Intelligent health management system and method for glandular hypertrophy patient

Abstract

The invention provides an intelligent health management system and method for glandular hypertrophy patients, and the method comprises the steps: carrying out the training and construction of face image features capable of recognizing different symptom degrees based on image data sets of adenoid faces of different degrees, and predicting and outputting the symptom degrees and clinical diagnosis and treatment strategy keywords on the face image features; and further generating an adenoid face pre-diagnosis scheme for the patient according to the symptom degree and the clinical diagnosis and treatment strategy keyword. Whether the user has the adenoid face or not and the symptom degree of the adenoid face are predicted through the AI technology, and the clinical diagnosis and treatment strategy is recommended, so that a doctor can be assisted to quickly diagnose the adenoid face, and the clinical diagnosis difficulty is reduced. According to the invention, parents can collect the face image of the child through the APP terminal at any time and upload the image to the hospital background server, predict the face of the adenoid body and obtain a pre-diagnosis prediction result, timely give risk pre-warning prompt to the face of the adenoid body, prevent in advance, intervene in treatment in time, reduce the risk, and care about the physical and psychological health development of the child in advance.

Description

Technical Field

[0001] The present invention relates to the field of intelligent medical management technology, and in particular to an intelligent health management system and method for patients with glandular hypertrophy. Background Art

[0002] Adenoids hypertrophy refers to a disease in which the adenoids are repeatedly stimulated by inflammation, resulting in pathological hyperplasia and hypertrophy, which causes corresponding symptoms. This disease is common in children, but it can also occur in some adults, often combined with chronic tonsillitis or tonsil hypertrophy.

[0003] The main causes of adenoid hypertrophy are repeated acute and chronic nasopharyngitis, or inflammation of the nasal cavity, sinuses, tonsils, etc., which affect the nasopharynx and stimulate the proliferation of adenoid tissue. Allergic reactions, gastroesophageal reflux, and active proliferation of the body's lymphatic tissue can also increase the risk of adenoid tissue hyperplasia. Adenoid hypertrophy is the main cause of airway obstruction and specific maxillofacial deformities, resulting in an adenoid face.

[0004] Adenoid facies are mostly caused by adenoid hypertrophy. Figure 1 The diagram shown is a comparison between adenoid facies and normal faces. The facial features of adenoid facies include short, thick and upturned upper lip, irregular teeth, protruding upper incisors, lack of expression, high arched jaws and drooping mandible.

[0005] The facial features of different degrees of adenoids and their symptomatic treatment plans are as follows:

[0006] 1. Mild adenoid facies

[0007] Features: The upper lip may be slightly raised and the teeth may be slightly uneven.

[0008] Treatment plan: First, observe whether the patient has symptoms such as mouth breathing or snoring. If so, seek medical attention in time to prevent prolonged mouth breathing from worsening the adenoid face. Secondly, follow the doctor's advice to use nasal aerosols, such as oxymetazoline hydrochloride spray, Fentel nasal spray, or give montelukast sodium treatment to improve the inflammatory response in the nasal cavity and reduce the stimulation to the adenoids.

[0009] 2. Moderate adenoid facies

[0010] Features: The upper lip is obviously upturned, the teeth are not neatly arranged, the upper incisors may protrude, and the facial expression is relatively stiff.

[0011] Treatment plan: On the basis of drug treatment, if the enlarged adenoids have caused obvious clinical symptoms, such as nasal congestion, mouth breathing, snoring, etc., and the drug treatment is not effective, surgical removal of the enlarged adenoids can be considered to fundamentally prevent the development of adenoid facies. After surgery, orthodontic treatment can be performed as needed to improve problems such as uneven tooth alignment.

[0012] 3. Severe adenoid facies

[0013] Features: The adenoid face has obvious features, such as severely upturned upper lip, severely uneven teeth, high arched jaw, drooping mandible, and severe lack of vividness in facial expression.

[0014] Treatment plan: For severe adenoids, surgical removal of the adenoids is the primary treatment. After surgery, the patient may need a longer period of orthodontic treatment to repair tooth defects and improve facial deformities. If the patient's facial deformity is too severe, orthognathic and orthodontic treatment should also be considered. In addition, patients should maintain an optimistic and positive attitude, exercise appropriately, improve their body's immunity, and promote the recovery of facial bones and soft tissues.

[0015] At present, most children with adenoids face usually do not notice it until they have moderate or severe adenoids face. Therefore, the treatment process is usually delayed, which delays the healthy development of children and seriously affects the physical and mental health of patients. If it can be prevented in advance, it can greatly improve the health care for children (or other patients).

[0016] In addition, the current clinical examination of adenoids facies is mainly judged by doctors based on medical data and experience (the proportion of the total number of cases is small, so a certain amount of experience is required to identify it), but this requirement is relatively high, so it is easy to cause diagnostic errors (degree of symptom) or ignore the diagnosis of adenoids facies (especially early symptoms are not obvious, especially in children under 3 years old, which may lead to misdiagnosis), and it is impossible to accurately diagnose adenoids facies and quickly give corresponding corrective treatment plans.

[0017] Finally, parents are unable to observe their children at any time and determine whether they have adenoids facies, so there is a lack of home-based adenoids facies monitoring and early warning programs. Summary of the invention

[0018] In order to solve the technical problems existing in the prior art, the present invention provides the following technical solutions:

[0019] On the one hand, a method for intelligent health management of patients with glandular hypertrophy is provided, the method being implemented by an electronic device, the method comprising:

[0020] S1, collect the user's facial image through the user terminal and upload it to the backend server;

[0021] S2, the backend server identifies and extracts a facial contour image from the facial image, and records the facial contour image into the user's electronic medical record file in the hospital information system;

[0022] S3, simultaneously inputting the facial contour image into a pre-deployed adenoids facial intelligent prognostic model, and having the adenoids facial intelligent prognostic model recognize facial image features in the facial contour image and output symptom levels and clinical diagnosis and treatment strategy keywords related to the facial image features;

[0023] S4, writing the facial image features and the corresponding symptom levels and clinical diagnosis and treatment strategy keywords into the user's electronic medical record file;

[0024] S5. Generate a first adenoids facial prognosis plan corresponding to the symptom level according to the clinical diagnosis and treatment strategy keywords; and retrieve a corresponding second adenoids facial prognosis plan from a medical case library based on the symptom level and the facial image features through a large language model; and integrate the first adenoids facial prognosis plan and the second adenoids facial prognosis plan and write them into the user's electronic medical record file;

[0025] S6. Regularly send the user's electronic medical record to the user end and / or the medical end.

[0026] Preferably, the method for generating the adenoids facial intelligent prognostic model comprises:

[0027] Collect and preprocess image datasets of faces with different degrees of adenoids;

[0028] Performing feature engineering on the image dataset, using CNN to identify and extract facial image features on each adenoid facial image in the image dataset, including facial contour, adenoid shape and size, lip and tooth shape, and jaw / palate shape;

[0029] Performing feature annotation on the facial image features, annotating corresponding symptom levels and clinical diagnosis and treatment strategy keywords, wherein the symptom levels include prompt information on whether the facial expression is lacking as determined based on the facial image features;

[0030] Counting the features of each facial image to obtain a feature set;

[0031] Dividing the feature set into a training set and a validation set according to a preset ratio;

[0032] Inputting the training set into a preset CNN model for feature training and learning, thereby constructing the adenoid facial intelligent pre-diagnosis model;

[0033] The recognition and prediction performance of the adenoids facial intelligent pre-diagnosis model was verified using the delivery validation set:

[0034] If the verification is passed, the adenoids facial intelligent pre-diagnosis model is deployed on the background server and put into the application service of adenoids facial pre-diagnosis after adjusting the parameters;

[0035] If the verification fails, the above steps are repeated to reconstruct the adenoid facial intelligent pre-diagnosis model.

[0036] Preferably, when collecting image datasets of adenoids with different degrees of appearance, the image datasets of adenoids with different degrees of appearance include:

[0037] frontal face images;

[0038] frontal smiling face images;

[0039] 45° image of the left side of the face and 45° image of the right side of the face;

[0040] 90° image of the left side of the face and 90° image of the right side.

[0041] On the other hand, an intelligent health management system for patients with glandular hypertrophy is provided, which is applied to an intelligent health management method for patients with glandular hypertrophy. The system includes:

[0042] The user end is used to collect the user's facial image and upload it to the backend server;

[0043] Backend server, including:

[0044] Hospital information system, used to generate and manage users' electronic medical records;

[0045] A contour recognition module, used for recognizing and extracting a facial contour image from the facial image, and recording the facial contour image into a user's electronic medical record file in a hospital information system;

[0046] Adenoids facial AI pre-diagnosis module, used for inputting the facial contour image into a pre-deployed adenoids facial intelligent pre-diagnosis model, and the adenoids facial intelligent pre-diagnosis model identifies the facial image features in the facial contour image and outputs the symptom degree and clinical diagnosis and treatment strategy keywords related to the facial image features;

[0047] A prediction result recording module, used for writing the facial image features and the corresponding symptom levels and clinical diagnosis and treatment strategy keywords into the user's electronic medical record file;

[0048] A strategy generation module, for generating a first adenoids facial prognosis plan corresponding to the symptom degree according to the clinical diagnosis and treatment strategy keywords; and for retrieving a corresponding second adenoids facial prognosis plan from a medical case library based on the symptom degree and the facial image features through a large language model;

[0049] A strategy integration module, used for integrating the first adenoids facial pre-diagnosis plan and the second adenoids facial pre-diagnosis plan and writing them into the user's electronic medical record file;

[0050] A recommendation module, used for regularly sending the user's electronic medical record to the user terminal and / or the medical terminal;

[0051] The medical terminal is used to log in and view the electronic medical records of the user;

[0052] The user end and the medical end are respectively connected to the background server for communication.

[0053] On the other hand, an electronic device is provided, comprising: a processor; a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, any one of the above-mentioned intelligent health management methods for patients with glandular hypertrophy is implemented.

[0054] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement any one of the above-mentioned intelligent health management methods for patients with glandular hypertrophy.

[0055] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0056] The present invention is based on the training of image data sets of adenoids of different degrees to construct facial image features that can identify different degrees of symptoms and predict and output the symptom degree and clinical diagnosis and treatment strategy keywords on the facial image features, and further generates an adenoids face pre-diagnosis plan for the patient based on the symptom degree and clinical diagnosis and treatment strategy keywords. By using AI technology to predict whether the user has adenoids face and the degree of symptoms of the adenoids face and recommend clinical diagnosis and treatment strategies, it can assist doctors in quickly diagnosing adenoids face and reduce the difficulty of clinical diagnosis. The present invention allows parents to collect children's facial images through the APP terminal at any time and upload them to the hospital's backend server, predict the adenoids face and obtain pre-diagnosis prediction results, give timely risk warning prompts for adenoids face, take precautions in advance, intervene in treatment in time, reduce risks, and care about the physical and mental health development of children in advance. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0058] Figure 1It is a schematic diagram comparing adenoids face and normal face;

[0059] Figure 2 This is a flow chart of an intelligent health management method for patients with glandular hypertrophy provided by an embodiment of the present invention;

[0060] Figure 3 It is a schematic diagram of the construction process of an adenoids facial intelligent pre-diagnosis model provided by an embodiment of the present invention;

[0061] Figure 4 This is a block diagram of an intelligent health management system for patients with glandular hypertrophy provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0063] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0064] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same. "of", "corresponding, relevant" and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same.

[0065] In the embodiments of the present invention, sometimes a subscript such as W1 may be mistakenly written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are consistent.

[0066] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0067] Terminology explanation:

[0068] 1. CNN

[0069] That is, Convolutional Neural Networks (CNN), a type of feedforward neural network that includes convolution calculations and has a deep structure, is one of the representative algorithms of deep learning. Convolutional neural networks have representation learning capabilities and can perform shift-invariant classification on input information according to their hierarchical structure, so they are also called "Shift-Invariant Artificial Neural Networks (SIANN)". The corresponding CNN model is also a traditional algorithm model, usually composed of the following parts:

[0070] ‌Convolutional layer‌: Extracts local features of input data through convolution operations. The convolution kernel slides over the input data, performs element-by-element dot product operations, and generates feature maps. The convolutional layer reduces the number of parameters and improves efficiency through local connections and weight sharing.

[0071] ‌Pooling layer‌: Downsamples the feature map to reduce the data dimension and retain important features. Common pooling methods include maximum pooling and average pooling.

[0072] ‌Fully connected layer‌: Integrates the features extracted by the convolutional layer and the pooling layer and maps them to the classification space, outputting the final classification result.

[0073] Activation function: such as ReLU, which introduces nonlinear characteristics and enables the model to learn complex features.

[0074] CNN workflow

[0075] ‌Input layer‌: input image data.

[0076] ‌Convolutional layer‌: Extracts local features of the image and gradually extracts higher-level features through multiple convolutional layers.

[0077] ‌Pooling layer‌: Downsample the feature map to reduce the data dimension.

[0078] Activation function: introduces nonlinear characteristics.

[0079] ‌Fully connected layer‌: Integrate and map the extracted features into the classification space.

[0080] ‌Output layer‌: The Softmax function is usually used for multi-classification tasks to output the probability distribution of each category.

[0081] 2. Hospital Information System

[0082] Hospital Information System (HIS) is the core system of hospital management. It aims to provide comprehensive medical information support for medical staff by integrating medical information from various departments of the hospital, such as patient information, doctor's order information, drug information, test information, etc. It uses modern technologies such as computer hardware and software, network communications, etc. to comprehensively manage the flow of people, logistics, and finance in the hospital, optimize the medical process, and improve medical efficiency and service quality.

[0083] Functions of Hospital Information System

[0084] ‌Information integration and management‌, the hospital information system integrates the data of various departments of the hospital, including patient diagnosis and treatment information, drug information, test information, etc., to achieve information sharing and collaborative work.

[0085] ‌Financial management‌, the system is integrated with the financial management system to achieve comprehensive management of hospital financial data and improve financial management efficiency.

[0086] ‌Clinical support‌ provides functions such as electronic medical records and medical order management to help doctors quickly obtain patient information and optimize the diagnosis and treatment process.

[0087] ‌Operational optimization‌: through real-time data collection and analysis, hospitals can accurately grasp business status, optimize business processes, and shorten patient consultation time.

[0088] The embodiment of the present invention provides a method for intelligent health management of patients with glandular hypertrophy, which can be implemented by an electronic device, which can be a terminal or a server. Figure 2 The flowchart of the intelligent health management method for patients with glandular hypertrophy is shown, and the processing flow of the method may include the following steps:

[0089] S1, collect the user's facial image through the user terminal and upload it to the backend server;

[0090] S2, the backend server identifies and extracts a facial contour image from the facial image, and records the facial contour image into the user's electronic medical record file in the hospital information system;

[0091] S3, simultaneously inputting the facial contour image into a pre-deployed adenoids facial intelligent prognostic model, and having the adenoids facial intelligent prognostic model recognize facial image features in the facial contour image and output symptom levels and clinical diagnosis and treatment strategy keywords related to the facial image features;

[0092] S4, writing the facial image features and the corresponding symptom levels and clinical diagnosis and treatment strategy keywords into the user's electronic medical record file;

[0093] S5. Generate a first adenoids facial prognosis plan corresponding to the symptom level according to the clinical diagnosis and treatment strategy keywords; and retrieve a corresponding second adenoids facial prognosis plan from a medical case library based on the symptom level and the facial image features through a large language model; and integrate the first adenoids facial prognosis plan and the second adenoids facial prognosis plan and write them into the user's electronic medical record file;

[0094] S6. Regularly send the user's electronic medical record to the user end and / or the medical end.

[0095] The present invention is based on the training of image data sets of adenoids of different degrees to construct facial image features that can identify different degrees of symptoms and predict and output the symptom degree and clinical diagnosis and treatment strategy keywords on the facial image features, and further generates an adenoids face pre-diagnosis plan for the patient based on the symptom degree and clinical diagnosis and treatment strategy keywords. By using AI technology to predict whether the user has adenoids face and the degree of symptoms of the adenoids face and recommend clinical diagnosis and treatment strategies, it can assist doctors in quickly diagnosing adenoids face and reduce the difficulty of clinical diagnosis. The present invention allows parents to collect children's facial images through the APP terminal at any time and upload them to the hospital's backend server, predict the adenoids face and obtain pre-diagnosis prediction results, give timely risk warning prompts for adenoids face, take precautions in advance, intervene in treatment in time, reduce risks, and care about the physical and mental health development of children in advance.

[0096] Combination Figure 4 As shown, the present invention deploys an adenoids facial AI pre-diagnosis module on the hospital background, which can input the facial contour image of the user (mainly children, but also other patients and other users) into the pre-deployed adenoids facial intelligent pre-diagnosis model, and the adenoids facial intelligent pre-diagnosis model recognizes the facial image features in the facial contour image and outputs the symptom degree and clinical diagnosis and treatment strategy keywords on the facial image features. The subsequent system searches and recommends adenoids facial clinical diagnosis and treatment plans of the corresponding degree based on the symptom degree and clinical diagnosis and treatment strategy keywords. The background can recommend the recognition results and treatment plans to the user end and the medical end (such as the PC end) for doctors to quickly pre-diagnose and for parents to check the children's adenoids facial screening results in time.

[0097] The user's electronic medical record file for pre-diagnosis can be pre-built on the background hospital information system (refer to the registration function of the existing hospital or the hospital information system file recording function), and the user's APP information can be bound in advance to facilitate the hospital to push adenoids facial screening notifications or results.

[0098] The communication solution between the user end or medical end and the backend server can be API communication, such as:

[0099] Overall architecture

[0100] ‌Client Layer‌:

[0101] It includes patient-side App (user-side) and medical-side App, supporting Android and iOS platforms.

[0102] The client app interacts with users through a graphical user interface (GUI) and collects user requests and data.

[0103] ‌Cloud service platform layer‌:

[0104] It includes data processing center, business logic processing, AI-assisted decision-making system, database, etc.

[0105] Responsible for processing requests sent by clients, executing corresponding business logic, and storing and managing data.

[0106] ‌Network and security requirements‌:

[0107] Adopt hybrid cloud deployment to ensure efficient and secure data processing.

[0108] Use CDN acceleration services to improve user access speed.

[0109] Implement data encryption transmission (HTTPS), data desensitization, access control and log auditing to ensure the security of data transmission and storage.

[0110] Communication protocols and technologies

[0111] ‌HTTP Communication‌:

[0112] One of the most common ways of communication is through the HTTP protocol. The client App sends data to the server through HTTP requests, and the server returns the corresponding data to the client through HTTP responses.

[0113] For example, a client App can send a GET request to obtain data, or send a POST request to submit form data.

[0114] ‌RESTful API‌:

[0115] RESTful API is a communication method based on HTTP protocol, which realizes communication between client and server by sending HTTP requests and receiving HTTP responses.

[0116] The client App can use different HTTP methods (such as GET, POST, PUT, DELETE, etc.) to send requests to the server. The server will process the request accordingly and return the corresponding data to the client.

[0117] ‌WebSocket Communication‌:

[0118] WebSocket is a real-time communication protocol that enables two-way communication between servers and clients.

[0119] Suitable for application scenarios that require real-time data push, such as instant messaging, remote monitoring, etc.

[0120] WebSocket connections are full-duplex, meaning that the server can actively push messages to the client, rather than just responding to requests sent by the client.

[0121] ‌Socket Communication‌:

[0122] Socket is a communication protocol based on TCP / IP that can realize two-way communication between server and client.

[0123] Suitable for applications with high performance requirements, such as online games, video calls, etc.

[0124] More code needs to be written to establish and maintain connections.

[0125] Communication process such as:

[0126] ‌Client initiates request‌:

[0127] Users initiate requests through the client App, such as querying medical records, making appointments, submitting forms, etc.

[0128] The client App encapsulates the request into an HTTP request or a request in another protocol format and sends it to the server over the network.

[0129] ‌Server processes the request‌:

[0130] After receiving the request, the server parses the request content and executes the corresponding business logic based on the request content.

[0131] The server may need to query a database, call other services, or perform complex calculations to generate the response data.

[0132] ‌Server returns response‌:

[0133] The server encapsulates the processing result into an HTTP response or a response in other protocol format and sends it back to the client through the network.

[0134] After receiving the response, the client App parses the response content and updates the user interface or performs other operations as needed.

[0135] Through the above solutions, the user end or medical end can effectively communicate with the backend server to achieve real-time data interaction and collaborative processing of business. At the same time, through reasonable network and security measures, the stability of communication and the security of data can be ensured.

[0136] like Figure 3 As shown, preferably, the method for generating the adenoids facial intelligent prognostic model comprises:

[0137] Collect and preprocess image datasets of faces with different degrees of adenoids;

[0138] Performing feature engineering on the image dataset, using CNN to identify and extract facial image features on each adenoid facial image in the image dataset, including facial contour, adenoid shape and size, lip and tooth shape, and jaw / palate shape;

[0139] Performing feature annotation on the facial image features, annotating corresponding symptom levels and clinical diagnosis and treatment strategy keywords, wherein the symptom levels include prompt information on whether the facial expression is lacking as determined based on the facial image features;

[0140] Counting the features of each facial image to obtain a feature set;

[0141] Dividing the feature set into a training set and a validation set according to a preset ratio;

[0142] Inputting the training set into a preset CNN model for feature training and learning, thereby constructing the adenoid facial intelligent pre-diagnosis model;

[0143] The recognition and prediction performance of the adenoids facial intelligent pre-diagnosis model was verified using the delivery validation set:

[0144] If the verification is passed, the adenoids facial intelligent pre-diagnosis model is deployed on the background server and put into the application service of adenoids facial pre-diagnosis after adjusting the parameters;

[0145] If the verification fails, the above steps are repeated to reconstruct the adenoid facial intelligent pre-diagnosis model.

[0146] The different degrees of facial distinctive features of adenoid facies can be expressed as mild, moderate, severe, etc., or as follows:

[0147] First-degree adenoid facies: The facial features do not change significantly, and may only show mild nasal congestion and runny nose, with little impact on the facial features;

[0148] Second degree adenoid facies: Obvious changes in facial features begin to appear, such as short, thick and upturned upper lip, irregular teeth, protruding upper incisors, etc. This is due to long-term mouth breathing caused by enlarged adenoids, which affects the normal development of facial muscles and bones;

[0149] ‌Third degree adenoid facies‌: Facial features change more significantly, including elongated jaw, high arched palate, lack of facial expression, etc. At this time, the adenoids almost completely block the nasopharynx, causing serious effects on facial appearance and respiratory function‌.

[0150] In order to extract more abundant facial features, the corresponding adenoid facial images collected here need to be collected from multiple angles, including:

[0151] frontal face images;

[0152] frontal smiling face images;

[0153] 45° image of the left side of the face and 45° image of the right side of the face;

[0154] 90° image of the left side of the face and 90° image of the right side.

[0155] Parents or nurses can assist children and other users in collecting corresponding adenoid facial images according to the above-mentioned angles and uploading them to the background, and pre-processing such as image cleaning and image enhancement can be performed to remove poor quality images, and perform brightness contrast enhancement and shadow removal on images with insufficient brightness and shadows.

[0156] Here we use convolutional neural network (CNN) to extract image features. Please refer to the following steps:

[0157] The following are the specific steps to refine the plan:

[0158] 1. Data preparation and preprocessing

[0159] 1.1 Image Dataset Collection

[0160] Collect a high-quality dataset of images of faces with varying degrees of adenoids.

[0161] Make sure the dataset covers patients of various ages and genders, as well as different lighting conditions, shooting angles, etc.

[0162] 1.2 Image Preprocessing

[0163] Scale and crop each image to ensure all images are of the same size and format.

[0164] Perform image enhancements such as rotation, flipping, adding noise, etc. to increase the diversity of the dataset.

[0165] Normalization is performed to scale the pixel values ​​to between 0 and 1 to improve the convergence speed of the model.

[0166] 2. Feature extraction and annotation

[0167] 2.1 Feature extraction using CNN

[0168] Build or select a pre-trained CNN model (such as VGG, ResNet, etc.) as a feature extractor.

[0169] A convolution operation is performed on each adenoid facial image to extract key features such as facial contour, adenoid shape and size, lip and tooth shape, and jaw / palate bone shape.

[0170] 2.2 Feature Annotation

[0171] Based on the extracted features, detailed feature annotation is performed on each image.

[0172] The annotation content should include: the specific symptom severity of adenoid facies (such as mild, moderate, severe), whether there is a lack of facial expression, and the corresponding clinical diagnosis and treatment strategy keywords.

[0173] Establish a labeling database to facilitate subsequent data management and use.

[0174] 3. Dataset Division

[0175] The labeled data set is divided into a training set and a validation set according to a preset ratio (such as 70% training set and 30% validation set).

[0176] Ensure that the training set and validation set are consistent in data distribution to avoid data skew.

[0177] 4. Model construction and training

[0178] 4.1 Building a CNN Model

[0179] According to the task requirements, design a suitable CNN model structure, including input layer, convolution layer, pooling layer, fully connected layer, etc.

[0180] Choose appropriate activation function, loss function, and optimization algorithm.

[0181] 4.2 Model Training

[0182] The training set is input into the CNN model for feature training and learning.

[0183] By iteratively optimizing the algorithm and continuously adjusting the model parameters, the performance of the model on the training set is gradually improved.

[0184] Monitor indicators such as loss function and accuracy during training to ensure the effectiveness of model training.

[0185] 5. Model validation and optimization

[0186] 5.1 Model Validation

[0187] The validation set is used to validate the trained CNN model and evaluate the recognition and prediction performance of the model.

[0188] Calculate the accuracy, recall, F1 score and other indicators on the validation set to measure the performance of the model. The administrator can verify it.

[0189] 5.2 Model Optimization

[0190] If the verification result is not ideal, analyze the problems of the model, such as overfitting, underfitting, etc.

[0191] Adjust the model structure, increase the amount of data, change hyperparameters, etc. to optimize the model.

[0192] Repeat the training-validation-optimization process until the model performance meets the expected requirements.

[0193] 6. Model deployment and application

[0194] 6.1 Model Deployment

[0195] The optimized CNN model was deployed on the background server to build an AI pre-diagnosis system for adenoids facial features.

[0196] Configure the server environment to ensure that the model can run stably.

[0197] 6.2 Tuning Participating Applications

[0198] Fine-tune the model according to actual application requirements, such as adjusting the threshold and optimizing the inference speed.

[0199] The model is connected to the front-end interface of the medical system to provide medical staff with adenoid facial pre-diagnosis services.

[0200] Monitor the performance of the model in actual applications, collect feedback and optimize it in a timely manner.

[0201] 7. Continuous iteration and update

[0202] New adenoid facial image data is collected regularly to update and iterate the model.

[0203] Focus on new technologies and methods in the medical field, continuously optimize model structure and algorithms, and improve model performance and accuracy.

[0204] Through the above detailed steps, an efficient and accurate adenoid facial intelligent pre-diagnosis model can be constructed to provide strong support for the medical field.

[0205] The model file can be loaded and deployed on the background for facial feature recognition and pre-diagnosis in the corresponding pediatric department to assist doctors in diagnosis.

[0206] S4, writing the facial image features and the corresponding symptom levels and clinical diagnosis and treatment strategy keywords into the user's electronic medical record file;

[0207] The user's identification results and the associated words of the recommended treatment strategy are promptly written into the file for easy review by doctors and parents.

[0208] To write facial image features and corresponding symptom severity and clinical diagnosis and treatment strategy keywords into the user's electronic medical record, you can follow the following steps:

[0209] First, it is necessary to ensure that the facial image features have been extracted through the CNN model or other related technologies, and the corresponding symptom severity and clinical diagnosis and treatment strategy keywords have been determined. These key information are the basis for writing into the electronic medical record.

[0210] Next, the specific steps are as follows:

[0211] ‌Data Arrangement‌:

[0212] The extracted facial image features, symptom severity, and clinical diagnosis and treatment strategy keywords are sorted out to ensure the accuracy and completeness of the data.

[0213] These data may exist in structured formats (such as JSON, XML, etc.) or unstructured text form, and need to be selected and organized according to actual conditions.

[0214] ‌Electronic medical record system access‌:

[0215] Determine how the user will access the electronic medical record system, which may be achieved through an API interface, database connection, or other means.

[0216] Ensure that communications with electronic medical record systems are secure, reliable, and comply with relevant medical data security and privacy protection regulations.

[0217] ‌Data writing‌:

[0218] According to the data model and requirements of the electronic medical record system, the sorted data is written into the corresponding fields.

[0219] This may include storing facial image features as binary data or image links, storing symptom severity and clinical treatment strategy keywords in text form, etc.

[0220] ‌Data verification and storage‌:

[0221] After the data is written, necessary data verification is performed to ensure the accuracy and consistency of the data.

[0222] After confirmation, the data will be saved in the electronic medical record system for subsequent medical services and research.

[0223] ‌Record and Track‌:

[0224] Record the time when data is written, operator and other information in the electronic medical record system to facilitate data tracking and auditing.

[0225] This helps ensure the integrity and traceability of data, which is of great significance for improving medical quality and handling medical disputes.

[0226] Through the above steps, facial image features, corresponding symptom severity and clinical diagnosis and treatment strategy keywords can be accurately written into the user's electronic medical record, providing strong support for the provision of medical services and the development of medical research.

[0227] S5. Generate a first adenoids facial prognosis plan corresponding to the symptom level according to the clinical diagnosis and treatment strategy keywords; and retrieve a corresponding second adenoids facial prognosis plan from a medical case library based on the symptom level and the facial image features through a large language model; and integrate the first adenoids facial prognosis plan and the second adenoids facial prognosis plan and write them into the user's electronic medical record file;

[0228] This department can use a large language model (which can be called by requesting a third-party large language model from the backend, such as Baidu Wenxin Yiyan) to assist in retrieving adenoid facial pre-diagnosis plans that meet the corresponding clinical diagnosis and treatment strategy keywords and symptom levels (because patients of the same type and degree in history have similar diagnosis plans in the database, so by looking at the data of the corresponding symptom levels and historical treatment strategy keywords, it is possible to retrieve and output treatment plans of similar symptom levels, that is, pre-diagnosis plans). Specifically:

[0229] ‌First, generate the first adenoid facial pre-diagnosis plan based on the clinical diagnosis and treatment strategy keywords‌.

[0230] Based on the degree of symptoms extracted from the patient's facial image features (such as the degree of adenoid hypertrophy, whether there is open mouth breathing, upturned upper lip and other adenoid facial features) and the corresponding clinical diagnosis and treatment strategy keywords (such as drug treatment, surgical treatment, etc.), a preliminary adenoid facial pre-diagnosis plan for the patient's symptoms can be formulated. This plan should list in detail the possible diagnosis, recommended treatment measures and subsequent follow-up plans.

[0231] ‌Secondly, the large language model is used to retrieve the second adenoid facial prognosis plan‌.

[0232] The patient's symptom severity and facial image features are used as input and retrieved from the medical case library through a large language model. The large language model can understand and process natural language text, so it can accurately find cases similar to the patient's situation from a large number of medical cases and extract the diagnosis and treatment strategies used in these cases, thereby forming a second adenoid facial pre-diagnosis plan. This plan should include best practices, treatment recommendations, and possible prognoses learned from similar cases.

[0233] ‌Finally, the two pre-diagnosis plans are integrated and written into the user's electronic medical record. Specifically, the first adenoids facial pre-diagnosis plan and the second adenoids facial pre-diagnosis plan are integrated using a large language model, the keyword similarity in the two plans is calculated, and the strategy is integrated. The method is as follows:

[0234] First, the two pre-diagnosis plans are text-processed separately to ensure that they are both presented in the form of natural language text to facilitate processing by the large language model.

[0235] Next, the text understanding capability of the large language model is used to extract keywords from the two solutions. This step can be achieved through named entity recognition (NER) or keyword extraction technology in the large language model to identify key medical terms, symptom descriptions, and treatment recommendations involved in the two solutions.

[0236] ‌Then, calculate the similarity of the keywords in the two solutions‌. This step can be achieved by using text similarity calculation methods such as cosine similarity, Jaccard similarity, or edit distance. Specifically, the keywords in the two solutions can be represented as vectors, and then the similarity between these vectors can be calculated. The higher the similarity, the greater the overlap between the two solutions at the keyword level, and the higher the possibility of fusion‌.

[0237] ‌Based on the similarity of keywords, the two pre-diagnosis plans are merged. The fusion method can be weighted average, union or intersection, etc., depending on the size of the similarity and the needs of the actual application. For example, if the keyword similarity of the two plans is very high, it means that they have a great consensus on diagnosis and treatment recommendations. At this time, the union method can be adopted to retain the diagnosis and treatment recommendations in the two plans; if the similarity is low, it means that the two plans have differences in some aspects. At this time, weighted average or intersection operations can be performed according to the actual situation to obtain a more comprehensive and accurate pre-diagnosis plan.‌

[0238] ‌Finally, the text generation capability of the large language model is used to present the fused pre-diagnosis plan in the form of natural language text‌. This step can be achieved through the text generation technology in the large language model, thereby generating a more comprehensive and accurate adenoid facial pre-diagnosis plan that contains both useful information of the first pre-diagnosis plan and useful information of the second pre-diagnosis plan‌.

[0239] Through the above steps, the first adenoids facial pre-diagnosis scheme and the second adenoids facial pre-diagnosis scheme can be integrated using a large language model, and the keyword similarity in the two schemes can be calculated for fusion.

[0240] The first adenoids facial pre-diagnosis plan and the second adenoids facial pre-diagnosis plan are integrated, and the diagnosis, treatment recommendations and prognosis evaluation in the two plans are comprehensively considered to form a more comprehensive and personalized pre-diagnosis plan. Then, this integrated pre-diagnosis plan is written into the patient's electronic medical record file for the doctor to refer to and use in the subsequent diagnosis and treatment process. At the same time, the security and privacy of the electronic medical record file should also be ensured to protect the patient's personal information from being leaked.

[0241] Through the above steps, the described plan can be implemented to provide patients with more accurate and personalized adenoid facial pre-diagnosis services.

[0242] S6. Regularly send the user's electronic medical record to the user end and / or the medical end.

[0243] The background can push the results to parents or doctors according to preset push rules, such as timely push or monthly push, so as to promptly intervene in the screening and pre-diagnosis of adenoids appearance of children and other users.

[0244] Figure 4 1 is a block diagram of an intelligent health management system for patients with glandular hypertrophy according to an exemplary embodiment, and the system is used for an intelligent health management method for patients with glandular hypertrophy. Figure 4 The system comprises:

[0245] The user end is used to collect the user's facial image and upload it to the backend server;

[0246] Backend server, including:

[0247] Hospital information system, used to generate and manage users' electronic medical records;

[0248] A contour recognition module, used for recognizing and extracting a facial contour image from the facial image, and recording the facial contour image into a user's electronic medical record file in a hospital information system;

[0249] Adenoids facial AI pre-diagnosis module, used for inputting the facial contour image into a pre-deployed adenoids facial intelligent pre-diagnosis model, and the adenoids facial intelligent pre-diagnosis model identifies the facial image features in the facial contour image and outputs the symptom degree and clinical diagnosis and treatment strategy keywords related to the facial image features;

[0250] A prediction result recording module, used for writing the facial image features and the corresponding symptom levels and clinical diagnosis and treatment strategy keywords into the user's electronic medical record file;

[0251] A strategy generation module, for generating a first adenoids facial prognosis plan corresponding to the symptom degree according to the clinical diagnosis and treatment strategy keywords; and for retrieving a corresponding second adenoids facial prognosis plan from a medical case library based on the symptom degree and the facial image features through a large language model;

[0252] A strategy integration module, used for integrating the first adenoids facial pre-diagnosis plan and the second adenoids facial pre-diagnosis plan and writing them into the user's electronic medical record file;

[0253] A recommendation module, used for regularly sending the user's electronic medical record to the user terminal and / or the medical terminal;

[0254] The medical terminal is used to log in and view the electronic medical records of the user;

[0255] The user end and the medical end are respectively connected to the background server for communication.

[0256] Please understand the functions of this system and the interactions between the subjects by referring to the corresponding step descriptions in the above method, which will not be repeated here.

[0257] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable system. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0258] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0259] In the present invention, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0260] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0261] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0262] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0263] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, systems and methods can be implemented in other ways. For example, the system 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 integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of the system or unit, which can be electrical, mechanical or other forms.

[0264] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, 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 solution of this embodiment.

[0265] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0266] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0267] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. An intelligent health management method for patients with glandular hypertrophy, characterized in that: The method comprises: S1, collect the user's facial image through the user-side APP and upload it to the backend server; S2, the backend server identifies and extracts the facial contour image in the facial image, and records the facial contour image into the user's electronic medical record file in the HIS system; S3, simultaneously inputting the facial contour image into a pre-deployed adenoid facial AI pre-diagnosis model, and having the adenoid facial AI pre-diagnosis model recognize facial image features in the facial contour image and output symptom levels and clinical diagnosis and treatment strategy keywords related to the facial image features; S4, writing the facial image features and the corresponding symptom levels and clinical diagnosis and treatment strategy keywords into the user's electronic medical record file; S5. Generate a first adenoids facial prognosis plan corresponding to the symptom level according to the clinical diagnosis and treatment strategy keywords; and retrieve a corresponding second adenoids facial prognosis plan from a medical case library based on the symptom level and the facial image features through the LLM large language model; and integrate the first adenoids facial prognosis plan and the second adenoids facial prognosis plan and write them into the user's electronic medical record file; S6. Regularly send the user's electronic medical record to the user APP and / or medical care end.

2. The intelligent health management method for patients with glandular hypertrophy according to claim 1, characterized in that: The method for generating the adenoid facial AI pre-diagnosis model comprises: Collect and preprocess image datasets of faces with different degrees of adenoids; Performing feature engineering on the image dataset, using CNN to identify and extract facial image features on each adenoid facial image in the image dataset, including facial contour, adenoid shape and size, lip and tooth shape, and jaw / palate shape; Performing feature annotation on the facial image features, annotating corresponding symptom levels and clinical diagnosis and treatment strategy keywords, wherein the symptom levels include prompt information on whether the facial expression is lacking as determined based on the facial image features; Counting the features of each facial image to obtain a feature set; Dividing the feature set into a training set and a validation set according to a preset ratio; Inputting the training set into a preset CNN model for feature training and learning, thereby constructing the adenoid facial AI pre-diagnosis model; The recognition and prediction performance of the adenoid facial AI pre-diagnosis model was verified using the delivery validation set: If the verification is passed, the adenoids facial AI pre-diagnosis model is deployed on the background server and put into the application service of adenoids facial pre-diagnosis after adjusting the parameters; If the verification fails, repeat the above steps to reconstruct the adenoid facial AI pre-diagnosis model.

3. The intelligent health management method for patients with glandular hypertrophy according to claim 2, characterized in that: When collecting image datasets of different degrees of adenoids, the image datasets of different degrees of adenoids include: frontal face images; frontal smiling face images; 45° image of the left side of the face and 45° image of the right side of the face; 90° image of the left side of the face and 90° image of the right side.

4. An intelligent health management system for patients with glandular hypertrophy, the intelligent health management system for patients with glandular hypertrophy is used to implement the intelligent health management method for patients with glandular hypertrophy as claimed in any one of claims 1 to 3, characterized in that: The system comprises: The user-side APP is used to collect the user's facial image and upload it to the backend server; Backend server, including: HIS system, used to generate and manage users' electronic medical records; A contour recognition module, used for recognizing and extracting a facial contour image in the facial image, and recording the facial contour image into the user's electronic medical record file in the HIS system; An adenoids facial AI pre-diagnosis module, used for inputting the facial contour image into a pre-deployed adenoids facial AI pre-diagnosis model, and the adenoids facial AI pre-diagnosis model identifies the facial image features in the facial contour image and outputs the symptom degree and clinical diagnosis and treatment strategy keywords related to the facial image features; A prediction result recording module, used for writing the facial image features and the corresponding symptom levels and clinical diagnosis and treatment strategy keywords into the user's electronic medical record file; A strategy generation module is used to generate a first adenoid facial prognosis plan corresponding to the symptom degree according to the clinical diagnosis and treatment strategy keywords; and retrieve a corresponding second adenoid facial prognosis plan from a medical case library based on the symptom degree and the facial image features through an LLM large language model; A strategy integration module, used for integrating the first adenoids facial pre-diagnosis plan and the second adenoids facial pre-diagnosis plan and writing them into the user's electronic medical record file; A recommendation module, used to regularly send the user's electronic medical record files to the user-side APP and / or the medical care side; The medical terminal is used to log in and view the electronic medical records of the user; The user-side APP and the medical-care side are respectively connected to the background server for communication.

5. An electronic device, characterized in that: The electronic device comprises: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 3 is implemented.

6. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program codes, which can be called by a processor to execute the method according to any one of claims 1 to 3.

Images