheimer's patient, and based on the submicron feature data to be detected and the diseased submicron feature data, respectively calculate the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ, and send the similarity to the doctor terminal. The doctor at the doctor terminal determines the medical consultation result of the subject to be detected based on the similarity.

[0040] Among them, the diseased facial tissues and organs include: high-definition facial tissues and organs such as eyes, nose, mouth, forehead, and chin. For the embodiment of the present application, after determining the submicro feature data to be detected of each facial tissue and organ in the facial image of the object to be detected, in order to provide the doctor with similarity, it is also necessary to obtain the diseased submicro feature data of each diseased facial tissue and organ of the Alzheimer's patient, and then calculate the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ based on each submicro feature data to be detected and each diseased submicro feature data, and finally show the similarity to the doctor, who determines the diagnosis result of the object to be detected based on the similarity. Thus, by determining the diagnosis result through an ordinary optical image of the face of the object to be detected, no psychological and physiological burden is added to the object to be detected, thereby improving the diagnosis experience of the object to be detected, and the ordinary optical facial image is simple to obtain and the cost is low. At the same time, identifying the facial features of the object to be detected through a preset feature extraction model is a convenient and fast recognition method that can assist doctors in improving the efficiency and accuracy of diagnosis.

[0041] According to a method for processing medical inquiry information provided by the present application, compared with the current method of determining medical inquiry results through oral communication between a doctor and an examinee, the present application obtains a facial image of the examinee at any medical inquiry moment by responding to a medical inquiry signal of the examinee; then determines that the facial image is a frontal image, inputs the facial image into a preset feature extraction model for feature extraction, and obtains submicron feature data of each facial tissue and organ to be detected of the examinee; finally obtains diseased submicron feature data of each diseased facial tissue and organ of an Alzheimer's patient, and based on the submicron feature data to be detected and the diseased submicron feature data, respectively calculates the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ, and sends the similarity to the doctor terminal, and the doctor Based on this similarity, the doctor at the terminal determines the diagnosis result for the patient under examination. This is done by obtaining a facial image of the patient under examination at any given time, extracting submicron feature data from the facial image using a feature extraction model, and then calculating similarity between this submicron feature data and the submicron feature data of a patient with Alzheimer's disease. The similarity is then transmitted to the doctor, who ultimately determines the diagnosis result for the patient under examination based on the similarity. This avoids lengthy consultations between the doctor and the patient, thereby improving consultation efficiency. Furthermore, the present application utilizes only standard optical facial images from any given time for analysis, enabling rapid processing of consultation information and reducing consultation costs. Furthermore, the present application utilizes a feature extraction model to analyze facial images, improving the accuracy of processing consultation information.

[0042] Furthermore, to better illustrate the above process of processing medical inquiry information, as a refinement and extension of the above embodiment, the present application embodiment provides another method for processing medical inquiry information, as shown in FIG2 , the method comprising:

[0043] 201. In response to a medical inquiry signal from a subject to be detected, obtain a facial image of the subject to be detected at any medical inquiry time.

[0044] Specifically, when the subject to be tested is undergoing a medical consultation, one or more ordinary optical high-definition cameras or video cameras and other devices are used to collect ordinary optical facial images of the subject to be tested at any time. The ordinary optical facial images are then analyzed to extract submicron feature data from the facial images. Finally, the doctor is assisted in determining the results of the medical consultation based on the submicron feature data. Since ordinary optical facial images can be collected quickly, conveniently, and at low cost, this application can improve medical consultation efficiency and reduce medical consultation costs.

[0045] Furthermore, in order to improve the accuracy of the medical consultation, after obtaining the facial image, it is necessary to perform noise reduction processing on the facial image. Based on this, the method includes: determining the pixel matrix corresponding to the facial image, and calculating the pixel median of each pixel point within a first preset range corresponding to each target pixel point in the pixel matrix; replacing the corresponding target pixel point with the pixel median to obtain a median pixel matrix; calculating the pixel mean of each pixel point within a second preset range corresponding to each median pixel point in the median pixel matrix; replacing the corresponding median pixel point with the pixel mean to obtain a mean pixel matrix; determining the spatial domain weight and pixel value domain weight corresponding to each mean pixel point in the mean pixel matrix; calculating the weighted average of each pixel point within a third preset range corresponding to each mean pixel point based on the spatial domain weight and the pixel value domain weight, and replacing the corresponding mean pixel point with the weighted average to obtain a denoised facial image; inputting the denoised facial image into a preset feature extraction model for feature extraction to obtain the sub-micro feature data to be detected of each facial tissue organ to be detected of the object to be detected.

[0046] Among them, the first preset range, the second preset range, and the third preset range are set according to actual needs, and the first preset range, the second preset range, and the third preset range can be the same or different. Specifically, first, a pixel matrix corresponding to the facial image is determined, and any point in the pixel matrix is ​​determined as a target pixel point. Then, the value of the target pixel point in the pixel matrix is ​​replaced by the median of the values ​​of each point in a range of the point. Each pixel point in the pixel matrix is ​​replaced in the above manner, and finally a median pixel matrix is ​​obtained. Then, any point in the median pixel matrix is ​​determined as a target pixel point. Then, the value of the target pixel point in the median pixel matrix is ​​replaced by the mean of the values ​​of each point in a range of the point. Each pixel point in the pixel matrix is ​​replaced in the above manner, and finally a mean pixel matrix is ​​obtained. Then, the spatial distance and pixel value similarity of each mean pixel point in the mean pixel matrix are comprehensively considered to determine the spatial domain weight and pixel value domain weight corresponding to each mean pixel point. Then, each pixel point in the neighborhood of each mean pixel point is multiplied by the corresponding spatial domain weight and pixel value domain weight respectively, and the multiplication results are added to obtain a weighted average value. Finally, the corresponding mean pixel point is replaced by the weighted average value, thereby obtaining a denoised facial image. Finally, the denoised facial image is input into the preset feature extraction model for feature extraction, thereby improving the accuracy of feature extraction and thus improving the accuracy of medical consultation.

[0047] 202. Determine that the facial image is a frontal image, input the facial image into a preset feature extraction model for feature extraction, and obtain submicron feature data of each facial tissue and organ to be detected of the object to be detected.

[0048] Among them, the preset feature extraction model includes: input layer, convolution layer, batch normalization processing layer, activation layer, maximum pooling layer, residual network layer, mean pooling layer, Flatten layer and fully connected layer.

[0049] For the embodiment of the present application, in order to improve the feature extraction accuracy of the preset feature extraction model, it is first necessary to train and construct the preset feature extraction model. Based on this, the method includes: constructing multiple preset initial feature extraction models, and obtaining sample facial images of sample objects and actual submicron feature data corresponding to each sample facial tissue and organ in the sample facial image, wherein the sample objects include objects without Alzheimer's disease and objects with Alzheimer's disease, and each of the sample facial images is a facial image of the same image type and specification; based on the correspondence between the sample facial images and the actual submicron feature data, constructing training data and test data; according to the number of the preset initial feature extraction models, dividing the training data into multiple groups of training sub-data; using each group of the training sub-data to train each of the preset initial feature extraction models respectively, to obtain each of the trained preset initial feature extraction models; according to the test data, respectively calculating the prediction accuracy of each of the trained preset initial feature extraction models; according to the prediction accuracy, determining the preset feature extraction model in each of the trained preset initial feature extraction models. Among them, each of the preset initial feature extraction models is trained respectively using each group of the training sub-data to obtain each of the preset initial feature extraction models after training, including: inputting the sample facial image in the training sub-data into the preset initial feature extraction model for feature extraction to obtain the predicted submicro feature data corresponding to each sample facial tissue organ in the sample facial image; constructing the loss function corresponding to each of the preset initial feature extraction models based on the actual submicro feature data and the predicted submicro feature data corresponding to the same sample facial tissue organ in the same sample object; and training each of the preset initial feature extraction models based on the loss function to obtain each of the preset initial feature extraction models after training.

[0050] Specifically, first, multiple preset initial feature extraction models are constructed. Specifically, the development of the preset initial feature extraction models can be implemented using the Python language. After completion, the software code is deployed on a high-performance graphics workstation. Among them, the model structures of the multiple preset initial feature extraction models can be the same or different. Then, under legal and compliant conditions, sample facial images are collected. For example, high-definition facial images of no less than 29 patients with Alzheimer's disease, totaling more than 1,100, and high-definition facial images of no less than 41 normal people (i.e., those who do not suffer from Alzheimer's disease), totaling more than 1,300, are randomly collected. The above-collected facial images are used as sample facial images, and the collected facial images are accurately labeled according to Alzheimer's patients and normal people. The collected and labeled sample facial images are unified in type and specification for easy processing. For example, the sample facial images are uniformly set to JPG format images with a width and height of 512 pixels. Then, the actual submicro feature data corresponding to the sample facial images are determined, and based on the correspondence between the sample facial images and the actual submicro feature data, a training data set, a verification data set, and a test data set are constructed according to a preset ratio. Then, the training data set is used according to the preset The number of initial feature extraction models is further divided into multiple sub-training sets, and different sub-training sets correspond to different preset initial feature extraction models. Then, different sub-training sets are used to train the corresponding preset initial feature extraction models. The specific training process is: first, the sample facial image in the sub-training set is input into the corresponding preset initial feature extraction model for feature extraction to obtain predicted sub-micro feature data, and then the predicted sub-micro feature data and the actual sub-micro feature data are compared. According to the comparison result, the loss function corresponding to the preset initial feature extraction model is constructed, and finally, the preset initial feature extraction model is trained according to the loss function, thereby training each preset initial feature extraction model. After the training is completed, the test data in the test data set are used to test each trained preset initial feature extraction model to determine the prediction accuracy corresponding to each preset initial feature extraction model, and finally the preset initial feature extraction model with the highest prediction accuracy is selected as the preset feature extraction model in the embodiment of the present application. For example, during the training process, the learning rate of the preset initial feature extraction model can be set to 1e-4, the number of iterations can be set to 80, the number of samples in a single pass (batch size) can be set to 8, and the optimization function can be set to Adam. Iterative training is performed, and the accuracy and loss values ​​output by the training verification can be shown in Figure 3. When the training verification results converge well, the iterative training is automatically terminated, and the weight parameters and bias parameters of each preset initial feature extraction model after training are finally output. In this way, by training and building a preset feature extraction model, the accuracy of the medical consultation can be improved.

[0051] Furthermore, after collecting the facial image of the object to be detected, when it is determined that the facial image is a frontal image, it is necessary to use a preset feature extraction model to extract the submicro feature data of the frontal image. Based on this, step 202 specifically includes: determining the first feature vector corresponding to each facial tissue organ to be detected in the facial image; inputting the first feature vector into the convolution layer through the input layer for convolution processing to obtain a convolution feature vector; inputting the convolution feature vector into the batch normalization processing layer for batch normalization processing to obtain a batch normalized feature vector; inputting the matching normalized feature vector into the activation layer for activation processing. Active processing is performed to obtain an activated feature vector; the activated feature vector is input into the maximum pooling layer for pooling processing to obtain a pooled feature vector; the pooled feature vector is added to the first image feature vector to obtain a fused feature vector; the fused feature vector is input into the residual network layer for residual processing to obtain a residual feature vector; the residual feature vector is sequentially input into the batch normalization layer, the activation layer, the mean pooling layer, the Flatten layer and the fully connected layer for feature extraction to obtain the sub-micro feature data to be detected of each facial tissue organ to be detected of the object to be detected.

[0052] Specifically, first, the first feature vector corresponding to each facial tissue and organ to be detected in the facial image is obtained, and then the first feature vector is sequentially passed through the convolution layer, batch normalization processing layer, activation layer, and maximum pooling layer for feature extraction to finally obtain a pooled feature vector. Then, the pooled feature vector is added to the first image feature vector to obtain a fused feature vector, and the fused feature vector is input into the residual network layer for residual processing to obtain a residual feature vector. At the same time, the residual network module layer operation is repeatedly used 5-10 times in the entire network model, and the fused feature vector obtained by the repeated residual network module layer operation is sequentially input into the batch normalization processing layer, activation layer, mean pooling layer, Flatten layer and fully connected layer for feature extraction to obtain the sub-micro feature data to be detected of each facial tissue and organ to be detected of the object to be detected.

[0053] In another embodiment of the present application, after collecting the facial image of the object to be detected, when it is determined that the facial image is a side image, in order to determine the diagnosis result, the method also includes: determining that the facial image is a side facial image of the object to be detected, inputting the side facial image into a preset side image extraction model for feature extraction, and obtaining correlated submicron feature data between each side tissue organ in the side facial image; obtaining correlated diseased submicron feature data between each diseased side tissue organ in the side facial image of the Alzheimer's patient, and based on the correlated submicron feature data and the correlated diseased submicron feature data, calculating the side similarity between the facial side of the Alzheimer's patient and the facial side of the object to be detected, and sending the side similarity to the doctor terminal, and the doctor at the doctor terminal determines the diagnosis result of the object to be detected based on the side similarity.

[0054] Among them, before performing feature extraction of the side image, it is necessary to first construct a preset side image extraction model. When constructing the preset side image extraction model, it is necessary to first construct at least one preset initial side image extraction model. Among them, the model structures of the multiple preset initial side image extraction models can be the same or different, and obtain a sample side facial image of the sample object and its corresponding actual associated sub-micro feature data, and based on the correspondence between the sample side facial image and the actual associated sub-micro feature data, according to a preset ratio, construct a training data set, a verification data set, and a test data set, and then divide the training data set into multiple sub-training sets according to the number of preset initial side image extraction models, different sub-training sets correspond to different preset initial side image extraction models, and then use different sub-training sets to train the corresponding preset initial side image extraction models, thereby training each preset initial side image extraction model. After the training is completed, use the test data in the test data set to test each trained preset initial side image extraction model to determine the prediction accuracy corresponding to each preset initial side image extraction model, and finally select the preset initial side image extraction model with the highest prediction accuracy as the preset side image extraction model in the embodiment of the present application. Furthermore, the side face image is input into a preset side face image extraction model for feature extraction. The model can output correlated submicron feature data between various side face tissues and organs in the side face image, and obtain correlated diseased submicron feature data between various diseased side face tissues and organs in the side face image of the Alzheimer's patient. Based on these correlated submicron feature data and the correlated diseased submicron feature data, the similarity between the side face of the Alzheimer's patient and the side face of the subject to be tested is calculated, for example, cosine similarity. The similarity is ultimately sent to the doctor, who uses the similarity to provide a diagnosis result. Therefore, using different models to extract different features for facial images in different orientations can improve diagnosis accuracy.

[0055] 203. Obtain diseased submicron feature data of each diseased facial tissue and organ of an Alzheimer's patient, and calculate the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ based on the submicron feature data to be detected and the diseased submicron feature data.

[0056] 204. Determine the performance weight of each facial tissue and organ to be tested for Alzheimer's disease.

[0057] 205. Based on the performance weight, each similarity is added together to obtain the recognition parameter corresponding to the object to be detected.

[0058] 206. Send the identification parameters to the doctor terminal. The doctor at the doctor terminal determines the diagnosis result of the subject to be detected based on the identification parameters.

[0059] Among them, the performance weight refers to the importance of each facial tissue and organ for the diagnosis of Alzheimer's disease.

[0060] Specifically, after determining the submicro feature data to be detected corresponding to each facial tissue and organ to be detected in the facial image of the object to be detected, it is also necessary to obtain the diseased submicro feature data of each diseased facial tissue and organ of the Alzheimer's patient, and then calculate the similarity between each facial tissue and organ to be detected and the corresponding diseased tissue and organ based on each submicro feature data to be detected and each diseased submicro feature number data. For example, calculate the similarity between each facial tissue and organ to be detected and the corresponding diseased tissue and organ, and determine the performance weight of each facial tissue and organ to be detected for Alzheimer's disease. For example, the similarity between the eyes of the object to be detected and the eyes of the Alzheimer's patient is 0.5, and the performance weight of the eyes for Alzheimer's disease is 0.2, and the similarity between the nose of the object to be detected and the nose of the Alzheimer's patient is 0. The similarity between the noses of the subjects to be tested and the lips of patients with Alzheimer's disease is 0.1, and the performance weight of the nose for Alzheimer's disease is 0.1; the similarity between the lips of the subjects to be tested and the lips of patients with Alzheimer's disease is 0.2, and the performance weight of the lips for Alzheimer's disease is 0.4; the similarity between the foreheads of the subjects to be tested and the foreheads of patients with Alzheimer's disease is 0.4, and the performance weight of the foreheads for Alzheimer's disease is 0.1; the similarity between the chins of the subjects to be tested and the chins of patients with Alzheimer's disease is 0.2, and the performance weight of the chins for Alzheimer's disease is 0.2. The above data can finally calculate the recognition parameter of the subjects to be tested to be 0.27, and finally show the recognition parameter to the consulting doctor, who can accurately give the consultation result of the subjects to be tested based on the recognition parameter.

[0061] According to another method for processing medical inquiry information provided by the present application, compared with the current method of determining medical inquiry results through oral communication between a doctor and an examinee, the present application obtains a facial image of the examinee at any medical inquiry moment by responding to a medical inquiry signal of the examinee; then determines that the facial image is a frontal image, inputs the facial image into a preset feature extraction model for feature extraction, and obtains submicron feature data of each facial tissue and organ to be detected of the examinee; finally obtains diseased submicron feature data of each diseased facial tissue and organ of an Alzheimer's patient, and based on the submicron feature data to be detected and the diseased submicron feature data, respectively calculates the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ, and sends the similarity to the doctor terminal, which The doctor at the patient terminal determines the diagnosis result of the object to be detected based on the similarity, thereby obtaining a facial image of the object to be detected at any diagnosis time, and using a feature extraction model to extract submicro feature data of the facial image, and then calculating the similarity between the submicro feature data and the facial submicro feature data of a patient with Alzheimer's disease, and sending the similarity to the doctor. The doctor finally determines the diagnosis result of the object to be detected based on the similarity, which can avoid long-term diagnosis communication between the doctor and the patient, thereby improving the diagnosis efficiency. At the same time, the present application only uses ordinary optical facial images at any diagnosis time for analysis, so that the diagnosis result can be determined quickly, and the diagnosis cost can also be reduced. In addition, the present application uses a feature extraction model to analyze facial images, which can improve the accuracy of determining the diagnosis result.

[0062] Furthermore, as a specific implementation of FIG1 , an embodiment of the present application provides a medical inquiry information processing device, as shown in FIG4 , which includes: an acquisition unit 31 , an extraction unit 32 and a calculation unit 33 .

[0063] The acquisition unit 31 is configured to acquire a facial image of the subject to be detected at any time of the inquiry in response to the inquiry signal of the subject to be detected.

[0064] The extraction unit 32 is configured to determine that the facial image is a frontal image, input the facial image into a preset feature extraction model to perform feature extraction, and obtain submicron feature data of each facial tissue and organ to be detected of the object to be detected.

[0065] The calculation unit 33 is configured to obtain the diseased submicron feature data of each diseased facial tissue and organ of an Alzheimer's patient, and based on the submicron feature data to be detected and the diseased submicron feature data, respectively calculate the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ, and send the similarity to the doctor terminal. The doctor at the doctor terminal determines the consultation result of the subject to be detected based on the similarity.

[0066] In a specific application scenario, as shown in Figure 5, in order to determine the sub-micro feature data to be detected of each facial tissue organ to be detected, the extraction unit 32 includes a determination module 321, a convolution module 322, a normalization module 323, an activation module 324, a pooling module 325, an addition module 326, a residual module 327, and a feature extraction module 328.

[0067] The determining module 321 is configured to determine a first feature vector corresponding to each facial tissue organ to be detected in the facial image.

[0068] The convolution module 322 is configured to input the first feature vector into the convolution layer through the input layer for convolution processing to obtain a convolution feature vector.

[0069] The normalization module 323 is configured to input the convolution feature vector into the batch normalization processing layer for batch normalization processing to obtain a batch normalized feature vector.

[0070] The activation module 324 is configured to input the matched normalized feature vector into the activation layer for activation processing to obtain an activated feature vector.

[0071] The pooling module 325 is configured to input the activation feature vector into the maximum pooling layer for pooling processing to obtain a pooled feature vector.

[0072] The adding module 326 is configured to add the pooled feature vector and the first image feature vector to obtain a fused feature vector.

[0073] The residual module 327 is configured to input the fused feature vector into the residual network layer for residual processing to obtain a residual feature vector.

[0074] The feature extraction module 328 is configured to sequentially input the residual feature vector into the batch normalization layer, the activation layer, the mean pooling layer, the Flatten layer, and the fully connected layer for feature extraction, thereby obtaining submicron feature data of each facial tissue and organ to be detected of the object to be detected.

[0075] In a specific application scenario, in order to construct a preset feature extraction model, the device further includes: a construction unit 34 , a division unit 35 , a training unit 36 ​​, and a determination unit 37 .

[0076] The construction unit 34 is configured to construct multiple preset initial feature extraction models and obtain sample facial images of sample objects and actual submicron feature data corresponding to each sample facial tissue organ in the sample facial images, wherein the sample objects include objects without Alzheimer's disease and objects with Alzheimer's disease, and each of the sample facial images is a facial image of a uniform image type and specification.

[0077] The construction unit 34 is configured to construct training data and test data based on the correspondence between the sample facial image and the actual sub-micro feature data.

[0078] The division unit 35 is configured to divide the training data into multiple groups of training sub-data according to the number of the preset initial feature extraction models.

[0079] The training unit 36 ​​is configured to use each group of the training sub-data to train each of the preset initial feature extraction models respectively to obtain each of the trained preset initial feature extraction models.

[0080] The calculation unit 33 is configured to calculate the prediction accuracy of each of the trained preset initial feature extraction models based on the test data.

[0081] The determining unit 37 is configured to determine the preset feature extraction model from the trained preset initial feature extraction models according to the prediction accuracy.

[0082] In a specific application scenario, in order to determine each of the preset initial feature extraction models after training, the training unit 36 ​​is configured to input the sample facial image in the training sub-data into the preset initial feature extraction model for feature extraction, and obtain the predicted submicron feature data corresponding to each sample facial tissue organ in the sample facial image; based on the actual submicron feature data and predicted submicron feature data corresponding to the same sample facial tissue organ in the same sample object, a loss function corresponding to each of the preset initial feature extraction models is constructed; based on the loss function, each of the preset initial feature extraction models is trained to obtain each of the preset initial feature extraction models after training.

[0083] In a specific application scenario, in order to perform noise reduction processing on facial images, the device further includes: a replacement unit 38.

[0084] The calculation unit 33 is configured to determine a pixel matrix corresponding to the facial image, and calculate a pixel median value of each pixel point within a first preset range corresponding to each target pixel point in the pixel matrix.

[0085] The replacement unit 38 is configured to replace the corresponding target pixel point with the pixel median to obtain a median pixel matrix.

[0086] The calculation unit 33 is configured to calculate the pixel mean of each pixel point within a second preset range corresponding to each median pixel point in the median pixel matrix.

[0087] The replacement unit 38 is configured to replace the corresponding median pixel point with the pixel mean to obtain a mean pixel matrix.

[0088] The determining unit 37 is configured to determine the spatial domain weight and the pixel value domain weight corresponding to each mean pixel point in the mean pixel matrix.

[0089] The calculation unit 33 is configured to calculate the weighted average of each pixel point within a third preset range corresponding to each mean pixel point based on the spatial domain weight and the pixel value domain weight, and use the weighted average value to replace the corresponding mean pixel point to obtain a denoised facial image.

[0090] The extraction unit 32 is configured to input the denoised facial image into a preset feature extraction model to perform feature extraction, thereby obtaining submicron feature data of each facial tissue and organ to be detected of the object to be detected.

[0091] In a specific application scenario, in order to determine the result of the medical consultation, the determining unit 37 is configured to respectively determine the performance weight of each facial tissue organ to be detected for Alzheimer's disease.

[0092] The calculation unit 33 is configured to add up the similarities based on the performance weight to obtain the identification parameters corresponding to the object to be detected; and send the identification parameters to the doctor terminal, and the doctor at the doctor terminal determines the consultation result of the object to be detected based on the identification parameters.

[0093] In a specific application scenario, in order to determine the medical consultation result corresponding to the side image, the extraction unit 32 is configured to determine that the facial image is a side facial image of the object to be detected, and input the side facial image into a preset side image extraction model for feature extraction to obtain the associated submicron feature data between the various side tissue organs in the side facial image.

[0094] The calculation unit 33 is configured to obtain associated diseased submicroscopic feature data between each diseased side tissue organ in the side face image of the Alzheimer's patient, and based on the associated submicroscopic feature data and the associated diseased submicroscopic feature data, calculate the side similarity between the side face of the Alzheimer's patient and the side face of the object to be detected, and send the side similarity to the doctor terminal. The doctor at the doctor terminal determines the consultation result of the object to be detected based on the side similarity.

[0095] It should be noted that for other corresponding descriptions of the functional modules involved in the medical inquiry information processing device provided in the embodiment of the present application, reference can be made to the corresponding description of the method shown in FIG1 , which will not be repeated here.

[0096] Based on the above method as shown in Figure 1, accordingly, the embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, which implements the following steps when executed by a processor: in response to a medical inquiry signal of the object to be detected, obtaining a facial image of the object to be detected at any medical inquiry time; determining that the facial image is a frontal image, inputting the facial image into a preset feature extraction model for feature extraction, and obtaining submicron feature data to be detected of each facial tissue and organ to be detected of the object to be detected; obtaining diseased submicron feature data of each diseased facial tissue and organ of an Alzheimer's patient, and based on the submicron feature data to be detected and the diseased submicron feature data, respectively calculating the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ, and sending the similarity to the doctor terminal, and the doctor at the doctor terminal determines the medical inquiry result of the object to be detected based on the similarity.

[0097] Based on the embodiment of the method shown in Figure 1 and the device shown in Figure 4, the embodiment of the present application also provides an entity structure diagram of a computer device, as shown in Figure 6, the computer device includes: a processor 41, a memory 42, and a computer program stored in the memory 42 and executable on the processor, wherein the memory 42 and the processor 41 are both arranged on a bus 43, and when the processor 41 executes the program, the following steps are implemented: in response to a medical inquiry signal of the subject to be detected, a facial image of the subject to be detected at any medical inquiry time is obtained; the facial image is determined to be a frontal image, and the facial image is input into a preset feature extraction model for feature extraction to obtain submicron feature data of each facial tissue and organ to be detected of the subject to be detected; the diseased submicron feature data of each diseased facial tissue and organ of an Alzheimer's patient is obtained, and based on the submicron feature data to be detected and the diseased submicron feature data, the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ is calculated, and the similarity is sent to a doctor terminal, and the doctor at the doctor terminal determines the medical inquiry result of the subject to be detected based on the similarity.

[0098] Through the technical solution of the present application, the present application obtains the facial image of the subject to be detected at any time of the inquiry by responding to the inquiry signal of the subject to be detected; then determines that the facial image is a frontal image, inputs the facial image into a preset feature extraction model for feature extraction, and obtains the submicro feature data of each facial tissue and organ to be detected of the subject to be detected; finally obtains the diseased submicro feature data of each diseased facial tissue and organ of the patient with Alzheimer's disease, and based on the submicro feature data to be detected and the diseased submicro feature data, respectively calculates the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ, and sends the similarity to the doctor terminal, and the doctor at the doctor terminal determines the submicro feature data of the patient to be detected based on the similarity. The medical consultation result of the object is obtained by obtaining a facial image of the object to be detected at any time of the consultation, and using a feature extraction model to extract submicro feature data of the facial image, and then calculating the similarity between the submicro feature data and the facial submicro feature data of a patient with Alzheimer's disease, and sending the similarity to the doctor. The doctor finally determines the medical consultation result of the object to be detected based on the similarity, which can avoid long-term consultation communication between the doctor and the patient, thereby improving the consultation efficiency. At the same time, the present application only uses ordinary optical facial images at any time of the consultation for analysis, so that the medical consultation result can be determined quickly, and the consultation cost can also be reduced. In addition, the present application uses a feature extraction model to analyze the facial image, which can improve the accuracy of determining the medical consultation result.

[0099] Obviously, those skilled in the art should understand that the modules or steps of the present application described above can be implemented using a general-purpose computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices. Alternatively, they can be implemented using program code executable by the computing device, so that they can be stored in a storage device and executed by the computing device. In some cases, the steps shown or described can be performed in a different order than herein, or they can be made into separate integrated circuit modules, or multiple modules or steps can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0100] The above description is merely a preferred embodiment of the present application and is not intended to limit the present application. Persons skilled in the art will readily appreciate that various modifications and variations are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

Claims

1. A method for processing consultation information, wherein, Including: In response to the interrogation signal of the object to be detected, obtain the facial image of the object to be detected at any interrogation moment; Determine that the facial image is a frontal image, input the facial image into a preset feature extraction model for feature extraction, and obtain the sub-micro feature data to be detected of each facial tissue organ to be detected of the object to be detected; Obtain the sub-micro feature data of each diseased facial tissue organ of Alzheimer's disease patients, and respectively calculate the similarity between each facial tissue organ to be detected and the corresponding diseased facial tissue organ based on the sub-micro feature data to be detected and the sub-micro feature data of the diseased, and send the similarity to the doctor terminal. The doctor at the doctor terminal determines the interrogation result of the object to be detected based on the similarity.

2. The method according to claim 1, wherein The preset feature extraction model includes: an input layer, a convolutional layer, a batch normalization processing layer, an activation layer, a max pooling layer, a residual network layer, an average pooling layer, a Flatten layer, and a fully connected layer; the step of inputting the facial image into the preset feature extraction model for feature extraction to obtain the sub-micro feature data to be detected of each facial tissue organ to be detected of the object to be detected includes: Determine the first feature vector corresponding to each facial tissue organ to be detected in the facial image; Input the first feature vector into the convolutional layer through the input layer for convolutional processing to obtain a convolutional feature vector; Input the convolutional feature vector into the batch normalization processing layer for batch normalization processing to obtain a batch-normalized feature vector; Input the matched normalized feature vector into the activation layer for activation processing to obtain an activation feature vector; Input the activation feature vector into the max pooling layer for pooling processing to obtain a pooled feature vector; Add the pooled feature vector to the first picture feature vector to obtain a fused feature vector; Input the fused feature vector into the residual network layer for residual processing to obtain a residual feature vector; Input the residual feature vector into the batch normalization processing layer, the activation layer, the average pooling layer, the Flatten layer, and the fully connected layer in sequence for feature extraction to obtain the sub-micro feature data to be detected of each facial tissue organ to be detected of the object to be detected.

3. The method according to claim 1, wherein, Before the step of inputting the facial image into the preset feature extraction model for feature extraction to obtain the sub-micro feature data to be detected of each facial tissue organ to be detected of the object to be detected, the method further includes: Construct a plurality of preset initial feature extraction models, and obtain the sample facial images of the sample objects and the actual sub-micro feature data corresponding to each sample facial tissue organ in the sample facial images, where the sample objects include objects without Alzheimer's disease and objects with Alzheimer's disease, and each of the sample facial images is a facial image of the same image type and specification; Construct training data and test data based on the correspondence between the sample facial images and the actual sub-micro feature data; Divide the training data into multiple groups of training sub-data according to the number of the preset initial feature extraction models; Train each of the preset initial feature extraction models using each group of the training sub-data to obtain the trained preset initial feature extraction models; Calculate the prediction accuracy of each of the trained preset initial feature extraction models based on the test data; Determine the preset feature extraction model from the trained preset initial feature extraction models according to the prediction accuracy; 4. The method according to claim 3, wherein, The step of training each of the preset initial feature extraction models using each group of the training sub-data to obtain the trained preset initial feature extraction models includes: Input the sample facial images in the training sub-data into the preset initial feature extraction model for feature extraction to obtain the predicted sub-micro feature data corresponding to each sample facial tissue and organ in the sample facial images; Construct the loss function corresponding to each preset initial feature extraction model based on the actual sub-micro feature data and the predicted sub-micro feature data corresponding to the same sample facial tissue and organ in the same sample object; Train each of the preset initial feature extraction models based on the loss function to obtain the trained preset initial feature extraction models; 5. The method according to claim 1, wherein, Before inputting the facial image into the preset feature extraction model for feature extraction to obtain the sub-micro feature data to be detected of each facial tissue and organ to be detected of the object to be detected, the method further includes: Determine the pixel matrix corresponding to the facial image and calculate the pixel median of each pixel point within a first preset range corresponding to each target pixel point in the pixel matrix; Replace the corresponding target pixel point with the pixel median to obtain a median pixel matrix; Calculate the pixel mean of each pixel point within a second preset range corresponding to each median pixel point in the median pixel matrix; Replace the corresponding median pixel point with the pixel mean to obtain a mean pixel matrix; Determine the spatial domain weight and the pixel value domain weight corresponding to each mean pixel point in the mean pixel matrix; Calculate the weighted average of each pixel point within a third preset range corresponding to each mean pixel point based on the spatial domain weight and the pixel value domain weight, and replace the corresponding mean pixel point with the weighted average to obtain a denoised facial image; The step of inputting the facial image into the preset feature extraction model for feature extraction to obtain the sub-micro feature data to be detected of each facial tissue and organ to be detected of the object to be detected includes: Input the denoised facial image into the preset feature extraction model for feature extraction to obtain the sub-micro feature data to be detected of each facial tissue and organ to be detected of the object to be detected; 6. The method according to claim 1, wherein After calculating the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ respectively based on the sub-micro feature data to be detected and the diseased sub-micro feature data, the method further includes: Determine the manifestation weight of each facial tissue and organ to be detected for Alzheimer's disease respectively; Add up each similarity based on the manifestation weight to obtain the recognition parameter corresponding to the object to be detected; Sending the similarity to the doctor terminal, and a doctor at the doctor terminal determining a medical consultation result of the object to be detected based on the similarity, includes: Sending the recognition parameters to the doctor terminal, and a doctor at the doctor terminal determining a medical consultation result of the object to be detected based on the recognition parameters.

7. The method according to claim 1, wherein After obtaining a facial image of the object to be detected at any medical consultation moment, the method further includes: Determining that the facial image is a side facial image of the object to be detected, inputting the side facial image into a preset side image extraction model for feature extraction, and obtaining associated sub-micro feature data between various side tissue organs in the side facial image; Obtaining associated diseased sub-micro feature data between various diseased side tissue organs in a side facial image of an Alzheimer's disease patient, calculating a side similarity between the side of the Alzheimer's disease patient's face and the side of the face of the object to be detected based on the associated sub-micro feature data and the associated diseased sub-micro feature data, and sending the side similarity to the doctor terminal, and a doctor at the doctor terminal determining a medical consultation result of the object to be detected based on the side similarity.

8. An inquiry information processing device, wherein, Including: An acquisition unit, configured to respond to a medical consultation signal of the object to be detected and acquire a facial image of the object to be detected at any medical consultation moment; An extraction unit, configured to determine that the facial image is a front image, input the facial image into a preset feature extraction model for feature extraction, and obtain sub-micro feature data of each to-be-detected facial tissue organ of the object to be detected data; A calculation unit, configured to obtain sub-micro feature data of each diseased facial tissue organ of an Alzheimer's disease patient, calculate a similarity between each to-be-detected facial tissue organ and the corresponding diseased facial tissue organ respectively based on the to-be-detected sub-micro feature data and the diseased sub-micro feature data, and send the similarity to the doctor terminal, and a doctor at the doctor terminal determining a medical consultation result of the object to be detected based on the similarity.

9. A computer-readable storage medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, it implements a medical consultation information processing method, including: Responding to a medical consultation signal of the object to be detected and acquiring a facial image of the object to be detected at any medical consultation moment; Determining that the facial image is a front image, inputting the facial image into a preset feature extraction model for feature extraction, and obtaining sub-micro feature data of each to-be-detected facial tissue organ of the object to be detected; Obtaining sub-micro feature data of each diseased facial tissue organ of an Alzheimer's disease patient, calculating a similarity between each to-be-detected facial tissue organ and the corresponding diseased facial tissue organ respectively based on the to-be-detected sub-micro feature data and the diseased sub-micro feature data, and sending the similarity to the doctor terminal, and a doctor at the doctor terminal determining a medical consultation result of the object to be detected based on the similarity.

10. The computer-readable storage medium according to claim 9, wherein, When the computer-readable instructions are executed by a processor, the implementation of the preset feature extraction model includes: an input layer, a convolutional layer, a batch normalization processing layer, an activation layer, a max pooling layer, a residual network layer, an average pooling layer, a Flatten layer, and a fully connected layer; the step of inputting the facial image into the preset feature extraction model for feature extraction to obtain the to-be-detected sub-micro feature data of each to-be-detected facial tissue and organ of the to-be-detected object includes: Determine the first feature vector corresponding to each to-be-detected facial tissue and organ in the facial image; Input the first feature vector into the convolutional layer through the input layer for convolutional processing to obtain a convolutional feature vector; Input the convolutional feature vector into the batch normalization processing layer for batch normalization processing to obtain a batch-normalized feature vector; Input the matched normalized feature vector into the activation layer for activation processing to obtain an activation feature vector; Input the activation feature vector into the max pooling layer for pooling processing to obtain a pooled feature vector; Add the pooled feature vector to the first picture feature vector to obtain a fused feature vector; Input the fused feature vector into the residual network layer for residual processing to obtain a residual feature vector; Input the residual feature vector into the batch normalization processing layer, the activation layer, the average pooling layer, the Flatten layer, and the fully connected layer in sequence for feature extraction to obtain the to-be-detected sub-micro feature data of each to-be-detected facial tissue and organ of the to-be-detected object.

11. The computer-readable storage medium according to claim 9, wherein, Before the step of inputting the facial image into the preset feature extraction model for feature extraction to obtain the to-be-detected sub-micro feature data of each to-be-detected facial tissue and organ of the to-be-detected object, when the computer-readable instructions are executed by a processor, the method further includes: Construct a plurality of preset initial feature extraction models, and obtain the sample facial images of the sample objects and the actual sub-micro feature data corresponding to each sample facial tissue and organ in the sample facial images, where the sample objects include objects without Alzheimer's disease and objects with Alzheimer's disease, and each of the sample facial images is a facial image of the same image type and specification; Construct training data and test data based on the correspondence between the sample facial images and the actual sub-micro feature data; Divide the training data into multiple groups of training sub-data according to the number of the preset initial feature extraction models; Use each group of the training sub-data to train each of the preset initial feature extraction models to obtain the trained preset initial feature extraction models; Calculate the prediction accuracy of each of the trained preset initial feature extraction models according to the test data; Determine the preset feature extraction model among the trained preset initial feature extraction models according to the prediction accuracy.

12. The computer-readable storage medium according to claim 11, wherein, When the computer-readable instructions are executed by a processor, the implementation of using each group of the training sub-data to train each of the preset initial feature extraction models to obtain the trained preset initial feature extraction models includes: Input the sample facial images in the training sub-data into the preset initial feature extraction model for feature extraction to obtain the predicted sub-micro feature data corresponding to each sample facial tissue and organ in the sample facial images; Based on the actual sub-micro feature data and the predicted sub-micro feature data corresponding to the same sample facial tissue and organ in the same sample object, construct the loss function corresponding to each preset initial feature extraction model; Based on the loss function, train each preset initial feature extraction model to obtain the trained preset initial feature extraction models.

13. The computer-readable storage medium according to claim 9, wherein, Before the computer-readable instructions are executed by the processor to input the facial image into the preset feature extraction model for feature extraction to obtain the to-be-detected sub-micro feature data of each to-be-detected facial tissue and organ of the to-be-detected object, the method further includes: Determine the pixel matrix corresponding to the facial image, and calculate the pixel median of each pixel point within a first preset range corresponding to each target pixel point in the pixel matrix; Replace the corresponding target pixel point with the pixel median to obtain a median pixel matrix; Calculate the pixel mean of each pixel point within a second preset range corresponding to each median pixel point in the median pixel matrix; Replace the corresponding median pixel point with the pixel mean to obtain a mean pixel matrix; Determine the spatial domain weight and the pixel value domain weight corresponding to each mean pixel point in the mean pixel matrix; Based on the spatial domain weight and the pixel value domain weight, calculate the weighted average of each pixel point within a third preset range corresponding to each mean pixel point, and replace the corresponding mean pixel point with the weighted average to obtain a denoised facial image; The step of inputting the facial image into the preset feature extraction model for feature extraction to obtain the to-be-detected sub-micro feature data of each to-be-detected facial tissue and organ of the to-be-detected object includes: Input the denoised facial image into the preset feature extraction model for feature extraction to obtain the to-be-detected sub-micro feature data of each to-be-detected facial tissue and organ of the to-be-detected object.

14. The computer-readable storage medium according to claim 9, wherein, Before the computer-readable instructions are executed by the processor to calculate the similarity between each to-be-detected facial tissue and organ and the corresponding diseased facial tissue and organ respectively based on the to-be-detected sub-micro feature data and the diseased sub-micro feature data, the method further includes: Respectively determine the performance weights of each to-be-detected facial tissue and organ for Alzheimer's disease; Based on the performance weights, add up each similarity to obtain the recognition parameter corresponding to the to-be-detected object; The step of sending the similarity to the doctor terminal, and the doctor of the doctor terminal determines the consultation result of the to-be-detected object based on the similarity, includes: Send the recognition parameter to the doctor terminal, and the doctor of the doctor terminal determines the consultation result of the to-be-detected object based on the recognition parameter.

15. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein, When the computer-readable instructions are executed by the processor, the consultation information processing method is implemented, including: In response to the consultation signal of the to-be-detected object, obtain the facial image of the to-be-detected object at any consultation moment; Determine that the facial image is a frontal image, input the facial image into a preset feature extraction model for feature extraction, and obtain the sub-micro feature data of each facial tissue and organ to be detected of the object to be detected; Obtain the sub-micro feature data of each diseased facial tissue and organ of Alzheimer's disease patients, and calculate the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ based on the sub-micro feature data to be detected and the sub-micro feature data of the disease, and send the similarity to the doctor's terminal. The doctor at the doctor's terminal determines the consultation result of the object to be detected based on the similarity.

16. The computer device according to claim 15, wherein, When the computer-readable instructions are executed by a processor, the preset feature extraction model includes: an input layer, a convolutional layer, a batch normalization processing layer, an activation layer, a max pooling layer, a residual network layer, an average pooling layer, a Flatten layer, and a fully connected layer; The process of inputting the facial image into a preset feature extraction model for feature extraction to obtain the sub-micro facial feature data of the object to be detected includes: Determine the first picture feature vector corresponding to the facial image; Input the first picture feature vector into the convolutional layer through the input layer for convolutional processing to obtain a convolutional feature vector; Input the convolutional feature vector into the batch normalization processing layer for batch normalization processing to obtain a batch normalization feature vector; Input the matching normalized feature vector into the activation layer for activation processing to obtain an activation feature vector; Input the activation feature vector into the max pooling layer for pooling processing to obtain a pooled feature vector; Add the pooled feature vector and the first picture feature vector to obtain a fused feature vector; Input the fused feature vector into the residual network layer for residual processing to obtain a residual feature vector; Input the residual feature vector into the batch normalization processing layer, the activation layer, the average pooling layer, the Flatten layer, and the fully connected layer in sequence for feature extraction to obtain the sub-micro facial feature data of the object to be detected.

17. The computer device according to claim 15, wherein, Before the computer-readable instructions are executed by a processor to implement the process of inputting the facial image into a preset feature extraction model for feature extraction to obtain the sub-micro feature data of each facial tissue and organ to be detected of the object to be detected, the method further includes: Construct multiple preset initial feature extraction models, and obtain the sample facial images of the sample objects and the actual sub-micro feature data corresponding to each sample facial tissue and organ in the sample facial images, where the sample objects include objects without Alzheimer's disease and objects with Alzheimer's disease, and each of the sample facial images is a facial image of the same image type and specification; Construct training data and test data based on the correspondence between the sample facial images and the actual sub-micro feature data; Divide the training data into multiple groups of training sub-data according to the number of the preset initial feature extraction models; Use each group of the training sub-data to train each of the preset initial feature extraction models to obtain each of the trained preset initial feature extraction models; Calculate the prediction accuracy of each of the trained preset initial feature extraction models respectively according to the test data; Determine the preset feature extraction model among the trained preset initial feature extraction models according to the prediction accuracy.

18. The computer device according to claim 17, wherein, When the computer-readable instructions are executed by a processor, the implementation of using each group of the training sub-data to train each of the preset initial feature extraction models to obtain the trained preset initial feature extraction models respectively includes: Input the sample facial images in the training sub-data into the preset initial feature extraction model for feature extraction to obtain the predicted sub-micro feature data corresponding to each sample facial tissue and organ in the sample facial images; Based on the actual sub-micro feature data and the predicted sub-micro feature data corresponding to the same sample facial tissue and organ in the same sample object, construct the loss function corresponding to each preset initial feature extraction model; Based on the loss function, train each of the preset initial feature extraction models to obtain the trained preset initial feature extraction models respectively.

19. The computer device according to claim 15, wherein, Before the step of inputting the facial image into the preset feature extraction model for feature extraction to obtain the sub-micro feature data to be detected of each facial tissue and organ to be detected of the object to be detected, the method further includes when the computer-readable instructions are executed by a processor: Determine the pixel matrix corresponding to the facial image, and calculate the pixel median of each pixel point within a first preset range corresponding to each target pixel point in the pixel matrix; Replace the corresponding target pixel point with the pixel median to obtain a median pixel matrix; Calculate the pixel mean of each pixel point within a second preset range corresponding to each median pixel point in the median pixel matrix; Replace the corresponding median pixel point with the pixel mean to obtain a mean pixel matrix; Determine the spatial domain weight and the pixel value domain weight corresponding to each mean pixel point in the mean pixel matrix; Based on the spatial domain weight and the pixel value domain weight, calculate the weighted average of each pixel point within a third preset range corresponding to each mean pixel point, and replace the corresponding mean pixel point with the weighted average to obtain a denoised facial image; The step of inputting the facial image into the preset feature extraction model for feature extraction to obtain the sub-micro feature data to be detected of each facial tissue and organ to be detected of the object to be detected includes: Input the denoised facial image into the preset feature extraction model for feature extraction to obtain the sub-micro feature data to be detected of each facial tissue and organ to be detected of the object to be detected.

20. The computer device according to claim 15, wherein, After the step of calculating the similarity between each facial tissue and organ to be detected and the corresponding diseased facial tissue and organ respectively based on the sub-micro feature data to be detected and the diseased sub-micro feature data, the method further includes when the computer-readable instructions are executed by a processor: Determine the manifestation weight of each facial tissue and organ to be detected for Alzheimer's disease respectively; Based on the manifestation weight, add up each similarity to obtain the recognition parameter corresponding to the object to be detected; Sending the similarity to the doctor terminal, and the doctor at the doctor terminal determines the consultation result of the object to be detected based on the similarity, including: Sending the recognition parameter to the doctor terminal, and the doctor at the doctor terminal determines the consultation result of the object to be detected based on the recognition parameter.

Images