Interaction method, device, equipment, storage medium and computer program product

Abstract

The embodiment of the invention provides an interaction method and device, equipment, a storage medium and a computer program product, and the method comprises the steps: displaying a first interaction interface which comprises an image display region and an answer region; displaying the medical image to be analyzed in the image display area; displaying at least one symptom identification question and a disease diagnosis question corresponding to the medical image in an answering area; and in response to an answering operation performed by the user in the answering area, obtaining first answering information of each symptom identification question and second answering information of the disease diagnosis question, and determining an answering result of the user for the medical image based on the first answering information, the second answering information and answer information corresponding to the medical image. According to the embodiment of the invention, the method helps to improve the capability of a user to perform symptom recognition according to a medical image and obtain a diagnosis result according to the symptom, and can provide a clear and understandable diagnosis basis and learning cases for the training of medical image diagnosis and the autonomous learning of students.

Description

Technical Field

[0001] The present application relates to, but is not limited to, the field of information technology, and in particular to an interaction method, apparatus, device, storage medium, and computer program product. Background Art

[0002] With the development of medical technology, the demand for medical imaging diagnosis training and student autonomous learning is increasing. However, the medical imaging diagnosis training or learning programs in related technologies provide poor interpretability of diagnostic results, which is not conducive to students' autonomous learning and understanding. Summary of the invention

[0003] In view of this, the embodiments of the present application provide at least one interactive method, apparatus, device, storage medium and computer program product, which help to improve the user's ability to identify disease symptoms based on medical images and draw diagnosis results based on the disease symptoms, and can provide clear and understandable diagnostic basis and learning cases for medical imaging diagnosis training and students' independent learning.

[0004] The technical solution of the embodiment of the present application is implemented as follows:

[0005] The present application provides an interactive method, the method comprising:

[0006] Displaying a first interactive interface; the first interactive interface includes an image display area and a question answering area;

[0007] Displaying the medical image to be analyzed in the image display area;

[0008] Displaying at least one disease symptom identification question and disease diagnosis question corresponding to the medical image in the question answering area;

[0009] In response to the user's answering operation in the answering area, the first answer information of each disease symptom identification question and the second answer information of the disease diagnosis question are obtained, and based on the first answer information, the second answer information and the answer information corresponding to the medical image, the user's answer result for the medical image is determined.

[0010] The present application provides an interactive device, the device comprising:

[0011] A first display module, used to display a first interactive interface; the first interactive interface includes an image display area and a question answering area;

[0012] A second display module, used for displaying the medical image to be analyzed in the image display area;

[0013] A third display module, configured to display at least one disease symptom identification question and a disease diagnosis question corresponding to the medical image in the question answering area;

[0014] A determination module is used to respond to the user's answering operation in the answering area, obtain the first answer information of each of the disease symptom identification questions and the second answer information of the disease diagnosis question, and determine the user's answer result for the medical image based on the first answer information, the second answer information and the answer information corresponding to the medical image.

[0015] An embodiment of the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the program, some or all of the steps in the above method are implemented.

[0016] An embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, some or all of the steps in the above method are implemented.

[0017] An embodiment of the present application provides a computer program product, including a computer program or instructions, which, when executed by a processor, implements some or all of the steps in the above method.

[0018] In the embodiment of the present application, a first interactive interface is displayed, and the first interactive interface includes an image display area and an answering area; the medical image to be analyzed is displayed in the image display area; at least one disease symptom identification question and a disease diagnosis question corresponding to the medical image are displayed in the answering area; in response to the answering operation performed by the user in the answering area, the first answer information of each disease symptom identification question and the second answer information of the disease diagnosis question are obtained, and based on each first answer information, the second answer information and the answer information corresponding to the medical image, the user's answer result for the medical image is determined. In this way, the user can be guided to complete the answer to each disease symptom identification question by analyzing the disease symptoms contained in the medical image, and answer the disease diagnosis question based on the answer to the disease symptom identification question, so as to obtain the final disease diagnosis result. In this way, it is helpful to improve the user's ability to identify disease symptoms based on medical images and obtain diagnostic results based on the disease symptoms, and can provide clear and understandable diagnostic basis and learning cases for medical image diagnosis training and students' autonomous learning.

[0019] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present application and are used together with the specification to illustrate the technical solution of the present application.

[0021] Figure 1A schematic diagram of an implementation flow of an interaction method provided in an embodiment of the present application;

[0022] Figure 2 A schematic diagram of a first interactive interface provided in an embodiment of the present application;

[0023] Figure 3 A schematic diagram of the implementation architecture of a medical imaging diagnosis autonomous learning system provided in an embodiment of the present application;

[0024] Figure 4 A schematic diagram of a display interface for answering questions provided in an embodiment of the present application;

[0025] Figure 5 A schematic diagram of a learning progress viewing interface provided for the implementation of this application;

[0026] Figure 6 A schematic diagram of the structure of an interactive device provided in an embodiment of the present application;

[0027] Figure 7 A schematic diagram of a hardware entity of a computer device in an embodiment of the present application. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solutions and advantages of the present application clearer, the technical solutions of the present application are further elaborated in detail below in conjunction with the drawings and embodiments. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0029] In the following description, reference is made to “some embodiments”, which describe a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0030] The terms "first / second / third" involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It is understandable that "first / second / third" can be interchanged with a specific order or sequence where permitted so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0031] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing this application and are not intended to limit this application.

[0032] The embodiment of the present application provides an interactive method, which can be executed by a computer device with a display function. The computer device may include but is not limited to at least one of a server, a laptop, a tablet computer, a desktop computer, a smart TV, a mobile device (such as a mobile phone, a portable video player, a personal digital assistant, a dedicated messaging device, a portable gaming device), etc. Figure 1 A schematic diagram of an implementation flow of an interactive method provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method may include the following steps S101 to S104:

[0033] Step S101, displaying a first interactive interface; the first interactive interface includes an image display area and a question answering area.

[0034] Here, the image display area may be an interface area in the first interactive interface for displaying medical images, and the question answering area may be an area in the first interactive interface for displaying test questions.

[0035] During implementation, the layout of the image display area and the question answering area in the first interactive interface can be any suitable layout, and the embodiments of the present application are not limited to this. For example, the image display area can be located on the left side of the first interactive interface, and the question answering area can be located on the right side of the first interactive interface. For another example, the image display area can be located on the upper side of the first interactive interface, and the question answering area can be located on the lower side of the first interactive interface. For another example, the image display area can be located on the lower side of the first interactive interface, and the question answering area can be located on the upper side of the first interactive interface.

[0036] Step S102: displaying the medical image to be analyzed in the image display area.

[0037] Here, the medical image to be analyzed may be any suitable medical image for users to conduct medical image training or learning, wherein the medical image may include but is not limited to X-ray images, computed tomography (CT) images, nuclear magnetic resonance imaging (MRI) images, and the like.

[0038] In some embodiments, the medical image corresponds to at least one disease type, and the medical image may include at least one disease symptom to be identified.

[0039] Step S103, displaying at least one disease symptom identification question and disease diagnosis question corresponding to the medical image in the question answering area.

[0040] Here, the disease symptom recognition questions are used to test the ability to recognize a disease symptom contained in a medical image, and the disease diagnosis questions are used to test the ability to diagnose the disease type corresponding to the medical image. It can be understood that the disease symptom in the embodiments of the present application refers to the characteristics of the disease or lesion manifested in the medical image.

[0041] During implementation, the question types of symptom identification questions and disease diagnosis questions can be determined based on actual conditions, and the embodiments of the present application are not limited to this.

[0042] In some embodiments, the question types of the disease symptom identification questions and disease diagnosis questions may include, but are not limited to, multiple choice questions, fill-in-the-blank questions, and / or essay questions. For example, the question type of the disease symptom identification question is a multiple choice question, and the question type of the disease diagnosis question is an essay question. For another example, the question type of the disease symptom identification question is a fill-in-the-blank question, and the question type of the disease diagnosis question is a fill-in-the-blank question.

[0043] In some embodiments, the medical image to be analyzed and at least one disease sign identification question and disease diagnosis question corresponding to the medical image can be selected from an examination question bank. During implementation, any suitable selection strategy can be used to select the medical image to be analyzed and at least one disease sign identification question and disease diagnosis question corresponding to the medical image from the examination question bank, and the embodiments of the present application are not limited to this. For example, a medical image to be analyzed and at least one disease sign identification question and disease diagnosis question corresponding to the medical image can be selected from the examination question bank in a random selection manner. For another example, according to a set disease type, a medical image matching the disease type can be selected from the examination question bank as the medical image to be analyzed, and at least one disease sign identification question and disease diagnosis question corresponding to the medical image can be obtained.

[0044] In some embodiments, the medical image to be analyzed and at least one disease sign identification question and disease diagnosis question corresponding to the medical image may be generated in real time based on existing medical diagnosis data. Medical diagnosis data may include, but are not limited to, disease diagnosis reports or disease diagnosis cases. For example, a medical image, at least one disease sign contained in the medical image, and the type of disease corresponding to the medical image may be extracted from a disease diagnosis report, and at least one disease sign identification question corresponding to the medical image may be generated based on the at least one disease sign contained in the medical image, and a disease diagnosis question corresponding to the medical image may be generated based on the disease type corresponding to the medical image.

[0045] In some embodiments, the medical image to be analyzed and at least one disease sign identification question and disease diagnosis question corresponding to the medical image can be automatically generated based on the set disease knowledge using a neural network model. For example, the medical image to be analyzed and at least one disease sign identification question and disease diagnosis question corresponding to the medical image can be generated based on typical brain image features of stroke disease.

[0046] Figure 2 This is a schematic diagram of a first interactive interface provided in an embodiment of the present application. Figure 2 As shown, the first interactive interface 100 includes an image display area 110 and a question answering area 120. The medical image 111 to be analyzed can be displayed in the image display area 110, and at least one disease symptom identification question 121 and a disease diagnosis question 122 corresponding to the medical image 111 can be displayed in the question answering area 120.

[0047] In some embodiments, see Figure 2 The medical image 111 displayed in the image display area 110 may include multiple associated sub-images 111a, such as multiple sub-images 111a taken from different angles of the same lesion. The image display area 110 also supports a comparison function for the multiple sub-images 111a. The user can view and compare the multiple sub-images 111a at the same time, conduct a comprehensive analysis, and improve the accuracy of diagnosis.

[0048] Step S104, in response to the user's answering operation in the answering area, obtain the first answer information of each of the disease symptom identification questions and the second answer information of the disease diagnosis question, and determine the user's answer result for the medical image based on the first answer information, the second answer information and the answer information corresponding to the medical image.

[0049] Here, the answering operations performed by the user in the question answering area may include but are not limited to answering operations to symptom identification questions and / or answering operations to disease diagnosis questions.

[0050] During implementation, the answering operation performed by the user in the question answering area may be determined based on actual conditions, and the embodiments of the present application do not limit this.

[0051] In some embodiments, the answering operation performed by the user in the answering area may be determined according to the question type of the disease symptom identification question and / or the disease diagnosis question. For example, when the question type of the disease symptom identification question is a multiple-choice question, the answering operation for the disease symptom identification question may be an operation of selecting an option. For another example, when the question type of the disease diagnosis question is a fill-in-the-blank question, the answering operation for the disease diagnosis question may be an operation of inputting text.

[0052] The answer information corresponding to the medical image may include answer information for disease symptom identification questions and / or answer information for disease diagnosis questions.

[0053] In some embodiments, the answer information corresponding to the medical image may include first answer information for disease sign identification questions and second answer information for disease diagnosis questions. Based on the first answer information and the first answer information of each disease sign identification question, the answer result of each disease sign identification question can be determined. Based on the second answer information and the second answer information of the disease diagnosis question, the answer result of the disease diagnosis question can be determined. Thus, based on the answer results of each disease sign identification question and the answer results of the disease diagnosis question, the user's answer result for the medical image can be determined.

[0054] In some embodiments, the answer information corresponding to the medical image may include second answer information for disease diagnosis questions; the first answer information for each disease symptom identification question may be determined based on the second answer information for the disease diagnosis question, and thus the user's answer results for the medical image may be determined based on each first answer information and each first answer information, second answer information, and second answer information.

[0055] In some embodiments, the answer information corresponding to the medical image may include the first answer information of each disease sign identification question; the second answer information of the disease diagnosis question can be determined based on the first answer information of each disease sign identification question, and thus the user's answer result for the medical image can be determined based on each first answer information and each first answer information, second answer information, and second answer information.

[0056] In some embodiments, the answer results of medical imaging questions may include, but are not limited to, at least one of the answer accuracy corresponding to each disease sign identification question, the overall answer accuracy of each disease sign identification question, the answer accuracy of the disease diagnosis question, the similarity between the answer result of the disease diagnosis question and the second answer information, etc.

[0057] In the embodiment of the present application, a first interactive interface is displayed, and the first interactive interface includes an image display area and an answering area; the medical image to be analyzed is displayed in the image display area; at least one disease symptom identification question and a disease diagnosis question corresponding to the medical image are displayed in the answering area; in response to the answering operation performed by the user in the answering area, the first answer information of each disease symptom identification question and the second answer information of the disease diagnosis question are obtained, and based on each first answer information, the second answer information and the answer information corresponding to the medical image, the user's answer result for the medical image is determined. In this way, the user can be guided to complete the answer to each disease symptom identification question by analyzing the disease symptoms contained in the medical image, and answer the disease diagnosis question based on the answer to the disease symptom identification question, so as to obtain the final disease diagnosis result. In this way, it is helpful to improve the user's ability to identify disease symptoms based on medical images and obtain diagnostic results based on the disease symptoms, and can provide clear and understandable diagnostic basis and learning cases for medical image diagnosis training and students' autonomous learning.

[0058] In some embodiments, the symptom identification question includes a symptom selection question having multiple symptom description options.

[0059] Here, the symptom description option may include text information for describing the symptom. During the answering process, the user can identify the symptom in the medical image and select at least one symptom description option from the symptom description options of the symptom selection question as the answer information of the symptom selector.

[0060] Before the above step S103, the method further includes the following steps S111 to S112:

[0061] Step S111: Analyze and process the medical image using the trained image analysis model to obtain a plurality of disease symptom description options corresponding to at least one disease symptom contained in the medical image.

[0062] Here, the image analysis model may be a trained neural network model. During implementation, the model structure and training method of the image analysis model may be predetermined by a person skilled in the art according to actual conditions, and the embodiments of the present application do not limit this. For example, the image analysis model may include, but is not limited to, a convolutional neural network model, a vision-based transformer model, a model based on a contrastive learning text-image pre-training (Constrastive Language-Image Pre-training, CLIP) architecture, and the like.

[0063] In some embodiments, the image analysis model is a model based on the CLIP architecture, and the image analysis model may include an image processing network, a text processing network, and an image-text alignment module. The image processing network can be used to extract features of medical images to obtain lesion image features corresponding to at least one symptom in the medical image, and the text processing network can be used to extract features of a large number of medical terms. Finally, the image-text alignment module is used to map the text features corresponding to a large number of medical terms and the lesion image features to the same vector space, and align the text features and the lesion image features in the vector space to obtain multiple text features related to the lesion image features, thereby obtaining multiple symptom descriptions corresponding to at least one symptom contained in the medical image, and then obtaining multiple symptom description options corresponding to at least one symptom contained in the medical image. Among them, the image processing network may include any suitable image encoder, such as a convolutional neural network, a recurrent neural network, a Transformer-based encoder, etc. The text processing network can include any suitable natural language processing model, such as a Bidirectional Encoder Representations from Transformers (BERT) model, a Generative Pre-Trained Transformer (GPT) model, etc.

[0064] In some embodiments, the image analysis model can be obtained by comparative learning and training using a large number of medical images and medical terms used in the medical field to describe symptoms and image diagnosis. For example, the text processing network in the image analysis model can be pre-trained using the commonly used diagnostic terms and descriptive texts in the medical field that have been pre-collected and organized to obtain a pre-trained text processing network; the image processing network in the image analysis model can be pre-trained using a large number of medical image samples to obtain a pre-trained image processing network; the image processing network, text processing network, and image-text alignment module in the image analysis model can be subjected to comparative learning and training using a large number of medical image samples with annotated information to obtain a trained image analysis model, wherein the annotated information of the medical image sample can include a pre-annotated description of at least one symptom corresponding to the medical image sample and a disease type corresponding to the medical image sample.

[0065] Step S112, for each of the symptom, based on a plurality of symptom description options corresponding to the symptom, a symptom multiple-choice question corresponding to the symptom is generated.

[0066] In the above embodiment, the trained image analysis model is used to analyze and process the medical image to obtain multiple symptom description options corresponding to at least one symptom contained in the medical image; for each symptom, based on the multiple symptom description options corresponding to the symptom, a symptom selection question corresponding to the symptom is generated. In this way, at least one symptom identification question corresponding to the medical image can be generated quickly and accurately.

[0067] In some embodiments, the above step S111 may include the following steps S121 to S122:

[0068] Step S121, using the trained image analysis model, analyzing and processing the medical image to obtain a set of disease symptom descriptions corresponding to at least one disease symptom contained in the medical image, and a confidence level of each disease symptom description in the set of disease symptom descriptions.

[0069] Step S122, for each of the symptoms, select multiple symptom descriptions from the symptom description set corresponding to the symptom as symptom description options corresponding to the symptom, and the selected multiple symptom descriptions at least include the symptom description with the highest confidence in the symptom description set.

[0070] In some implementations, multiple symptom descriptions with the highest confidence levels may be selected from a set of symptom descriptions corresponding to the symptom as symptom description options corresponding to the symptom.

[0071] In some embodiments, the symptom description with the highest confidence in a symptom description set corresponding to the symptom can be used as the symptom description option corresponding to the symptom, and at least one symptom description other than the symptom description with the highest confidence in the symptom description set can be randomly selected as the symptom description option corresponding to the symptom.

[0072] In some embodiments, the symptom description option corresponding to the symptom description with the highest confidence among the symptom description options corresponding to the symptom is the correct option, and the remaining symptom description options can be used as interference options.

[0073] In the above embodiment, the trained image analysis model is used to analyze and process the medical image to obtain a set of symptom descriptions corresponding to at least one symptom contained in the medical image, and the confidence of each symptom description in the symptom description set; for each symptom, multiple symptom descriptions are selected from the symptom description set corresponding to the symptom as the symptom description options corresponding to the symptom, and the selected multiple symptom descriptions at least include the symptom description with the highest confidence in the symptom description set. In this way, since the multiple symptom description options corresponding to the symptom at least include the symptom description with the highest confidence output by the image analysis model, the correct option corresponding to the symptom can be obtained quickly and accurately, thereby improving the rationality of each symptom selection question.

[0074] In some embodiments, the answer information corresponding to the medical image includes first answer information of each disease sign identification question corresponding to the medical image, and second answer information of the disease diagnosis question corresponding to the medical image, and the answer results include the first answer result of each disease sign identification question, and the second answer result of the disease diagnosis question.

[0075] Determining the answer result of the user for the medical image based on the first answer information, the second answer information and the answer information corresponding to the medical image in the above step S104 may include the following steps S131 to S132:

[0076] Step S131, for each of the disease symptom identification questions, based on the first answer information and the first answer information of the disease symptom identification question, determine the first answer result of the disease symptom identification question.

[0077] Here, the first answer information of each disease symptom identification question can be set in any appropriate manner according to actual conditions, and the embodiment of the present application is not limited to this.

[0078] In some implementations, the first answer information of the disease symptom identification question corresponding to the medical image may be obtained by manually annotating the medical image in advance.

[0079] In some embodiments, the medical image model can be input into a trained image analysis model to obtain a set of symptom descriptions corresponding to at least one symptom contained in the medical image; for each symptom, based on the symptom description with the highest confidence in the symptom description set corresponding to the symptom, the first answer information of the symptom identification question corresponding to the symptom is determined. For example, the symptom description with the highest confidence can be determined as the first answer information. For another example, the symptom description with the highest confidence can be manually confirmed, and the confirmed symptom description with the highest confidence can be used as the first answer information.

[0080] In some embodiments, when the first answer information and the first answer information of the disease symptom identification question match, the first answer result of the disease symptom identification question can be determined to be a correct answer; when the first answer information and the first answer information of the disease symptom identification question do not match, the first answer result of the disease symptom identification question can be determined to be an incorrect answer. The matching of the first answer information and the first answer information can include that the first answer information and the first answer information match in words and / or match in semantics.

[0081] Step S132: determining a second answer result of the disease diagnosis question based on the second answer information and the second answer information.

[0082] Here, the second answer information of the disease diagnosis question can be set in any appropriate manner according to the actual situation, and the embodiment of the present application is not limited to this.

[0083] In some implementations, the second answer information of the disease diagnosis question corresponding to the medical image may be obtained by manually annotating the medical image in advance.

[0084] In some embodiments, the medical image model can be input into a trained image analysis model to obtain the disease type corresponding to the medical image; based on the disease type, the second answer information of the disease diagnosis question corresponding to the medical image is determined. For example, the disease type can be determined as the second answer information. For another example, the disease type can be manually confirmed, and the confirmed disease type is used as the second answer information.

[0085] In some embodiments, when the second answer information and the second answer information of the disease diagnosis question match, the second answer result of the disease diagnosis question can be determined to be a correct answer; when the second answer information and the second answer information of the disease diagnosis question do not match, the second answer result of the disease diagnosis question can be determined to be an incorrect answer. The matching of the second answer information and the second answer information can include that the second answer information and the second answer information match in words and / or match in semantics.

[0086] In some embodiments, determining the first answer result of the disease symptom identification question based on the first answer information and the first answer information of the disease symptom identification question in the above step S131 may include at least one of the following steps S141 to S143:

[0087] Step S141, when the first answer information and the first answer information of the disease symptom identification question are the same, determining that the first answer result of the disease symptom identification question is correct.

[0088] Step S142, when the semantic similarity between the first answer information and the first answer information of the disease symptom identification question is greater than or equal to a similarity threshold, determining that the first answer result of the disease symptom identification question is a correct answer.

[0089] Step S143, when the first answer information of the disease symptom identification question is different from the first answer information, and the semantic similarity between the first answer information of the disease symptom identification question and the first answer information is less than the similarity threshold, determine that the first answer result of the disease symptom identification question is an incorrect answer.

[0090] Here, any suitable natural language model may be used to determine the semantic similarity between the first reply information and the first answer information, and the embodiments of the present application are not limited to this.

[0091] The similarity threshold can be preset according to actual conditions.

[0092] In the above embodiment, by comparing whether the first answer information of the disease symptom identification question is the same as the first answer information, and / or whether the semantic similarity between the first answer information of the disease symptom identification question and the first answer information is greater than or equal to the similarity threshold, it is determined whether the first answer result of the disease symptom identification question is correct, and the first answer result can be judged more accurately.

[0093] It should be noted that the method of determining the second answer result of the disease diagnosis question based on the second answer information and the second answer information in the above step S132 can be the same as the method of determining the first answer result of the symptom identification question based on the first answer information and the first answer information of the symptom identification question in the above step S131. When implementing it, reference can be made to at least one of the above steps S141 to S143.

[0094] In some embodiments, each of the disease symptom identification questions is used to assess the ability to identify a disease symptom contained in the medical image, and the disease diagnosis questions are used to assess the ability to diagnose the type of disease corresponding to the medical image.

[0095] Before the above step S102, the method may further include the following steps S151 to S152:

[0096] Step S151, determining the learning ability parameter of the user.

[0097] Here, the learning ability parameters may include but are not limited to the accuracy rate of disease identification questions corresponding to each disease symptom, the degree of mastery of the identification ability of each disease symptom, the learning efficiency of the identification ability of each disease symptom, the expected upper limit of learning efficiency corresponding to each disease symptom, the accuracy rate of disease diagnosis questions corresponding to each disease type, the degree of mastery of the diagnostic ability of each disease type, the learning efficiency of the diagnostic ability of each disease type, the expected upper limit of learning efficiency corresponding to each disease type, etc.

[0098] The accuracy rate of the symptom identification questions corresponding to the symptom refers to the accuracy rate of the user's answers to the symptom identification questions corresponding to the symptom in the historical question-answering process.

[0099] The accuracy rate of the disease diagnosis questions corresponding to a disease type refers to the accuracy rate of the user's answers to the disease diagnosis questions corresponding to the disease type in the historical question-answering process.

[0100] The mastery of the ability to identify symptoms can be determined based on the accuracy of the symptom identification questions corresponding to the symptom. For example, the mastery of the ability to identify symptoms can include poor, good, excellent, or precise, etc.; when the accuracy of the symptom identification questions corresponding to the symptom is less than the first accuracy threshold, the mastery of the ability to identify the symptom is determined to be poor; when the accuracy of the symptom identification questions corresponding to the symptom is less than the second accuracy threshold and greater than or equal to the first accuracy threshold, the mastery of the ability to identify the symptom is determined to be good; when the accuracy of the symptom identification questions corresponding to the symptom is less than the third accuracy threshold and greater than or equal to the second accuracy threshold, the mastery of the ability to identify the symptom is determined to be excellent; when the accuracy of the symptom identification questions corresponding to the symptom is greater than or equal to the third accuracy threshold, the mastery of the ability to identify the symptom is determined to be precise; wherein the first accuracy threshold is lower than the second accuracy threshold, and the second accuracy threshold is lower than the third accuracy threshold. The first accuracy threshold, the second accuracy threshold, and the third accuracy threshold can all be set according to actual conditions.

[0101] The learning efficiency of the ability to identify a disease symptom refers to the efficiency of the user's ability to identify the disease symptom during the historical question-answering process. For example, the faster the user's correct answer rate for the disease symptom identification question corresponding to the disease symptom during the historical question-answering process improves, the higher the learning efficiency of the user's ability to identify the disease symptom.

[0102] In some embodiments, a machine learning algorithm can be used to accurately evaluate the user's learning ability based on the accuracy of the disease identification questions corresponding to each disease symptom, and obtain the upper limit of the learning efficiency that can be achieved for each disease symptom. For example, the machine learning algorithm used may include but is not limited to support vector machines and / or random forests.

[0103] In some implementations, the user's learning ability parameter may be obtained by collecting statistics on the user's historical answer data.

[0104] Step S152: based on the learning ability parameter, obtaining the medical image and at least one disease sign identification question and disease diagnosis question corresponding to the medical image from an examination question bank.

[0105] Here, according to the user's learning ability parameters, medical images matching the learning ability parameters and disease symptom identification questions and disease diagnosis questions corresponding to the medical images can be obtained from the assessment question bank.

[0106] In some embodiments, any suitable task recommendation algorithm may be used to recommend a medical image to be analyzed and at least one disease sign identification question and disease diagnosis question corresponding to the medical image from the assessment question bank based on the user's learning ability parameter. For example, the task recommendation algorithm may include but is not limited to a gradient boosting algorithm and / or reinforcement learning.

[0107] In the above embodiment, based on the user's learning ability parameters, medical images and at least one disease sign identification question and disease diagnosis question corresponding to the medical images are obtained from the assessment question bank. In this way, the user can be provided with medical images to be analyzed that are more matched with the current learning ability and disease sign identification questions and disease diagnosis questions corresponding to the medical images, which is more conducive to the user's learning of medical image analysis capabilities.

[0108] In some embodiments, the above step S152 may include the following steps S161 to S162:

[0109] Step S161, determining the target assessment difficulty based on the learning ability parameter.

[0110] Step S162: based on the target assessment difficulty, obtaining the medical image and at least one disease sign identification question and disease diagnosis question corresponding to the medical image from the assessment question bank.

[0111] Here, the target assessment difficulty may be an assessment difficulty that matches the user's learning ability.

[0112] In some implementations, a learning ability model of the user may be constructed based on the learning ability parameters of the user, and the model parameters of the learning ability model may include the target assessment difficulty. For example, the current target assessment difficulty of the user may be dynamically adjusted based on the learning ability parameters of the user, so that the current target assessment difficulty is adapted to the current learning ability of the user.

[0113] In some embodiments, the target assessment difficulty may include a target difficulty level, and the assessment question bank may include multiple medical images with different difficulty levels and at least one disease sign identification question and disease diagnosis question corresponding to each medical image. Based on the target assessment difficulty, a medical image with a difficulty level consistent with the target assessment difficulty may be selected from the assessment question bank as the current medical image to be analyzed, and at least one disease sign identification question and disease diagnosis question corresponding to the medical image may be obtained.

[0114] In implementation, the difficulty level corresponding to the medical image can be defined by those skilled in the art according to actual conditions, and the embodiments of the present application do not limit this. For example, the difficulty level can be defined by the complexity of the medical image, the coverage of the symptom knowledge points corresponding to the medical image, and / or the prevalence of the disease category corresponding to the medical image.

[0115] In the above embodiment, the target assessment difficulty is determined based on the learning ability parameter, and based on the target assessment difficulty, a medical image and at least one disease sign identification question and disease diagnosis question corresponding to the medical image are obtained from the assessment question bank. In this way, medical images and corresponding disease sign identification questions and disease diagnosis questions with a difficulty that matches the user's current learning ability can be obtained, which is conducive to improving the user's learning enthusiasm and learning effect.

[0116] In some embodiments, the above step S162 may include the following steps S171 to S172:

[0117] Step S171, determining the prevalence of a target disease that matches the target assessment difficulty.

[0118] Step S172, based on the prevalence of the target disease, obtaining the medical image and at least one disease symptom identification question and disease diagnosis question corresponding to the medical image from the assessment question bank.

[0119] Here, each medical image in the test question bank has a corresponding disease prevalence, and the disease prevalence represents the prevalence of the disease type corresponding to the medical image. Based on the target disease prevalence, a medical image whose disease prevalence matches the target disease prevalence can be obtained from the test question bank, and the disease symptom identification questions and disease diagnosis questions corresponding to the medical image can be obtained.

[0120] In some embodiments, a matching relationship between at least one disease process degree and an assessment difficulty can be predetermined; based on the matching relationship, a target disease prevalence that matches the target assessment difficulty can be determined. It is understandable that the lower the disease prevalence corresponding to the medical image, the higher the assessment difficulty corresponding to the medical image. Therefore, the lower the disease prevalence in the matching relationship, the higher the assessment difficulty that matches the disease process degree.

[0121] In the above embodiment, the prevalence of the target disease that matches the target assessment difficulty is determined; and based on the prevalence of the target disease, the medical image and the disease sign identification questions and disease diagnosis questions corresponding to the medical image are obtained from the assessment question bank. In this way, in the process of selecting medical images, considering the correlation between the assessment difficulty corresponding to the medical image and the prevalence of the disease, the difficulty of the selected medical image can be made more accurate, so that the disease sign identification questions and disease diagnosis questions corresponding to the medical image are more adapted to the user's current learning ability.

[0122] In some embodiments, the above step S152 may include the following steps S181 to S182:

[0123] Step S181, based on the learning ability parameter, obtaining a candidate medical image sequence and at least one disease symptom recognition question and a disease diagnosis question corresponding to each candidate medical image in the candidate medical image sequence from a test question bank.

[0124] Step S182, taking each medical image in the candidate medical image sequence as the medical image to be analyzed in turn, and obtaining at least one disease symptom identification question and disease diagnosis question corresponding to the medical image.

[0125] Here, the candidate medical image sequence may include multiple medical images to be analyzed in a specific order. Each medical image in the candidate medical image sequence may be sequentially used as a medical image to be analyzed, and the medical image may be displayed in the image display area, and at least one disease sign identification question and disease diagnosis question corresponding to the medical image may be displayed in the question answering area; in response to the user's answering operation in the question answering area, after obtaining the first answer information of each disease sign identification question corresponding to the medical image and the second answer information of the disease diagnosis question corresponding to the medical image, the next medical image may be displayed in the image display area, and at least one disease sign identification question and disease diagnosis question corresponding to the next medical image may be displayed in the question answering area for the user to answer.

[0126] In some embodiments, based on the user's learning ability parameters, a set number of medical image sets and at least one disease sign identification question and disease diagnosis question corresponding to each medical image in the medical image set can be selected from the examination question bank, and the examination order corresponding to each medical image in the medical image set can be determined; based on the examination order, each medical image in the medical image set can be sorted to obtain a candidate medical image sequence.

[0127] In some embodiments, the assessment order corresponding to each medical image in the medical image set may be determined according to the difficulty level corresponding to the medical image. For example, the lower the difficulty level corresponding to the medical image, the higher the assessment order corresponding to the medical image. In this way, the user can analyze the medical images with lower difficulty first, improve the user's enthusiasm for answering in the early stage, and thus can promote the user to actively complete the subsequent answers to the disease symptom identification questions and disease diagnosis questions corresponding to the medical images.

[0128] In the above embodiment, based on the user's learning ability parameter, a candidate medical image sequence and at least one disease sign identification question and disease diagnosis question corresponding to each candidate medical image in the candidate medical image sequence are obtained from the test question bank; each medical image in the candidate medical image sequence is sequentially used as a medical image to be analyzed, and at least one disease sign identification question and disease diagnosis question corresponding to the medical image is obtained. In this way, appropriate medical images can be recommended to the user in order for analysis according to the user's learning ability.

[0129] In some embodiments, the above method may further include the following steps S191 to S192:

[0130] Step S191, updating the learning ability parameter of the user based on the answer result of the user to the medical image.

[0131] Step S192, based on the updated learning ability parameter, obtain the next medical image and at least one disease sign recognition question and disease diagnosis question corresponding to the next medical image from the examination question bank.

[0132] In this way, the user's learning ability parameters can be continuously updated according to the user's answers to medical images, and then based on the updated learning ability parameters, the next medical image that better matches the user's current learning ability and the symptom identification questions and disease diagnosis questions corresponding to the next medical image can be obtained from the assessment question bank, thereby continuously improving the user's learning effect on medical image analysis.

[0133] The following takes the medical imaging diagnosis autonomous learning scenario of medical students as an example to illustrate the application of the interactive method provided in the embodiment of the present application in an actual scenario.

[0134] In related technologies, autonomous learning systems for medical imaging diagnosis usually rely on a large amount of annotated medical imaging data. However, due to the complexity and diversity of medical imaging data, as well as the scarcity of annotated data, these systems often fail to achieve ideal performance. In addition, when faced with new and unseen cases, these systems often lack sufficient adaptability, making it difficult for students to conduct effective learning and diagnostic analysis.

[0135] In response to the needs of medical image diagnosis training and student autonomous learning, the embodiment of the present application provides a medical image diagnosis autonomous learning system based on an interpretable model. By introducing an interpretable model, the system not only improves the interpretability of the diagnostic results, but also makes significant progress in autonomous learning. The system provides clear and understandable diagnostic basis and cases for medical image diagnosis training and student autonomous learning. With the help of real medical image data and corresponding diagnostic results, the system can obtain diagnostic basis through a trained image analysis model, allowing students to understand and identify key features, thereby helping students to deeply understand the principles and methods of medical image diagnosis. The medical image diagnosis autonomous learning system provided in the embodiment of the present application has at least the following improvements:

[0136] 1) Model interpretability assists autonomous learning: Since the image analysis model used in the medical image diagnosis autonomous learning system provided in the embodiment of the present application is highly interpretable, the system can clearly display the key features extracted from the medical images (corresponding to the symptoms in the aforementioned embodiment) and the relationship between these key features and the diagnosis results (corresponding to the disease type in the aforementioned embodiment). This enables users (such as students, etc.) to better understand the basis for diagnosis, so that when faced with new and unseen cases, they can more accurately identify key features and conduct effective autonomous learning.

[0137] 2) Dynamic feedback mechanism: The medical imaging diagnosis autonomous learning system provided in the embodiment of the present application also introduces a dynamic feedback mechanism, which can monitor the user's diagnostic ability in real time and dynamically adjust the learning strategy and parameters according to the diagnostic ability feedback. When the system recognizes that the user performs poorly on a specific type of case, it can automatically re-recommend these cases to the user to improve the user's diagnostic ability.

[0138] 3) Continuous learning and updating: As medical research deepens and new medical imaging data continues to be generated, the system can continuously learn new knowledge from this data and update its own imaging analysis model. This enables the system to always keep pace with the latest advances in the medical field and improve the accuracy and reliability of diagnosis.

[0139] In addition, the system itself has an autonomous learning function, which can adjust the learning content and difficulty according to the students' learning progress and feedback, and provide personalized learning paths and feedback. After the system learns the relevant parameters used to characterize a user's autonomous learning ability, through continuous learning and optimization, the system can continuously adapt to the students' learning needs, gradually understand and recommend the most suitable learning materials to users, and improve learning effects and efficiency.

[0140] In some embodiments, an image analysis model based on the CLIP architecture can be used to perform actual medical image analysis. Each input medical image has a pre-labeled disease category label. The image analysis model is based on the CLIP architecture and is pre-trained by comparative learning using a large number of medical images and common terms used in the medical field to describe symptoms and image diagnosis.

[0141] In some embodiments, the medical image to be analyzed, as well as at least one symptom multiple-choice question and a disease diagnosis question corresponding to the medical image, can be displayed in the user interface. The user can complete the answers to each symptom multiple-choice question through analysis, and use the answers selected as the basis for diagnosis, and input the final disease diagnosis result (corresponding to the disease type in the aforementioned embodiment). The disease diagnosis result input by the user is compared with the disease category label corresponding to the medical image to obtain the accuracy of the user's analysis of the medical image. Each symptom multiple-choice question corresponds to a symptom, and the options in the symptom multiple-choice question are multiple candidate words corresponding to the symptom (corresponding to the symptom description in the aforementioned embodiment). Each option is generated by a trained image analysis model based on the input medical image, and the candidate word with the highest confidence is the correct option.

[0142] In some embodiments, the system has an intelligent autonomous learning function, which can adjust the learning content and difficulty according to the students' mastery of disease symptom knowledge points (i.e., the ability to identify disease symptoms), provide personalized learning paths and feedback, so as to optimize learning effects and improve learning efficiency; it can customize learning tasks according to the user's learning ability, and after the system learns the relevant parameters used to characterize a student's autonomous learning ability, it can gradually understand and recommend students to obtain the most appropriate learning materials (including medical images and disease symptom selection questions and disease diagnosis questions corresponding to the medical images). Among them, the learning difficulty is inversely correlated with the prevalence of the disease.

[0143] Figure 3 This is a schematic diagram of the implementation architecture of a medical imaging diagnosis autonomous learning system provided in an embodiment of the present application. Figure 3 As shown, the medical imaging diagnosis autonomous learning system includes: a data input and preprocessing module 210, a stable accuracy training module 220, a learning task difficulty matching module 230, a learning ability upper limit calculation and task recommendation module 240, an output and feedback module 250, and a system optimization and iteration module 260.

[0144] The data input and preprocessing module 210 can implement at least one of the following functions:

[0145] 1) Medical imaging data input

[0146] Data source selection: Various types of medical imaging data are collected from different hospitals and medical institutions, including X-ray images, CT images, MRI images, etc.

[0147] Data cleaning: Preprocess the collected medical imaging data to remove noise and incomplete data and ensure the quality of input data.

[0148] Data format conversion: Convert medical imaging data in various formats into a standard format suitable for image analysis model processing, such as DICOM format.

[0149] 2) Deep learning convolutional neural network processing

[0150] Model architecture selection: The image analysis model uses pre-trained deep learning convolutional neural networks (such as ResNet, VGG, etc.) to extract features from medical images.

[0151] Feature extraction: Medical images are processed through convolutional layers and pooling layers to extract key features of medical images, such as lesion areas and edge features.

[0152] Model training: Use large-scale labeled datasets to train the deep learning convolutional neural network in the image analysis model to improve its performance on different medical images. The training process includes techniques such as data augmentation, batch normalization, and early stopping.

[0153] In some embodiments, in addition to using a convolutional neural network to extract features from medical images, the image analysis model may also use a recurrent neural network (RNN) or a generative adversarial network (GAN) to extract features from medical images.

[0154] 3) Text alignment pre-training

[0155] Medical Terminology Collection: Collect and organize diagnostic terms and descriptive texts commonly used in the medical field.

[0156] Pre-trained models: Natural language processing models (such as BERT, GPT, etc.) in image analysis models are pre-trained using medical terminology to enable them to understand and generate medical text.

[0157] Alignment process: Align image features with corresponding medical terms to establish a mapping relationship between input images and output text. First, CLIP technology is used, which combines deep convolutional neural networks (CNN) and natural language processing models (such as BERT or GPT) to simultaneously encode medical images and description texts and extract their features. CLIP maps image features and text features into the same vector space through a shared embedding space. In this space, related image features and text features are close together, while irrelevant image features and text features are pushed apart. During the training process, the model is optimized through the contrastive loss function to minimize the distance between related image features and text features and maximize the distance between irrelevant image features and text features. CLIP uses a large amount of paired medical images and corresponding description text data for pre-training to ensure that the image analysis model can effectively align image features and text features. The validation set is used to evaluate the alignment performance of the model and improve the accuracy and stability of the model through parameter tuning. Finally, the trained image analysis model is applied to perform actual medical image analysis tasks. The system displays the medical images to be analyzed and the corresponding generated descriptive text through the user interface, assisting medical professionals in diagnosis and decision-making, and achieving efficient alignment between input images and output text.

[0158] In some embodiments, in addition to text alignment, other modal data (such as sound, video, etc.) may also be considered to be included in the image analysis model training process to improve the generalization ability and robustness of the model.

[0159] The training module 220 for stable accuracy can enable the user to achieve a stable accuracy at a specific learning task difficulty. By using deep learning convolutional neural network and text alignment pre-training technology. By processing medical image data, the image analysis model can learn the alignment relationship between the input medical image and the output symptom description text and disease type.

[0160] The user can observe the medical image to be analyzed through the user interface and select the corresponding text from the given symptom description options to describe the symptoms in the medical image. The system calculates the accuracy between the user's selection and the actual description text, thereby reflecting the user's mastery of relevant medical knowledge points.

[0161] The training stable accuracy module 220 can implement at least one of the following functions:

[0162] 1) User interface image display

[0163] In terms of interface design, an intuitive user interface (corresponding to the first interactive interface in the aforementioned embodiment) is first designed to facilitate users to browse and select medical images. The interface design focuses on user experience and adopts a simple and clear layout so that users can easily find the required functions. The interface includes image zoom, pan and rotate functions, and users can view the image in detail through mouse wheel, drag and rotation gestures. These functions ensure that users can clearly view image details and identify important features.

[0164] In terms of image display, the pre-processed medical images are displayed through the interface for users to perform further interactive operations. Users can select and mark images on the interface, and the system will update and feedback the user's operation results in real time. In addition, the medical images displayed in the user interface can include multiple associated sub-images, such as multiple sub-images of the same lesion taken from different angles. The user interface also supports the comparison function of the multiple sub-images. Users can view and compare the multiple sub-images at the same time, conduct comprehensive analysis, and improve the accuracy of diagnosis. The overall interface design not only improves the user's operating efficiency, but also provides rich interactive functions to ensure that users can complete various answering tasks conveniently and quickly during use.

[0165] 2) User text selection interaction

[0166] Option provision: Multiple symptom description options are provided for each medical image, allowing users to select the most appropriate symptom description. These symptom description options can be output by the image analysis model or selected from a common medical term library.

[0167] Interaction records: Record each user's selection and interaction behavior for subsequent analysis and evaluation. Interaction data includes the text selected by the user (such as symptom description, disease type, etc.), selection time, and selection frequency.

[0168] 3) Accuracy calculation

[0169] In terms of the comparison algorithm, the comparison algorithm is first used to compare the text selected by the user with the real corresponding text (corresponding to the answer information in the aforementioned embodiment). This process includes two main methods: exact matching and semantic similarity calculation. Exact matching refers to directly comparing the text selected by the user with the standard answer to determine whether the two are completely consistent. This method is fast and simple and is suitable for situations where the terminology is standardized and clear. However, medical text descriptions often have diversity, and relying solely on exact matching may not fully reflect the user's understanding. Therefore, semantic similarity calculation becomes a necessary supplement. Through natural language processing technology (such as BERT or Word2Vec), the text is converted into a vector representation, and the semantic similarity between the user's selected text and the standard answer is calculated. This method can identify semantically similar descriptions, even if they differ in expression. Next is the accuracy evaluation, and the system calculates the accuracy of the user's selection based on the comparison results. The calculation of the accuracy not only considers exact matching, but also combines semantic similarity scoring to form a comprehensive evaluation index. In order to help users understand their learning progress, the system provides real-time feedback. The feedback form includes text prompts and graphical representations. The text prompts directly tell the user the accuracy of the selection and provide improvement suggestions (including the interpretation of knowledge points, etc.). Graphical representation uses progress bars, pie charts, or line graphs to intuitively display the user's changes in accuracy and knowledge points. This real-time feedback mechanism can not only enhance the user's learning experience, but also help users find and correct errors in a timely manner, gradually improving learning results. Overall, by combining the comparison algorithm of exact matching and semantic similarity calculation, as well as a variety of feedback forms, the system can accurately and comprehensively evaluate the user's learning performance and help them effectively master medical knowledge.

[0170] In some implementations, after the user completes the answer operation corresponding to the medical image, the user's answer result for the medical image may be displayed. Figure 4 A schematic diagram of a display interface for answering questions provided in an embodiment of the present application, such as Figure 4 As shown, the display interface 30 of the answer results may include the first answer result 31 of each disease symptom identification question, the second answer result 32 of the disease diagnosis question, and a learning progress viewing control 33 corresponding to each disease symptom identification question, wherein the learning progress viewing control 33 corresponding to the disease symptom identification question is used to view the learning progress of the disease symptom knowledge point corresponding to the disease symptom identification question, that is, the degree of mastery of the ability to identify the disease symptom.

[0171] In some implementations, in response to a user clicking on a learning progress viewing control 33 corresponding to a disease symptom identification question, a learning progress viewing interface for the disease symptom knowledge point corresponding to the disease symptom identification question may be displayed. Figure 5 A schematic diagram of a learning progress viewing interface provided for the implementation of this application, such as Figure 5As shown, the learning progress viewing interface 40 of a symptom knowledge point can display the correct rate 41 of the user's answer to the symptom identification question corresponding to the symptom knowledge point, as well as the user's learning progress 42 of the symptom knowledge point.

[0172] 4) User mastery assessment

[0173] Mastery model: Based on the user's accuracy data, a user's knowledge mastery model is established. The model takes into account the user's historical choices, accuracy, and learning time.

[0174] Evaluation Report: Generate a detailed evaluation report to show the user's mastery of each knowledge point. The report includes charts and text analysis.

[0175] The learning task difficulty matching module 230 can further construct a user model on the basis of a stable accuracy rate (stable accuracy rate means that the accuracy rate remains within a certain fluctuation range) to match it with the difficulty of the learning task (corresponding to the assessment difficulty in the aforementioned embodiment). According to the user's accuracy rate in the first step, the system will search for a set of medical images that match the corresponding text (including symptom descriptions and / or disease types) selected by the user as the medical image material subsequently displayed to the user for analysis. These medical images themselves cover a number of different symptom knowledge points. Through this process, accurate information about the user's mastery of each knowledge point can be obtained. By gradually adjusting the medical images displayed to the user, the system will help the user improve the accuracy rate on each symptom knowledge point, thereby establishing a stable mapping relationship between the medical image and the correct symptom description and disease type in the user's mind.

[0176] The learning task difficulty matching module 230 can implement at least one of the following functions:

[0177] 1) User model construction

[0178] Personalized model: Build a personalized user model based on the user's learning behavior and accuracy data. The user model uses collaborative filtering and clustering algorithms to analyze user data.

[0179] Dynamic adjustment: Dynamically adjust the model based on the user's learning progress to ensure that the medical images displayed to the user for analysis are of moderate difficulty. Adjustment parameters include learning rate and task difficulty. For example, the smaller the learning rate corresponding to a disease symptom knowledge point, the more medical images corresponding to the disease symptom knowledge point are recommended to the user.

[0180] In some embodiments, a rule-based matching method and / or an adaptive learning algorithm may be used to match the difficulty of learning tasks. For example, a series of rules may be formulated to determine the difficulty of learning tasks suitable for the user based on the user's historical learning records or performance. In another example, an adaptive learning algorithm may be used to dynamically adjust the difficulty of learning tasks based on the user's real-time feedback during the learning process. This method may more accurately match the user's actual ability level.

[0181] 2) Analysis of knowledge mastery

[0182] In terms of fine-grained analysis, we first conduct a comprehensive and detailed analysis of the user's accuracy data to determine their level of mastery of each knowledge point.

[0183] For example, the analysis method can include a variety of advanced statistical and machine learning techniques, mainly using decision trees and Bayesian networks. Decision trees construct a series of rules to gradually segment the data and intuitively reveal the user's performance on different knowledge points (i.e., symptom knowledge points). Each node represents a decision point, through which you can clearly see the user's accuracy on a specific knowledge point and which knowledge points have weak links. The Bayesian network uses a probabilistic graphical model to express the conditional dependency between various knowledge points. By learning the user's historical data, the Bayesian network can infer the user's mastery probability of each knowledge point and quantify the user's learning situation. In addition, methods such as cross-validation can be used to evaluate the accuracy and reliability of the model to ensure the credibility of the analysis results.

[0184] In the knowledge point evaluation part, through the above-mentioned fine-grained analysis, a detailed knowledge point evaluation report is generated to help users fully understand their learning situation. The report is rich and detailed, mainly including the following aspects: First, the percentage of mastery, which quantitatively displays the user's correct rate on each knowledge point. For example, a user's mastery of the knowledge point "stroke diagnosis" is 85%, while the mastery of "Parkinson's disease diagnosis" is only 60%. This percentage display allows users to intuitively see their learning achievements and shortcomings. Secondly, there are weakness prompts, which provide specific improvement suggestions for knowledge points with low user mastery. For example, if a user performs poorly on "Parkinson's disease diagnosis", the system will prompt the user to focus on reviewing related content and recommend appropriate learning resources, such as textbook chapters, online courses or exercises. In addition, the report also includes a learning progress curve, which shows the user's learning progress over a period of time and the changing trend of the mastery of each knowledge point.

[0185] 3) Image material search

[0186] Material library construction: Establish an image material library covering different knowledge points (corresponding to the test question library in the above embodiment) to ensure the diversity and coverage of the materials. The material library is regularly updated and contains the latest medical images.

[0187] Intelligent search: Intelligently search for suitable medical imaging materials based on the user's mastery level and learning needs. Search algorithms include vector search and semantic retrieval.

[0188] 4) Adaptation difficulty image display

[0189] Difficulty matching: According to the user's knowledge, medical images of appropriate difficulty are selected for display. The difficulty level is defined by the complexity of the medical image and the coverage of knowledge points.

[0190] Progressive learning: Help users gradually improve their knowledge level by gradually increasing the difficulty of the medical images displayed. The difficulty of each medical image displayed is adjusted based on the user's learning curve, number of learning times and / or accuracy curve.

[0191] After successfully matching the difficulty of the learning task, the system can use the learning ability upper limit calculation and task recommendation module 240 to independently calculate the user's learning ability upper limit, and give learning task recommendations under maximum learning efficiency. Based on the user accuracy data accumulated in the first two steps, the model can accurately evaluate the user's mastery of each knowledge point. When the user learns different images, the model will record and analyze the accuracy data of multiple knowledge points judged by the user, and draw the corresponding accuracy curve. Through this curve, the model can intelligently adjust the image content displayed to the user later to ensure the user's continuous progress in the learning process. At the same time, the model will also give the most suitable learning task recommendations for the current stage based on the user's learning ability (such as learning efficiency, the highest level achievable) and learning progress (such as poor, good, excellent, and fine progress status) to maximize the user's learning efficiency.

[0192] The learning ability upper limit estimation and task recommendation module 240 can implement at least one of the following functions:

[0193] 1) Recording of accuracy data of multiple knowledge points

[0194] The recording and analysis of the accuracy data of multiple knowledge points is an important part of evaluating the learning effect of users. First of all, in terms of data recording, the system will record the accuracy data of users on different knowledge points in detail to form a comprehensive data set.

[0195] For example, the recorded data may include the following: timestamp, which is used to mark the specific time of each user interaction to help analyze learning progress and time distribution; image identification (ID), which is used to uniquely identify each medical image viewed by the user to ensure the accuracy and traceability of the data; user selection, which records the text description selected by the user in each interaction for subsequent comparison and analysis; correctness, which calculates and records the correctness of each user's selection as an important indicator for evaluating their knowledge mastery. In addition, the number of user interactions, dwell time, and operation path are recorded to provide a more comprehensive understanding of the user's learning behavior.

[0196] In terms of data analysis, the system conducts in-depth analysis of the recorded data to discover the learning patterns and bottlenecks of users. First, the data is preprocessed and cleaned through statistical analysis tools (such as Python's Pandas and R language's dplyr and other libraries) to ensure data quality. Then, a variety of analysis methods are used to conduct a detailed discussion of the data. For example, time series analysis can reveal the changes in users' learning effects in different time periods, helping to understand users' learning progress and peak periods; frequency analysis can count the number of interactions and accuracy of users on different knowledge points, and identify areas where they have better and worse mastery; cluster analysis can divide users into different groups based on their learning behavior and provide personalized learning suggestions. In addition, the system can also use machine learning algorithms to perform more advanced analysis of data. Decision tree analysis identifies the key factors that affect users' accuracy by constructing decision paths; Bayesian networks reveal the interdependencies between different knowledge points and the probability of users mastering each knowledge point through probabilistic graph models.

[0197] Based on the comprehensive analysis results, the system generates a detailed learning report, including learning progress curve, knowledge point mastery percentage, and weakness prompts. The learning progress curve shows the changing trend of the user's learning effect over a period of time, helping them to intuitively understand their progress; the knowledge point mastery percentage quantitatively shows the user's correct rate on each knowledge point, allowing them to clearly see their learning results and shortcomings; the weakness prompts provide specific improvement suggestions and recommended resources for the knowledge points that the user has a low mastery of, helping them to review and improve in a targeted manner.

[0198] 2) Accuracy curve drawing

[0199] Curve drawing: Draw the user's accuracy curve for each knowledge point to intuitively show their learning progress. The curve chart includes the time axis and the change in accuracy.

[0200] Data Visualization: Use data visualization techniques to generate easy-to-understand charts and reports. Visualization tools may include but are not limited to Matplotlib and / or D3.js.

[0201] 3) Calculation of the upper limit of learning ability

[0202] Ability assessment model: Use machine learning algorithms to accurately assess the user's learning ability. The machine learning algorithms used may include but are not limited to support vector machines and / or random forests.

[0203] Upper limit calculation: Calculate the upper limit of the user's ability at the current learning stage and determine the highest level they can reach. The calculation results are used to adjust the difficulty of subsequent learning tasks.

[0204] 4) Intelligent image adjustment

[0205] Dynamic adjustment: Intelligently adjust the medical images displayed for analysis based on the user's learning ability and progress. Adjustment parameters include the disease type, difficulty, and knowledge points covered by the medical image.

[0206] Personalized recommendations: Provide personalized medical image recommendations to ensure that the user's learning content matches their abilities. The recommendation algorithm uses deep learning and collaborative filtering technology.

[0207] 5) Learning task recommendation algorithm

[0208] Algorithm design: Design an efficient learning task recommendation algorithm to recommend the most suitable learning task based on the user's learning ability and learning progress. The learning task may include specific learning objectives and the display order of each medical image in the learning objective. The learning objective includes at least one medical image to be analyzed and studied, and the symptom identification questions and disease diagnosis questions corresponding to each medical image. The learning task recommendation algorithm may include but is not limited to gradient boosting and / or reinforcement learning.

[0209] Task recommendation: Based on the algorithm's recommendation results, provide users with scientific and reasonable learning task suggestions. Recommendations include the display order and difficulty adjustment of medical images.

[0210] In some implementations, the recommendation of learning tasks can be implemented based on a recommendation algorithm based on machine learning. In addition to estimating learning ability based on the user's accuracy, other machine learning algorithms (such as collaborative filtering, content filtering, etc.) can also be combined to recommend learning tasks that are more suitable for the user. These algorithms can comprehensively consider factors such as the user's historical learning records, interest preferences, etc. to give more accurate recommendations.

[0211] In some implementations, the recommendation of learning tasks can be implemented based on expert system recommendations. The learning tasks are recommended in combination with the knowledge and experience of domain experts. The expert system can give more targeted and personalized recommendations based on the actual situation and needs of the user.

[0212] The output and feedback module 250 can implement at least one of the following functions:

[0213] 1) Personalized learning task recommendation

[0214] First, based on the user's learning ability and progress, the system generates personalized learning tasks. This process combines the user's historical learning data and real-time performance to ensure that the recommended tasks can maximize the learning effect. The task content includes specific learning goals and the order of presentation of medical images. Specifically, the learning goals are set based on the user's mastery of different knowledge points, aiming to help users gradually improve their understanding and memory of each knowledge point. For example, if the user performs poorly in "stroke diagnosis", the system will set relevant learning goals, such as "identifying typical symptoms of stroke" and "understanding treatment options for stroke." These goals are clear and specific, allowing users to study and review in a targeted manner.

[0215] In terms of the display order of medical images, the system will intelligently arrange the display order of medical images according to the user's learning progress and knowledge mastery. First, the system will give priority to displaying images that the user is more familiar with or less difficult to help them build confidence and consolidate existing knowledge. Subsequently, the difficulty will be gradually increased to display images with more details and complex features to challenge the user's analysis and judgment ability. In this way, through step-by-step learning, users can gradually improve their understanding and diagnosis of complex medical images. In addition, the system will regularly review and display images that the user has learned before to help them consolidate their memory and prevent knowledge from being forgotten. The recommendation system is a key component in the task generation process.

[0216] Through the recommendation system, users are provided with the most suitable learning tasks for the current stage. The recommendation system makes recommendations based on real-time and historical data to ensure the personalization and accuracy of tasks. First, the system monitors the user's learning progress in real time, recording data such as the accuracy of each interaction, operation time, and selection path. These real-time data reflect the user's current learning status and provide the latest source of information for task recommendations. Secondly, the system comprehensively analyzes the user's historical data, including their long-term performance on different knowledge points, learning frequency and habits. Through these historical data, the system can understand the user's learning patterns and preferences and provide more targeted task recommendations.

[0217] The recommendation system uses a variety of machine learning algorithms to recommend learning tasks, such as collaborative filtering and deep learning. Collaborative filtering algorithms predict the learning tasks that users may be interested in and need by analyzing the learning behaviors of other similar users. Deep learning algorithms build complex neural network models to deeply mine the potential patterns and rules in user data and provide more accurate task recommendations. The recommendation system will also continuously learn and optimize itself, adjust the recommendation strategy based on user feedback and learning effects, and ensure dynamic optimization and continuous improvement of task recommendations.

[0218] Recommending suitable learning tasks based on the user's learning ability and progress can enable the user to make continuous progress in the learning process and improve learning efficiency.

[0219] 2) Learning progress tracking

[0220] Progress tracking: real-time tracking of user's learning progress and recording of their learning status. The data includes the tasks completed by the user, accuracy rate and learning time.

[0221] Progress Report: Generate detailed learning progress reports to help users understand their learning outcomes. The reports include charts and text descriptions to show the user's progress and deficiencies.

[0222] 3) User feedback collection

[0223] Feedback mechanism: Establish a user feedback mechanism to collect users’ opinions on learning tasks and systems. Feedback methods include questionnaires and scoring systems.

[0224] Feedback indicators: collect indicators such as user learning satisfaction, task difficulty perception, and system response speed.

[0225] Feedback analysis: Analyze user feedback to find out the shortcomings of the system and the direction of improvement. Analysis methods include text analysis and statistical analysis.

[0226] Feedback implementation: Based on the analysis results, make corresponding improvements to the system to enhance user experience.

[0227] The system optimization and iteration module 260 can implement at least one of the following functions:

[0228] 1) Model performance evaluation

[0229] Performance evaluation: Regularly evaluate the performance of the model to ensure its performance on different tasks. Evaluation methods may include but are not limited to A / B testing and / or cross-validation.

[0230] Metrics monitoring: Monitor the various metrics of the model to identify and resolve potential issues. Monitoring metrics may include, but are not limited to, accuracy, recall, and / or F1 score.

[0231] 2) Algorithm optimization

[0232] Optimize algorithms: Continuously optimize recommendation algorithms to improve their accuracy and efficiency. Optimization methods include parameter tuning and algorithm upgrades.

[0233] The medical imaging diagnosis autonomous learning system provided in the embodiment of the present application can continuously adapt to the latest progress and changes in the medical field through continuous learning and optimization, and improve the accuracy and reliability of diagnosis. At the same time, due to the use of an interpretable model, the system is particularly good at training and learning abnormal or rare cases.

[0234] The system can continuously learn new knowledge from new medical image data and optimize its own performance. This function not only helps students understand the update and evolution of medical knowledge, but also stimulates their interest in learning and encourages them to actively participate in learning and exploration.

[0235] The system can intelligently adjust the learning content and difficulty according to the students' learning progress and feedback, and provide personalized learning paths and feedback. This personalized learning method can better meet the different learning needs of students and improve their learning effects and motivation.

[0236] Through continuous learning and optimization, the system can constantly adapt to the latest developments and changes in medical knowledge, enabling students to handle new medical knowledge with ease.

[0237] For example, in a medical imaging diagnosis training course at a medical school, students and teachers encountered a common challenge: how to impart diagnostic knowledge within limited classroom time while ensuring that students can deeply understand and apply this knowledge. In addition, with the rapid development of medical technology, new disease characteristics and diagnostic methods continue to emerge, and traditional training methods are difficult to adapt to these changes in a timely manner.

[0238] At this time, the medical school introduced the medical imaging diagnosis training and learning system based on the interpretable model provided in the embodiment of the present application. This system not only provides a large amount of real medical image data and corresponding diagnosis results, but also clearly demonstrates the key features of the model in identifying the disease, making inferences and drawing conclusions during the diagnosis process.

[0239] Suppose Xiao Wang is a student majoring in medical imaging diagnosis. He finds that he is particularly interested in the diagnosis of brain diseases. In the system, he chooses a special topic for brain disease study. Based on Xiao Wang's choice, the system generates personalized medical imaging diagnosis case exercises of special difficulty for him.

[0240] Every time Xiao Wang completes a learning task, the system will dynamically adjust the learning difficulty and content based on his learning progress and answering situation, ensuring that he can master diagnostic knowledge step by step.

[0241] When learning how to interpret MRI images of brain diseases, Xiao Wang encountered an MRI brain image that was difficult to judge. He could refer to the concept comparison results displayed in detail in the system, extract key features such as brain structure and abnormal areas from the CT image, and make inferences and draw diagnostic conclusions based on these key features.

[0242] Through this process, Xiao Wang not only understood the diagnosis results, but also gained a deep understanding of the principles and processes of diagnosis, which enhanced his understanding and application of diagnostic knowledge.

[0243] As medical research deepens, new brain disease characteristics and diagnostic methods are constantly being discovered. The system automatically updates and incorporates the latest diagnostic knowledge and cases into the material library.

[0244] Xiao Wang learned the new knowledge under the guidance of the system and consolidated his learning results through new diagnostic cases. He found that he could now more accurately judge the type and degree of brain diseases.

[0245] Through this medical imaging diagnosis training and learning system based on an interpretable model, Xiao Wang not only quickly mastered the basic knowledge and skills of medical imaging diagnosis, but also improved his learning ability and comprehensive quality. At the same time, the system also provides a new teaching model and method for the medical imaging diagnosis training courses of medical schools, promoting the innovation and development of medical imaging diagnosis education.

[0246] The interactive method provided in the embodiments of the present application can be applied not only to medical image diagnosis training systems, but also to medical image analysis software, intelligent diagnosis auxiliary equipment, online learning platforms, virtual reality (VR) and augmented reality (AR) applications and other scenarios.

[0247] Medical imaging diagnosis training systems usually involve a large amount of medical imaging data and complex diagnostic processes. The interactive method provided in the embodiment of the present application can make the diagnostic process clearer through an interpretable model, which is easier for trainees to understand and remember.

[0248] For example, in a medical image interpretation training system, the interactive method provided in the embodiment of the present application can be integrated so that residents can more intuitively understand and learn diagnostic logic during the clinical image thinking ability training process.

[0249] Medical image analysis software usually needs to process and analyze a large amount of medical image data to assist doctors in making diagnoses. Introducing an interpretable model can improve the usability and credibility of the software and help doctors better understand the analysis results.

[0250] In medical image analysis software, by integrating the interactive method provided in the embodiment of the present application, more detailed explanations can be provided in tasks such as image segmentation and registration, thereby enhancing doctors' confidence in the results.

[0251] In the teaching process of intelligent diagnostic auxiliary equipment, such as portable ultrasound diagnostic equipment, mobile X-ray equipment, etc., integrating the interactive method provided by the embodiment of the present application can provide more accurate and reliable diagnostic results. Through the interpretable model, the device can show the key features and reasoning process of the diagnostic process to the doctor, improving the teaching effect.

[0252] The online learning platform can provide remote education and training services for medical imaging diagnosis. The interactive method provided in the embodiment of the present application can provide students and doctors with personalized learning paths and feedback, and dynamically adjust the learning content and difficulty according to the learning progress and understanding level.

[0253] For example, in an online course on medical imaging diagnosis, the interactive method provided in the embodiment of the present application can intelligently recommend relevant learning materials and exercises based on students' answers and feedback, thereby improving learning efficiency.

[0254] In VR and AR applications of medical imaging, the interactive method provided by the embodiments of the present application can provide users with a more immersive and intuitive learning experience. Users can simulate real diagnostic scenarios in a virtual environment and understand the key information and reasoning process in the diagnostic process through an interpretable model.

[0255] Based on the foregoing embodiments, an embodiment of the present application provides an interactive device, which includes the units included and the modules included in the units, which can be implemented by a processor in a computer device; of course, it can also be implemented by a specific logic circuit; in the implementation process, the processor can be a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP) or a field programmable gate array (FPGA), etc.

[0256] Figure 6 A schematic diagram of the structure of an interactive device provided in an embodiment of the present application is shown in FIG. Figure 6 As shown, the interactive device 500 includes: a first display module 510, a second display module 520, a third display module 530 and a determination module 540, wherein:

[0257] The first display module 510 is used to display a first interactive interface; the first interactive interface includes an image display area and a question answering area;

[0258] A second display module 520, used for displaying the medical image to be analyzed in the image display area;

[0259] A third display module 530 is used to display at least one disease sign identification question and disease diagnosis question corresponding to the medical image in the question answering area;

[0260] The determination module 540 is used to obtain the first answer information of each disease symptom identification question and the second answer information of the disease diagnosis question in response to the answer operation performed by the user in the answer area, and determine the user's answer result for the medical image based on the first answer information, the second answer information and the answer information corresponding to the medical image.

[0261] In some embodiments, the disease symptom identification question includes a disease symptom selection question, and the disease symptom selection question has multiple disease symptom description options; the above-mentioned device also includes: an analysis module, which is used to: before displaying at least one disease symptom identification question and disease diagnosis question corresponding to the medical image in the answer area, use the trained image analysis model to analyze and process the medical image to obtain multiple disease symptom description options corresponding to at least one disease symptom contained in the medical image; for each of the disease symptoms, based on the multiple disease symptom description options corresponding to the disease symptom, generate a disease symptom selection question corresponding to the disease symptom.

[0262] In some embodiments, the analysis module is also used to: use a trained image analysis model to analyze and process the medical image to obtain a set of symptom descriptions corresponding to at least one symptom contained in the medical image, and the confidence of each symptom description in the symptom description set; for each symptom, select multiple symptom descriptions from the symptom description set corresponding to the symptom as the symptom description options corresponding to the symptom, and the selected multiple symptom descriptions include at least the symptom description with the highest confidence in the symptom description set.

[0263] In some embodiments, the answer information corresponding to the medical image includes first answer information of each disease sign identification question corresponding to the medical image, and second answer information of the disease diagnosis question corresponding to the medical image, and the answer results include the first answer result of each disease sign identification question, and the second answer result of the disease diagnosis question; the determination module is also used to: for each disease sign identification question, based on the first answer information and the first answer information of the disease sign identification question, determine the first answer result of the disease sign identification question; based on the second answer information and the second answer information, determine the second answer result of the disease diagnosis question.

[0264] In some embodiments, the determination module is further configured to:

[0265] When the first answer information and the first answer information of the disease symptom identification question are the same, determining that the first answer result of the disease symptom identification question is correct;

[0266] When the semantic similarity between the first answer information and the first answer information of the disease symptom identification question is greater than or equal to the similarity threshold, determining that the first answer result of the disease symptom identification question is a correct answer;

[0267] When the first answer information and the first answer information of the disease symptom identification question are different and the semantic similarity between the first answer information and the first answer information of the disease symptom identification question is less than the similarity threshold, it is determined that the first answer result of the disease symptom identification question is an incorrect answer.

[0268] In some embodiments, each of the disease sign identification questions is used to assess the ability to identify a disease sign contained in the medical image, and the disease diagnosis question is used to assess the ability to diagnose the type of disease corresponding to the medical image; the device also includes: an acquisition module, used to: determine the learning ability parameter of the user before displaying the medical image to be analyzed in the image display area; based on the learning ability parameter, acquire the medical image and at least one disease sign identification question and disease diagnosis question corresponding to the medical image from the assessment question bank.

[0269] In some embodiments, the acquisition module is also used to: determine the target assessment difficulty based on the learning ability parameter; and acquire the medical image and at least one disease sign identification question and disease diagnosis question corresponding to the medical image from the assessment question bank based on the target assessment difficulty.

[0270] In some embodiments, the acquisition module is also used to: determine the prevalence of the target disease that matches the target assessment difficulty; based on the prevalence of the target disease, acquire the medical image and at least one disease sign identification question and disease diagnosis question corresponding to the medical image from the assessment question bank.

[0271] In some embodiments, the acquisition module is also used to: based on the learning ability parameter, obtain a candidate medical image sequence and at least one disease sign identification question and disease diagnosis question corresponding to each candidate medical image in the candidate medical image sequence from the assessment question bank; take each medical image in the candidate medical image sequence as the medical image to be analyzed in turn, and obtain at least one disease sign identification question and disease diagnosis question corresponding to the medical image.

[0272] In some embodiments, the above-mentioned device also includes: an updating module, which is used to: update the learning ability parameters of the user based on the user's answer results for the medical image; based on the updated learning ability parameters, obtain the next medical image and at least one disease sign identification question and disease diagnosis question corresponding to the next medical image from the assessment question bank.

[0273] The description of the above device embodiment is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. In some embodiments, the functions or modules included in the device provided in the embodiment of the present application can be used to execute the method described in the above method embodiment. For technical details not disclosed in the device embodiment of the present application, please refer to the description of the method embodiment of the present application for understanding.

[0274] It should be noted that in the embodiment of the present application, if the above method is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, which is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM), a disk or an optical disk. In this way, the embodiment of the present application is not limited to any specific hardware, software or firmware, or any combination of hardware, software, and firmware.

[0275] An embodiment of the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the program, some or all of the steps in the above method are implemented.

[0276] The embodiment of the present application provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, some or all of the steps in the above method are implemented. The computer-readable storage medium can be transient or non-transient.

[0277] An embodiment of the present application provides a computer program, including a computer-readable code. When the computer-readable code is run in a computer device, a processor in the computer device executes some or all of the steps for implementing the above method.

[0278] An embodiment of the present application provides a computer program product, including a computer program or instructions, which, when executed by a processor, implements some or all of the steps in the above method.

[0279] The embodiment of the present application provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and when the computer program is read and executed by a computer, some or all of the steps in the above method are implemented. The computer program product can be implemented specifically by hardware, software or a combination thereof. In some embodiments, the computer program product is specifically embodied as a computer storage medium, and in other embodiments, the computer program product is specifically embodied as a software product, such as a software development kit (SDK) and the like.

[0280] It should be noted here that the description of the various embodiments above tends to emphasize the differences between the various embodiments, and the same or similar aspects can be referenced to each other. The description of the above device, storage medium, computer program and computer program product embodiments is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. For technical details not disclosed in the embodiments of the device, storage medium, computer program and computer program product of this application, please refer to the description of the method embodiment of this application for understanding.

[0281] It should be noted that Figure 7 A schematic diagram of a hardware entity of a computer device in an embodiment of the present application is shown in FIG. Figure 7 As shown, the hardware entity of the computer device 600 includes: a processor 601, a communication interface 602 and a memory 603, wherein:

[0282] Processor 601 generally controls the overall operation of computer device 600 .

[0283] The communication interface 602 enables the computer device to communicate with other terminals or servers through a network.

[0284] The memory 603 is configured to store instructions and applications executable by the processor 601, and can also cache data to be processed or processed by the processor 601 and each module in the computer device 600 (for example, image data, audio data, voice communication data, and video communication data), which can be implemented by flash memory (FLASH) or random access memory (Random Access Memory, RAM). Data transmission can be performed between the processor 601, the communication interface 602, and the memory 603 through the bus 604.

[0285] It should be understood that "one embodiment" or "an embodiment" mentioned throughout the specification means that specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present application, the size of the serial number of each step / process mentioned above does not mean the order of execution, and the execution order of each step / process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application. The serial numbers of the embodiments of the present application mentioned above are for description only and do not represent the advantages and disadvantages of the embodiments.

[0286] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0287] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0288] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, the functional units in the embodiments of the present application may be all integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0289] A person skilled in the art can understand that all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM), magnetic disks or optical disks, etc., various media that can store program codes.

[0290] Alternatively, if the above-mentioned integrated unit of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can essentially or in other words, the part that contributes to the relevant technology can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0291] The above is only an implementation method of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed in the present application should be included in the protection scope of the present application.

Claims

1. An interactive method, characterized in that: The method comprises: Displaying a first interactive interface; the first interactive interface includes an image display area and a question answering area; Displaying the medical image to be analyzed in the image display area; Displaying at least one disease symptom identification question and disease diagnosis question corresponding to the medical image in the question answering area; In response to the user's answering operation in the answering area, the first answer information of each disease symptom identification question and the second answer information of the disease diagnosis question are obtained, and based on the first answer information, the second answer information and the answer information corresponding to the medical image, the user's answer result for the medical image is determined.

2. The method according to claim 1, characterized in that The disease symptom identification questions include disease symptom selection questions, and the disease symptom selection questions have multiple disease symptom description options; Before displaying at least one disease symptom identification question and disease diagnosis question corresponding to the medical image in the question answering area, the method further includes: Analyzing and processing the medical image using the trained image analysis model to obtain a plurality of disease symptom description options corresponding to at least one disease symptom contained in the medical image; For each of the symptom, a symptom multiple-choice question corresponding to the symptom is generated based on a plurality of symptom description options corresponding to the symptom.

3. The method according to claim 2, characterized in that The method of using the trained image analysis model to analyze and process the medical image to obtain a plurality of symptom description options corresponding to at least one symptom contained in the medical image, including: Analyzing and processing the medical image using the trained image analysis model to obtain a set of disease symptom descriptions corresponding to at least one disease symptom contained in the medical image, and a confidence level of each disease symptom description in the set of disease symptom descriptions; For each of the symptoms, multiple symptom descriptions are selected from the symptom description set corresponding to the symptom as symptom description options corresponding to the symptom, and the selected multiple symptom descriptions at least include the symptom description with the highest confidence in the symptom description set.

4. The method according to claim 1, characterized in that: The answer information corresponding to the medical image includes the first answer information of each disease sign identification question corresponding to the medical image, and the second answer information of the disease diagnosis question corresponding to the medical image, and the answer result includes the first answer result of each disease sign identification question, and the second answer result of the disease diagnosis question; The determining, based on the first answer information, the second answer information, and the answer information corresponding to the medical image, the answer result of the user for the medical image includes: For each of the disease symptom identification questions, determining a first answer result of the disease symptom identification question based on the first answer information and the first answer information of the disease symptom identification question; Based on the second answer information and the second answer information, a second answer result of the disease diagnosis question is determined.

5. The method according to claim 4, characterized in that The determining of the first answer result of the disease symptom identification question based on the first answer information and the first answer information of the disease symptom identification question includes at least one of the following: When the first answer information and the first answer information of the disease symptom identification question are the same, determining that the first answer result of the disease symptom identification question is correct; When the semantic similarity between the first answer information and the first answer information of the disease symptom identification question is greater than or equal to the similarity threshold, determining that the first answer result of the disease symptom identification question is a correct answer; When the first answer information and the first answer information of the disease symptom identification question are different and the semantic similarity between the first answer information and the first answer information of the disease symptom identification question is less than the similarity threshold, it is determined that the first answer result of the disease symptom identification question is an incorrect answer.

6. The method according to any one of claims 1 to 5, characterized in that Each of the disease symptom recognition questions is used to assess the ability to recognize a disease symptom contained in the medical image, and the disease diagnosis question is used to assess the ability to diagnose the disease type corresponding to the medical image; Before the image display area displays the medical image to be analyzed, the method further includes: Determining a learning ability parameter of the user; Based on the learning ability parameter, the medical image and at least one disease symptom identification question and disease diagnosis question corresponding to the medical image are obtained from a test question bank.

7. The method according to claim 6, characterized in that The obtaining, based on the learning ability parameter, the medical image and at least one disease symptom identification question and disease diagnosis question corresponding to the medical image from a test question bank comprises: Determining the difficulty of the target assessment based on the learning ability parameter; Based on the target assessment difficulty, the medical image and at least one disease symptom identification question and disease diagnosis question corresponding to the medical image are obtained from the assessment question bank.

8. The method according to claim 7, characterized in that The obtaining, based on the target assessment difficulty, the medical image and at least one disease symptom identification question and disease diagnosis question corresponding to the medical image from the assessment question bank comprises: Determine the prevalence of the target disease that matches the difficulty of the target assessment; Based on the prevalence of the target disease, the medical image and at least one disease symptom identification question and disease diagnosis question corresponding to the medical image are obtained from the assessment question bank.

9. The method according to claim 6, characterized in that The obtaining, based on the learning ability parameter, the medical image and at least one disease symptom identification question and disease diagnosis question corresponding to the medical image from a test question bank comprises: Based on the learning ability parameter, obtaining a candidate medical image sequence and at least one disease symptom recognition question and a disease diagnosis question corresponding to each candidate medical image in the candidate medical image sequence from a test question bank; Each medical image in the candidate medical image sequence is sequentially used as the medical image to be analyzed, and at least one disease symptom recognition question and disease diagnosis question corresponding to the medical image is obtained.

10. The method according to claim 6, characterized in that The method further comprises: Based on the user's answer results for the medical image, updating the user's learning ability parameter; Based on the updated learning ability parameter, the next medical image and at least one disease symptom identification question and disease diagnosis question corresponding to the next medical image are obtained from the examination question bank.

11. An interactive device, characterized in that: The device comprises: A first display module, used to display a first interactive interface; the first interactive interface includes an image display area and a question answering area; A second display module, used for displaying the medical image to be analyzed in the image display area; A third display module, configured to display at least one disease symptom identification question and a disease diagnosis question corresponding to the medical image in the question answering area; A determination module is used to respond to the user's answering operation in the answering area, obtain the first answer information of each of the disease symptom identification questions and the second answer information of the disease diagnosis question, and determine the user's answer result for the medical image based on the first answer information, the second answer information and the answer information corresponding to the medical image.

12. A computer device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, wherein: When the processor executes the program, the steps in the method according to any one of claims 1 to 10 are implemented.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

14. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps in the method according to any one of claims 1 to 10 are implemented.

Images