Intelligent interaction system and method for oral health management

Abstract

The invention relates to the technical field of intelligent health management, and discloses an intelligent interaction system and method for oral health management, an intelligent shooting guide unit is used for selecting an intraoral shooting situation, generating a dynamic viewing frame and a verification prompt, generating information used for guiding shooting of a standard intraoral image, and optimizing the intraoral image; the multi-dimensional disease risk analysis unit is used for extracting high-dimensional disease features of the intraoral image, performing multi-omics data fusion based on deep learning and outputting a structured oral disease risk judgment result; the personalized interaction unit carries out result interpretation according to the risk judgment result, generates a disease intervention interaction task, pushes scenarized knowledge and generates a dynamic health card; and the full-period health archive unit is used for performing time sequence archive analysis on key output information of each unit and performing data interaction with an external health management system. According to the invention, image guide acquisition, intelligent risk analysis, personalized interactive feedback and full-period health data management are effectively integrated, and integrated oral health management is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent health management, in particular to an intelligent interaction system and method for oral health management. BACKGROUND

[0002] In the current field of oral health management, the conventional technical system mainly covers traditional clinical diagnosis methods, basic saliva detection tools and simple health management application programs. However, these methods have obvious limitations in terms of popularity, accuracy and continuous management capability.

[0003] Traditional oral disease diagnosis mainly relies on manual operations such as visual inspection and palpation by doctors, and must be completed in professional medical institutions. Not only is the cost of single examination high and the overall process time-consuming, but also has a high requirement for the distribution of medical resources. In addition, the traditional method has limited sensitivity in identifying early asymptomatic lesions, and a large number of hidden oral problems develop into severe cases due to the failure to detect them in time.

[0004] In terms of detection technology, the saliva detection products on the market are mostly limited to single biomarker analysis, and have not integrated multi-omics data such as genome and proteome. The recognition sensitivity of early periodontitis and other diseases is significantly lower than that of traditional methods. At the same time, the sampling device is complicated to operate and relies on manual interpretation of results, which makes it difficult to adapt to home testing scenarios, limiting its popular application. In terms of digital management, the existing oral health management APPs mostly have single functions, focusing on basic data recording or general knowledge pushing, and lack effective shooting guidance mechanisms, resulting in unstable quality of user self-shot images and high AI analysis misjudgment rate. In addition, the detection data is usually scattered in the phone album or paper documents, and cannot form a continuous and traceable health record, making it difficult for doctors to assess the disease dynamics. The weak user interaction mechanism and the lack of personalized design of the suggestion content lead to low detection initiative and management compliance of teenagers and the elderly.

[0005] From the overall industry, the domestic market has not yet formed a complete precision screening system technical solution. The existing system products have obvious shortcomings in data quality guarantee and user long-term participation maintenance in the home scenario, and it is difficult to meet the goal of early screening of chronic diseases. SUMMARY

[0006] The present application aims to provide an intelligent interaction system and method for oral health management, which can effectively integrate image-guided acquisition, intelligent risk analysis, personalized interactive feedback and long-term health data management to realize integrated oral health management, in order to fill the technical gap of current technical system in home applicability, data continuity and user participation.

[0007] The basic scheme provided by the present application is: an intelligent interaction system for oral health management, comprising: The intelligent photographing guiding unit comprises a photographing scene decision module for selecting an intraoral photographing scenario, a real-time viewfinder module for generating a dynamic viewfinder and a verification prompt, a multi-modal prompt module for generating information used for guiding photographing of a standard intraoral image, and an image preprocessing module for intraoral image optimization; The multi-dimensional disease risk analysis unit comprises an image feature extraction module for extracting high-dimensional disease features from the intraoral image, a multi-omics data fusion module for fusing non-image oral health data and high-dimensional disease features, and a risk assessment module for outputting a structured oral disease risk determination result according to the fusion result. The personalized interactive unit comprises a risk interpretation module for interpreting the risk determination result, a task challenge module for generating a disease intervention interactive task, a health knowledge push module for pushing scenario-based knowledge, and a social sharing module for generating a dynamic health card. The whole-cycle health record unit comprises a time axis archiving module for performing time sequence archiving analysis on key output information of the intelligent photographing guiding unit, the multi-dimensional disease risk analysis unit and the personalized interactive unit, and a data linkage module for interacting with external health management system data.

[0008] The application also provides an intelligent interactive method for oral health management, which utilizes an intelligent interactive system for oral health management. S1, a user selects an intraoral photographing scenario, and photographs a standard intraoral image according to a dynamic viewfinder, a verification prompt and guiding information; after preprocessing the standard intraoral image, the user is prompted that the photographing is completed; S2, high-dimensional disease features are extracted from the intraoral image; non-image oral health data is inputted, and risk assessment is performed by fusing the high-dimensional disease features, to output a structured oral disease risk determination result; S3, at least one of the following is performed: interpreting the risk determination result, generating a disease intervention interactive task, pushing scenario-based knowledge, and generating a dynamic health card; S4, time sequence archiving analysis is performed on key output information in the processes of S1-S3; S5, data interaction is performed with an external health management system.

[0009] The working principle and advantages of the application are as follows: The system adopts a four-layer closed-loop architecture of intelligent photographing guidance, multi-dimensional disease risk analysis, personalized interactive feedback and whole-cycle health archiving, effectively realizes high-quality collection, accurate identification and continuous management of oral health data in a family scenario by constructing a standardized photographing assistance mechanism and an interactive health management process.

[0010] In the data acquisition link, the system adopts intelligent shooting guidance technology, through dynamic framing, multi-modal prompting and real-time verification mechanism, plus image rapid optimization, which significantly improves the quality of non-professional user selfie images. After community user group testing, the effective area of intraoral photography of the guided shooting reaches 92%, the light uniformity is improved by 40%, providing high-quality data for subsequent disease recognition, and the misjudgment rate is reduced from 15% to less than 8%.

[0011] In the recognition and analysis layer, the system integrates at least 512-dimensional image features and multi-omics data to construct a high-dimensional feature fusion model, which improves the recognition sensitivity of early periodontitis and other occult diseases by 40%, and the overall disease risk assessment accuracy is significantly better than that of traditional single modal detection method.

[0012] In the user interaction and health management dimension, the system automatically generates interpretation content and task challenges based on the risk determination results, and converts health suggestions into executable interactive tasks, effectively enhancing user participation and compliance. Actual application data shows that the user task completion rate is improved by more than 50%, the health knowledge awareness rate is improved from less than 35% to 80%, and the active detection willingness of key groups such as teenagers is also improved by about 45%.

[0013] In addition, the system constructs a complete time sequence health record, supports one-key backtracking and cross-platform interaction of detection data and intervention records, and the data query efficiency is improved by more than 70%, doctors can quickly grasp the patient's whole health changes through the system before seeing a doctor, effectively shortening the preparation time for diagnosis and treatment, and improving the efficiency of medical services.

[0014] In terms of economic and social benefits, the system provides a feasible path to solve the problems of poor quality of family self-test data and low user stickiness, and is expected to promote the proportion of active detection of oral diseases in China from less than 1% to more than 10%, promote early screening and intervention, and reduce the medical burden of patients from mild to severe. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 The structure diagram of the intelligent interactive system for oral health management provided by the embodiment of the application; Figure 2 The interface diagram of the intelligent shooting guidance unit provided by the embodiment of the application; Figure 3 The risk interpretation result diagram provided by the embodiment of the application; Figure 4 The flowchart of the intelligent interactive method for oral health management provided by the embodiment of the application. DETAILED DESCRIPTION

[0016] The following detailed explanation illustrates the specific implementation methods: The basic implementation examples are as follows: Figure 1 As shown: An intelligent interactive system for oral health management, comprising: The intelligent shooting guidance unit includes a shooting scene decision module for selecting intraoral shooting scenarios, a real-time framing module for generating dynamic framing frames and verification prompts, a multimodal prompting module for generating information used to guide the shooting of standard intraoral images, and an image preprocessing module for optimizing intraoral images. The multidimensional disease risk analysis unit includes an image feature extraction module for extracting high-dimensional disease features from intraoral images, a multi-omics data fusion module for fusing non-image-based oral health data and high-dimensional disease features, and a risk assessment module for outputting structured oral disease risk determination results based on the fusion results. Personalized interactive units include a risk interpretation module that interprets the results based on the risk assessment results, a task challenge module that generates interactive disease intervention tasks, a health knowledge push module that pushes contextualized knowledge, and a social sharing module that generates dynamic health cards. The full-cycle health record unit includes a timeline archiving module for performing time-series archiving analysis of key output information from the intelligent shooting guidance unit, the multi-dimensional disease risk analysis unit, and the personalized interaction unit, as well as a data linkage module for data interaction with external health management systems.

[0017] Specifically: The intelligent shooting guidance unit includes a shooting scene decision module, a real-time viewfinder module, a multimodal prompting module, and an image preprocessing module.

[0018] The shooting scene decision module is used to select the intraoral shooting context. Shooting scenes can be defined in a structured manner, including behavioral categories and time information. Behavioral categories include eating behaviors (such as eating, drinking water or beverages) and cleaning behaviors (such as brushing teeth, using dental floss, rinsing mouth). Figure 2 As shown, in this embodiment, the shooting scenarios include eating - breakfast (which can be understood as shooting after breakfast), eating - lunch, eating - dinner; brushing teeth - morning (which can be understood as shooting after brushing teeth in the morning), brushing teeth - noon, brushing teeth - evening. Other embodiments can be adjusted according to actual needs. Since the type of oral intervention or exposure event plays an important role in the identification of oral diseases, and the oral microenvironment (such as saliva flow rate, bacterial activity, and food debris type) varies significantly at different time periods, selecting the shooting scenario before shooting facilitates the provision of differentiated analysis strategies for oral disease risk identification based on different scenarios. It also supports personalized feedback: for example, the system can push stronger intervention suggestions based on the "dinner + brushing teeth" scenario.

[0019] A real-time framing module for generating a dynamic framing and a verification prompt. The dynamic framing is generated by using an AI real-time image segmentation technology, and meets the requirement of covering the key areas of the user's teeth, including the occlusal surface, buccal surface and lingual surface. The real-time framing module is also used for a dynamic angle verification prompt, which detects the user's shooting angle in real time and calculates the deviation, and outputs a verification prompt when the deviation exceeds a threshold. The verification prompt information includes a dynamic framing body change and specific adjustment behavior information prompt, and the specific adjustment behavior information includes adjustment direction, adjustment angle and target shooting area. For example, when the user's shooting angle deviation exceeds 5°, the framing automatically turns red and displays an adjustment prompt (and voice): "Please raise the phone by 10° to shoot the occlusal surface". This way can ensure that the image covers the key detection area, and the effective area ratio is ≥90%.

[0020] A multi-modal prompt module for generating information for guiding the user to shoot standard intraoral images, that is, analyzing and processing the image in the current dynamic framing to output guiding information. The guiding information includes a voice prompt (such as "Please keep the light sufficient, avoid shadow obstruction") and a graphical annotation, which includes a screen display of a tooth contour schematic diagram to guide the user to complete the shooting of three standard intraoral images (occlusal surface, maxillary buccal surface and mandibular buccal surface), and the shooting time is ≤2 minutes, the image clarity is ≥1080P, and the light uniformity is improved by 40%. In addition, the graphical annotation can also distinguish the target shooting area (occlusal surface, maxillary buccal surface and mandibular buccal surface) on the tooth contour schematic diagram.

[0021] It should be noted that the real-time framing module focuses on real-time quality verification and fine tuning during shooting. Function: Based on AI segmentation, it determines whether the current picture meets the standard when the user is shooting. Content: Dynamically display a frame (generated by AI in real time) that can cover the key areas of the user's teeth. If the angle / coverage is insufficient, give precise adjustment instructions (such as "turn a little bit to the left"). Only intervene when verification fails to avoid constant interference. The multi-modal prompt module focuses on process guidance before and during shooting. Function: Tell the user "which one to shoot next" and "general shooting conditions". Content: Voice: "Please shoot the occlusal surface now, please lightly bite the upper and lower teeth." Graphics: User's real-time tooth contour; Prompt: "Keep the light sufficient, don't block it with your hand." It does not involve real-time image analysis, but only provides preset guidance.

[0022] An image preprocessing module for intraoral image optimization, which automatically crops redundant backgrounds such as hands and lips (with a cropping accuracy of 95%), optimizes backlight and dark light images through a brightness equalization algorithm, and the preprocessing time is <3 seconds, providing high-quality data for subsequent disease recognition.

[0023] A multi-dimensional disease risk analysis unit, including an image feature extraction module, a multi-omics data fusion module and a risk scoring module. An image feature extraction module is configured to extract high-dimensional disease features of intraoral images. Based on a ResNet50 network, disease features such as dental caries (e.g., enamel demineralization white spots) and gingivitis (e.g., swollen gums) in intraoral images are extracted, and a 512-dimensional feature vector (high-dimensional disease feature) is output, with an accuracy of feature extraction of greater than or equal to 88%.

[0024] A multi-omics data fusion module is configured to fuse non-image oral health data and high-dimensional disease features based on deep learning. The non-image oral health data includes saliva detection data (e.g., Streptococcus mutans concentration, inflammatory factor level), and the fusion is performed through Figure 2 a saliva test paper colorimetric input. Through an attention mechanism, the weights of the non-image oral health data and the high-dimensional disease features are adjusted based on the correlation degree data, and higher weights are given to the high-correlation data (e.g., saliva data accounts for 40% of the weight of the gingivitis risk). After fusion, the disease recognition accuracy is improved by 15%.

[0025] A risk assessment module is configured to output a structured oral disease risk determination result based on the fusion result. Through Figure 2 an "analysis" to start the evaluation, and through "viewing" to display the result. This module inputs a pre-trained risk assessment model after multi-omics data fusion, combines a rule engine to determine a high-risk threshold (greater than or equal to 70 points), and traces back the feature weights to automatically label the dominant risk inducement. The risk determination result includes the oral disease and its risk score, the risk level (0-100 points, greater than or equal to 70 points is high risk), and the risk inducement (e.g., "dental caries risk 82 points, due to Streptococcus mutans concentration exceeding 2 times"). A single analysis takes less than 30 seconds, and a total accuracy of 87.6% is achieved through verification of 1,200 samples from a dental hospital. The sensitivity of early periodontitis detection is improved by 40% compared to traditional methods.

[0026] A personalized interactive unit includes a risk interpretation module, a task challenge module, a health knowledge push module, and a social sharing module. Through Figure 2 the "analysis" to start the evaluation at the same time, and through "viewing" to enter the interaction.

[0027] A risk interpretation module is configured to interpret the risk assessment result. This module automatically converts structured disease risk indicators into user-oriented, popular, and causal health interpretation texts based on a mapping mechanism of a pre-set medical knowledge graph and a natural language generation (NLG) template, explains the risk result in popular language (e.g., "high risk of gingivitis may cause gum bleeding, and oral cleaning should be strengthened"), avoids the accumulation of professional terms, and improves user understanding by 60%. As Figure 3 shown.

[0028] A task challenge module is configured to generate a disease intervention interactive task according to the risk determination result. The module dynamically matches and generates differentiated and executable interactive intervention tasks, such as “complete 3-day tooth brushing check-in” for a low-risk user and “make a dental examination appointment within 1 week” for a high-risk user, by using a preset risk level-task mapping rule library in combination with a user portrait and a behavior context. The user can unlock points (10 points / task) and health medals (such as “oral health guard”) by completing the task, and the user's task completion rate is improved by 50%. The task challenge module further includes a family health competition sub-module, which introduces a “family health competition” mode (such as competing in tooth brushing check-in days among family members), enhances social attributes, and improves the stickiness of family users by 40%, which is suitable for family health management scenarios.

[0029] A health knowledge push module is configured to push scenario-based knowledge according to the risk determination result. The module matches corresponding scenario-based knowledge points based on the disease type in the user risk determination result by constructing an association mapping table of oral disease types and scenario-based health knowledge, and generates personalized health education content in the form of a comic or a short video by calling a preset multimedia template library to push, such as pushing “relationship between sweet food intake and dental caries” for a high dental caries risk, which can be presented in the form of a comic or a short video, and the knowledge acceptance of teenagers is improved by 45%.

[0030] A social sharing module is configured to generate a dynamic health card according to the risk assessment result. The module dynamically renders structured disease types, risk scores, health suggestions, and timestamps into an interactive health card containing a system entry, and integrates a third-party social platform SDK to realize one-key sharing (supporting one-key sharing to platforms such as WeChat and Weibo) to promote the socialization of health management content.

[0031] A whole-cycle health record unit includes a time sequence archiving analysis module and a data linkage module, and further includes a data security module.

[0032] The time sequence archiving analysis module is configured to perform time sequence archiving analysis on key output information of the intelligent shooting guide unit, the multi-dimensional disease risk analysis unit, and the personalized interaction unit. The module performs time sequence archiving on structured health output information according to detection dates, dynamically generates a multi-dimensional health trend graph based on disease types and time range screening conditions by calling a visualization engine. The key output information includes intraoral images, risk scores, and interaction task completion conditions. The key generated information in a statistical analysis time period is analyzed to generate a visual health trend graph (such as a “dental caries risk change trend curve in the past 6 months”), and data query efficiency is improved by 70% by supporting data screening according to disease types and time ranges.

[0033] The time sequence archiving analysis module further includes an audit sub-module configured to adjust the archived information, and a report generation sub-module configured to generate a corresponding health report based on the archived analysis data.

[0034] The data linkage module is used for data interaction with external health management systems. This module achieves secure integration with external health management systems (such as hospital electronic medical record systems and community health platforms) through standardized medical data interfaces and a tiered authorization mechanism. It also supports users in setting differentiated data access permissions for doctors and family members, collaboratively supporting clinical diagnosis and treatment as well as family health management. For example, doctors can view a user's historical test data to develop more precise treatment plans, and users can authorize family members to view their records for collaborative family health management.

[0035] Data security module: Data is stored using the AES-256 encryption algorithm and access is only permitted after user authorization, complying with the requirements of the Personal Information Protection Law, reducing the risk of data leakage to below 0.1%. For research institutions, it provides a data anonymization and export function (removing personal privacy information), supporting CSV / Excel format export, facilitating researchers to conduct oral health big data research, and increasing data utilization by 35%.

[0036] like Figure 4 As shown, this embodiment also provides an intelligent interactive method for oral health management, utilizing an intelligent interactive system for oral health management; the method includes: S1, the user selects the intraoral shooting scenario and takes 3 standard intraoral images according to the dynamic viewfinder, verification prompts and guidance information; after preprocessing the standard intraoral images, the user is prompted that the shooting is complete; S2 extracts high-dimensional disease features from intraoral images; inputs non-image-based oral health data, integrates high-dimensional disease features for risk assessment, and outputs structured oral disease risk determination results; results are displayed within 30 seconds. S3 involves at least one of the following: interpreting the risk assessment results, generating interactive disease intervention tasks, pushing contextualized knowledge, and generating dynamic health cards. S4, perform time-series archiving analysis on the key output information in the S1-S3 process; S5 interacts with external health management systems for data exchange.

[0037] Specifically, in practical applications (home version system (mini-program deployment)): 1) System parameter configuration Compatible devices: iOS 12.0+ and Android 8.0+ phones, supporting front / rear cameras (resolution ≥1080P); Core modules: Shooting guidance (real-time framing, voice prompts), disease recognition (lightweight ResNet50 model, 50MB in size), interactive feedback (task challenges, points badges), data storage (cloud encrypted storage, 10GB of free storage space). Response time: Pre-processing of images <3 seconds, disease analysis <30 seconds, data archiving <5 seconds.

[0038] 2) Operating steps (1) Users register and fill in basic information (age, history of oral diseases); (2) Select “Oral Examination” and take 3 intraoral photos (occlusal surface, maxillary buccal surface, and mandibular buccal surface) according to the real-time viewfinder and voice prompts. (3) You can optionally upload the saliva test results (such as a photo of the test strip), and the system will output the risk score and cause within 30 seconds; (4) View the risk interpretation and interactive tasks (such as "3-day brushing check-in"), and earn 10 points for completing the task; (5) Generate a health card and share it to the family group. The system will automatically archive the data to the timeline.

[0039] 3) Validate the data Test population: 300 residents from the community surrounding a dental hospital (aged 20-75, including 150 patients with oral diseases); Results: Image effectiveness rate was 92%, disease identification accuracy rate was 87.6%, user task completion rate was 50%, and satisfaction score was 4.8 out of 5.

[0040] Specifically, in practical applications (community healthcare institution version of the system): 1) System parameter configuration (1) Deployment method: Community hospital intranet server + tablet terminal, supporting 10 terminals online at the same time; (2) New features: Batch detection (import data of 20 people at a time), doctor review channel (review time < 2 minutes), community health report generation (statistical distribution of diseases by region); (3) Data linkage: Connect to the community health management platform to support the synchronization of residents' health records.

[0041] 2) Operating steps (1) Community doctors use tablet terminals to take intraoral photos of residents (system guidance optimization); (2) The system automatically analyzes and generates an initial screening report, which is then reviewed by the doctor who provides supplementary suggestions. (3) Generate personalized tasks for residents (such as "follow-up examination in 1 week"), and residents can scan the code to view and complete them; (4) The system generates community oral health reports quarterly to provide a basis for public health decision-making.

[0042] 3) Application effect Community coverage: Three community health service centers in the High-tech Zone are piloting the program, serving 5,000 residents; Efficiency improvement: community screening efficiency increased from 20 people per day to 50 people per day, and doctor review time shortened by 60%.

[0043] The comparison results of the system with the prior art are shown in Table 1: Table 1 Comparison of the system with the prior art

[0044] Example two Different from example one, the intelligent shooting guide unit further includes a user portrait recognition module for adjusting the old-age exclusive guide mode. The specific process is as follows: The user portrait recognition submodule is used to identify whether the user belongs to the old-age group (such as age ≥ 65 years old or enable the elder mode). When it is determined that the user is old, the system automatically switches the guide strategy, including disabling the dynamic angle verification prompt and color feedback mechanism of the real-time viewfinder module, and instead outputting a step-by-step static example guide sequence from the multi-modal prompt module, that is, sequentially displaying standard intraoral image examples (such as occlusal surface and buccal surface) with high contrast and large size, each with a ≥ 28pt bold font action instruction (such as "please take the back teeth like this"), and supporting manual triggering of voice reading. At the same time, the AI image segmentation capability of the real-time viewfinder module in the background is retained, which is only used to judge whether the key area is covered, and if it is not up to standard, a parallel comparison feedback of the user's actual shooting picture and the standard example is generated, supplemented by concrete adjustment prompts, such as using animation.

[0045] This optimization reduces the operation cognitive load without adding an independent evaluation module, significantly improves the shooting success rate and compliance of old-age users with limited vision and digital skills, and realizes inclusive oral health management.

[0046] Example three Different from examples one and two, the task challenge module includes a family collaborative intervention module, which is used to generate collaborative intervention tasks with role division and health causal relationship based on a pre-constructed family oral health coupling model according to the oral risk transmission path and behavior dependence relationship among members, and to encourage continuous participation through a family health contribution value feedback mechanism.

[0047] The specific application process includes: 1) Identify the user's home group (through account binding or device sharing), including member basic information, family role, cohabitation situation, etc. 2) Collect the oral risk assessment results of each member (such as high risk of childhood caries and high risk of old-age gingivitis).

[0048] 3) Construct a family oral health coupling model. This model is based on medical evidence and predefines the following association rules: If there are school-age children at home, their tooth brushing compliance is strongly related to parental supervision behavior; If the frequency of family meals is high, there is a risk of cross-infection of cariogenic bacteria (such as Streptococcus mutans) among family members; The oral cleaning habits of the elderly wearing dentures are easily affected by the frequency of reminders from their cohabiting children.

[0049] 4) Input the oral risk data of each member into the pre-trained family oral behavior coupling model to dynamically generate asymmetric, role-based, and causally linked interactive tasks, such as: Push to parents: "Your child's risk of dental caries is high, please accompany brushing for 3 consecutive nights and take photos to confirm, which can reduce their plaque index"; Push to teenagers: "Dad's risk of gingival bleeding is high, please remind him to use dental floss 2 times this week, and the whole family's oral inflammation risk will decrease"; Push to the whole family: "Please complete the post-meal gargling check-in for more than 4 times this week, to block the transmission of cariogenic bacteria".

[0050] The completion of the task can be visualized in the form of family health contribution value (not just ranking), and a family-specific oral health report can be unlocked (such as "this month, the family's plaque transmission risk has decreased by 32%").

[0051] The intelligent interactive system and method for oral health management provided in this embodiment deeply integrates the unique characteristics of oral health, such as bacterial transmission, behavior concealment, and intergenerational influence, introduces a family oral health coupling model, breaks through the traditional general health check-in mode, dynamically generates role-based collaborative tasks (such as "accompany brushing confirmation" and "post-meal gargling check-in for the whole family") by identifying the cross-relationships of dental caries or gingivitis risk among family members (such as the homology of cariogenic bacteria between parents and children, and the dependence of the elderly on reminders for cleaning), not only improves the depth of family interaction and emotional connection, but also improves the intervention compliance of high-risk members and reduces the overall oral health risk of the family, significantly enhancing the stickiness of family users and the effectiveness of health management.

[0052] Embodiment Four Unlike embodiments one, two, and three, the risk assessment module further includes a microenvironment imbalance determination sub-module, which further integrates time-series intraoral image features and behavior logs based on conventional risk assessment to construct a quantitative indicator reflecting the accelerating trend of oral ecological imbalance.

[0053] A plaque activity time-series module is constructed, which calculates the plaque activity based on the plaque optical feature change rate in the occlusal surface / gingival margin area in the last N (such as the last 3) intraoral images by the following formula:

[0054] wherein, ​is the plaque coverage intensity for the ith detection (output by the image feature extraction module, normalized to 0-1); is the interval time between two detections (days); N is the total number of detections.

[0055] The Oral Microenvironment Imbalance Index (OMII) is introduced to combine the plaque activity is fused with the current risk score R and the behavior compliance decay factor D (such as the number of days of interruption in tooth brushing check-in):

[0056] wherein, is a Sigmoid function that maps the activity to [0, 1]; , is a clinical calibration weight, and the sum is 1.

[0057] Dynamic intervention tasks are generated. When OMII ≥ threshold (such as 0.75) and shows an upward trend, the task challenge module does not push the regular check-in task, but generates an urgent intervention task that breaks the deterioration chain, such as: “Gingival inflammation is detected to be developing rapidly! Please use a chlorhexidine-containing mouthwash within 48 hours, and upload a new image for review.”

[0058] The intelligent interactive system and method for oral health management provided in the embodiment construct a deterioration trend prediction index (OMII) through plaque dynamic growth rate, risk score, and behavior decay, realize the transition from static risk score to dynamic deterioration trend early warning, and drive the generation of forward-looking intervention tasks.

[0059] The above only describes the embodiments of the present application, and the common knowledge of specific structures and characteristics in the scheme is not described in detail. The person skilled in the art knows all the ordinary technical knowledge in the field of the application before the application date or the priority date, can know all the prior art in the field, and has the ability to apply conventional experimental means before that date. The person skilled in the art can improve and implement the present scheme based on their own ability under the guidance of the present application, and some typical known structures or known methods should not be an obstacle for the person skilled in the art to implement the present application. It should be noted that for those skilled in the art, without departing from the structure of the present application, a number of modifications and improvements can be made, which should be considered as the protection scope of the present application, and these will not affect the effect and practicality of the patent.

Claims

1. An intelligent interactive system for oral health management, characterized by, Comprise: Intelligent shooting guide unit, including shooting scene decision module for selecting intraoral shooting situation, real-time framing module for generating dynamic framing and verification prompt, multi-modal prompt module for generating information used to guide shooting standard intraoral images, and image preprocessing module for intraoral image optimization; Multi-dimensional disease risk analysis unit, including image feature extraction module for high-dimensional disease feature extraction of intraoral images, multi-omics data fusion module for fusion of non-image oral health data and high-dimensional disease features, and risk assessment module for outputting structured oral disease risk determination results according to fusion results; Personalized interactive unit, including risk interpretation module for interpreting results according to risk determination results, task challenge module for generating disease intervention interactive tasks, health knowledge push module for pushing scenario-based knowledge, and social sharing module for generating dynamic health cards; Whole-cycle health record unit, including time axis archiving module for time sequence archiving analysis of key output information of intelligent shooting guide unit, multi-dimensional disease risk analysis unit and personalized interactive unit, and data linkage module for interaction with external health management system data.

2. The intelligent interactive system for oral health management according to claim 1, wherein, In the shooting scene decision module, the shooting situation is defined in a structured manner, including behavior categories and time information, wherein the behavior categories include dietary behavior and cleaning behavior.

3. The intelligent interactive system for oral health management according to claim 1, wherein, In the real-time framing module, AI real-time image segmentation technology is used to generate dynamic framing, and key areas including tooth occlusal surface, buccal surface and lingual surface are covered.

4. The intelligent interactive system for oral health management according to claim 1, wherein, In the real-time framing module, the user's shooting angle is detected in real time and the deviation is calculated, and when the deviation exceeds the threshold, a verification prompt is output, including dynamic framing body change and specific adjustment behavior information prompt, including adjustment direction, adjustment angle and target shooting area.

5. The intelligent interactive system for oral health management according to claim 1, wherein, In the multi-modal prompt module, the guided information includes voice prompts and graphical annotations, and the graphical annotations include tooth contour schematic diagrams.

6. The intelligent interactive system for oral health management according to claim 1, wherein, In the multi-omics data fusion module, the weights of non-image oral health data and high-dimensional disease features are adjusted based on correlation data through attention mechanism; the non-image oral health data includes saliva detection data.

7. The intelligent interactive system for oral health management according to claim 1, wherein, The risk determination results include oral diseases, risk scores, risk levels and risk causes.

8. The intelligent interactive system for oral health management according to claim 1, wherein, In the task challenge module, through a pre-set risk level-task mapping rule library, combined with user portraits and behavior context, differential and executable interactive intervention tasks are dynamically matched and generated. 9.The smart interactive system for oral health management of claim 1, wherein, In the health knowledge push module, by constructing an association mapping table of oral disease types and scenario-based health knowledge, corresponding scenario-based knowledge points are matched based on the disease types in the user's risk determination results, and a pre-set multimedia template library is called to generate personalized health education content in the corresponding form for pushing.

10. An intelligent interactive method for oral health management, characterized in that, An intelligent interactive system for oral health management according to any one of claims 1-9; the method comprises: S1, the user selects the intraoral shooting situation, and shoots the standard intraoral image according to the dynamic framing, the verification prompt and the guide information; after the standard intraoral image is preprocessed, it is prompted that the shooting is completed; S2, high-dimensional disease feature extraction is performed on the intraoral image; non-image oral health data is input, and high-dimensional disease features are fused for risk assessment, and a structured oral disease risk judgment result is output; S3, at least one of result interpretation of the risk judgment result, generation of disease intervention interaction task, pushing of scenario knowledge, and generation of dynamic health card is performed; S4, time sequence archiving analysis is performed on the key output information in the processes of S1-S3; S5, data interaction is performed with an external health management system.

Images