System and method for assessing and analyzing postural health of the neck and shoulder region of a teenager

Abstract

The application discloses a system and method for evaluating and analyzing the health condition of the neck and shoulder posture of teenagers, which comprises a registration interaction module, an inquiry information collection module, a visual inspection information collection module and an AI comprehensive information processing module, wherein the inquiry information collection is mainly through online answering of preset questions; the visual inspection information collection is mainly through a self-provided camera of a smart phone or a PAD or the like to collect the neck and shoulder posture; the AI comprehensive information processing module is fused with computer vision AI technology to realize quantitative extraction and intelligent analysis of the neck and shoulder posture parameters, and the core comprises five-layer architecture of a data access layer, an AI preprocessing layer, an AI core analysis layer, a report generation layer and a data backflow layer, so that automatic calculation and quantitative evaluation of core posture parameters such as ear-shoulder horizontal offset distance can be realized.

Description

Technical Field

[0001] This invention relates to a medical testing cloud platform, and more particularly to a system and method for assessing and analyzing the health status of the neck and shoulder posture of adolescents. It can use a combination of traditional Chinese medicine techniques of "inspection" and "inquiry" to remotely diagnose and assess the health status of the neck and shoulder posture and provide professional advice. It can be widely used in families or schools to assess and analyze the health status of the neck and shoulder posture of adolescents. Background Technology

[0002] With the arrival of the 4G and 5G internet era, the learning and lifestyles of teenagers have undergone significant changes. Besides spending their days hunched over books and homework, they are also addicted to electronic devices such as tablets and mobile phones in their spare time, leading to a growing number of teenagers becoming "the most glued-to-the-head" generation. Prolonged periods of looking down at desks easily cause neck and shoulder muscle fatigue. Some teenagers will tilt their bodies or necks to temporarily relieve neck and shoulder muscle pain. Some teenagers are growing rapidly, and the height of their desks and chairs is disproportionate to their height, causing excessive flexion of the neck and back while reading and writing. Others carry heavy backpacks. Under these unfavorable factors, more and more teenagers are losing their original graceful or elegant posture, exhibiting deformities such as uneven shoulders, hunchback, scoliosis, and forward head posture (cervical flexion). These deformities not only seriously affect the physical appearance of teenagers but, more importantly, severely impact their spinal health and even cause psychological anxiety.

[0003] Traditional Chinese medicine has long emphasized "prevention before disease occurs, and prevention of disease progression." To prevent postural deformities and spinal sub-health, it's crucial to utilize specialized examinations to detect and correct postural deformities in adolescents as early as possible. Simultaneously, parents and teachers need to cultivate and encourage good ergonomic reading and writing postures in teenagers. However, most parents lack knowledge of these specialized examinations and basic ergonomic posture guidelines, often requiring them to take their children to specialist medical institutions. This necessitates both the child and parent taking time off work, and can lead to long waiting times, especially when hospitals are crowded. Therefore, developing a remote diagnostic cloud system that can automatically and accurately detect and assess the health of adolescents' neck and shoulder posture is essential. This system would effectively alleviate the aforementioned difficulties in assessing adolescent posture and perfectly meet the needs of the rapidly developing artificial intelligence era. Summary of the Invention

[0004] This invention provides a system and method for assessing and analyzing the health status of the neck and shoulder posture in adolescents. It addresses the shortcomings of existing clinical practices that cannot provide efficient, systematic, and intelligent assessments of the health status of the neck and shoulder posture in adolescents. The aim is to provide society with a cloud platform that can apply the "combination of inquiry and observation" technique of traditional Chinese medicine for musculoskeletal disorders, integrate computer vision AI intelligent quantitative analysis, and realize remote diagnosis and assessment of the health status of the neck and shoulder posture, and provide professional advice and guidance reports.

[0005] The technical solution of this invention is: a health assessment and analysis system for neck and shoulder posture in adolescents, including a registration and interaction module, a consultation information collection module, a visual diagnosis information collection module, and an AI comprehensive information processing module;

[0006] The registration interaction module is used to assist subjects in completing real-name registration and to provide subjects with an introduction to system functions and precautions.

[0007] The consultation information collection module is used to collect four questions from the subject and upload them to the AI ​​comprehensive information processing module. The four questions are general information, neck, shoulder and back symptoms, impact on learning or life, and factors that induce neck, shoulder and back pain.

[0008] The diagnostic information collection module assists subjects in capturing and recording the cervical spine mobility in all directions, the collinearity or coplanarity of body surface landmarks, and the positional characteristics of the ear center and acromion point in frontal and lateral views under natural body position through voice prompts and virtual animation examples. After storage, the information is uploaded to the AI ​​integrated information processing module.

[0009] The AI ​​integrated information processing module includes a data access layer, an AI preprocessing layer, an AI core analysis layer, and a report generation layer.

[0010] The data access layer inputs structured data collected by the consultation information collection module, image and video information collected by the visual diagnosis information collection module, and gravity sensor data from the electronic product, and associates them with the user QR code generated by the subject in the early stage; it outputs a standard format AI input dataset.

[0011] The AI ​​preprocessing layer takes into account the standard format AI input dataset output by the data access layer; it locates the core skeletal points of the neck and shoulders using the MediaPipe Pose lightweight posture model and outputs two-dimensional coordinates and confidence scores. The core skeletal points of the neck and shoulders include the center point of the external auditory canal (i.e., the center of the ear) and the acromion apex; it adds an AI posture compliance pre-detection function. If it detects that the subject's body rotation causes the sagittal plane to be not perpendicular to the camera or the ear is obscured by hair, resulting in a confidence score <0.6, the system will interrupt the analysis and immediately provide feedback to the front end requesting that the subject turn to the side of the camera and expose the ear; it outputs a quantitative feature dataset.

[0012] The AI ​​core analysis layer takes into account the quantitative feature dataset output by the AI ​​preprocessing layer, and uses dual-model parallel processing to realize body posture visual AI analysis and symptom risk AI analysis.

[0013] The posture visual AI analysis is based on skeletal point coordinates, combined with preset reference object pixel size ratios or device depth information for spatial scale calibration; it establishes an absolute vertical coordinate system using data from the electronic product's built-in gravity sensor, automatically calculating the ear-shoulder horizontal offset distance. The calculation formula is: in the absolute vertical coordinate system, ear-shoulder horizontal offset distance = vertical distance from the ear center coordinate to the gravity vertical line - absolute value of the vertical distance from the same side acromion coordinate to the gravity vertical line; simultaneously, it calculates posture parameters, compares them with backend standard thresholds, and outputs an AI posture score (0-100 points, above 90 points is healthy), a list of abnormal items (including the abnormal level of ear-shoulder horizontal offset distance), and the degree of abnormality (mild / obvious); the posture parameters include the height difference between the two shoulders, cervical spine lateral flexion / rotation angle, and kyphosis angle.

[0014] The symptom risk AI analysis uses a logistic regression classification model to transform the structured data collected by the consultation information collection module into quantitative features and output the AI ​​risk level. The quantitative features include pain score, duration of aches and pains, and number of accompanying symptoms.

[0015] The report generation layer is used to output the final neck and shoulder posture health assessment and analysis report of the subject. The report includes the horizontal offset distance between the ear and shoulder, and matches a special correction plan for abnormal horizontal offset distance between the ear and shoulder.

[0016] This invention also provides a method for assessing and analyzing the neck and shoulder posture health of adolescents, the specific steps of which are as follows:

[0017] Step 1: Assist participants in completing real-name registration through the registration interaction module, and provide participants with an introduction to the system functions and precautions;

[0018] Step 2: Collect four questions from the subject through the consultation information collection module and upload them to the AI ​​comprehensive information processing module. The four questions are general information, neck, shoulder and back symptoms, impact on study or life, and factors that induce neck, shoulder and back pain.

[0019] Step 3: Use the visual diagnosis information collection module to provide voice prompts and virtual animation examples to assist subjects in capturing and recording the cervical spine mobility in all directions, the collinearity or coplanarity of body surface landmarks, and the positional characteristics of the ear center and acromion point in the subject's natural body position from the front and side views using the built-in camera and gravity sensor of the electronic product. After storage, upload the data to the AI ​​integrated information processing module.

[0020] Step 4: The AI ​​integrated information processing module includes a data access layer, an AI preprocessing layer, an AI core analysis layer, and a report generation layer.

[0021] The data access layer inputs structured data collected by the consultation information collection module, image and video information collected by the visual diagnosis information collection module, and gravity sensor data from the electronic product, and associates them with the user QR code generated by the subject in the early stage; it outputs a standard format AI input dataset.

[0022] The AI ​​preprocessing layer takes into account the standard format AI input dataset output by the data access layer; it locates the core skeletal points of the neck and shoulders using the MediaPipe Pose lightweight posture model and outputs two-dimensional coordinates and confidence scores. The core skeletal points of the neck and shoulders include the center point of the external auditory canal (i.e., the center of the ear) and the acromion apex; it adds an AI posture compliance pre-detection function. If it detects that the subject's body rotation causes the sagittal plane to be not perpendicular to the camera or the ear is obscured by hair, resulting in a confidence score <0.6, the system will interrupt the analysis and immediately provide feedback to the front end requesting that the subject turn to the side of the camera and expose the ear; it outputs a quantitative feature dataset.

[0023] The AI ​​core analysis layer takes into account the quantitative feature dataset output by the AI ​​preprocessing layer, and uses dual-model parallel processing to realize body posture visual AI analysis and symptom risk AI analysis.

[0024] The posture visual AI analysis is based on skeletal point coordinates, combined with preset reference object pixel size ratios or device depth information for spatial scale calibration; it establishes an absolute vertical coordinate system using data from the electronic product's built-in gravity sensor, automatically calculating the ear-shoulder horizontal offset distance. The calculation formula is: in the absolute vertical coordinate system, ear-shoulder horizontal offset distance = vertical distance from the ear center coordinate to the gravity vertical line - absolute value of the vertical distance from the same side acromion coordinate to the gravity vertical line; simultaneously, it calculates posture parameters, compares them with backend standard thresholds, and outputs an AI posture score (0-100 points, above 90 points is healthy), a list of abnormal items (including the abnormal level of ear-shoulder horizontal offset distance), and the degree of abnormality (mild / obvious); the posture parameters include the height difference between the two shoulders, cervical spine lateral flexion / rotation angle, and kyphosis angle.

[0025] The symptom risk AI analysis uses a logistic regression classification model to transform the structured data collected by the consultation information collection module into quantitative features and output the AI ​​risk level. The quantitative features include pain score, duration of aches and pains, and number of accompanying symptoms.

[0026] The report generation layer is used to output the final neck and shoulder posture health assessment and analysis report of the subject. The report includes the horizontal offset distance between the ear and shoulder, and matches a special correction plan for abnormal horizontal offset distance between the ear and shoulder.

[0027] Furthermore, the AI ​​integrated information processing module also includes a data feedback layer;

[0028] The data backflow layer desensitizes the evaluation results and periodically backflows them to the background database to incrementally train the MediaPipe Pose lightweight pose model, outputting the optimized MediaPipe Pose lightweight pose model.

[0029] Furthermore, the general information includes gender, age, height, and the type of school attended;

[0030] The neck, shoulder, and back symptoms can be selected as either "no neck, shoulder, or back pain" or "neck, shoulder, and back pain". If you select "neck, shoulder, and back pain", you can then select whether there are any accompanying symptoms and provide the duration and score of the neck, shoulder, and back pain. If there are accompanying symptoms, you can then select one or more of the following options: dizziness, headache, numbness in the hands, chest tightness, poor sleep, and memory loss.

[0031] The impact on learning or life includes seven items, namely, stiff neck, clicking neck, neck, shoulder and back pain affecting learning or reading, neck, shoulder and back pain affecting sports or reading, neck, shoulder and back pain affecting concentration, neck, shoulder and back pain affecting normal sleep, and neck, shoulder and back pain causing bad mood. Each item can be selected as one of the following four severity levels: none, mild, moderate and severe.

[0032] The factors that induce neck, shoulder and back pain can be selected individually or in combination from the following seven common influencing factors: heavy study load, unsuitable height of reading and writing desks and chairs, computer or mobile phone use, incorrect reading and writing posture, neck injury, musical instrument classes, and lack of physical exercise.

[0033] The beneficial effects of this invention are as follows: The adolescent neck and shoulder posture health assessment and analysis system and analysis and information processing method of this invention can apply the combination of traditional Chinese medicine's "inspection" and "inquiry" techniques for musculoskeletal diagnosis, and integrate computer vision AI intelligent quantitative analysis to realize remote diagnosis and assessment of neck and shoulder posture health status and provide professional advice. For the first time, the horizontal offset distance between the ear and shoulder is included as a core indicator in the adolescent neck and shoulder posture assessment, realizing the objective quantitative judgment of head forward extension (cervical flexion), and solving the defects of traditional manual assessment that are highly subjective and difficult to identify minor abnormalities.

[0034] This platform can be widely used in families and schools to assess and analyze the neck and shoulder posture health of teenagers. No professional equipment is required; data collection and quantitative assessment can be completed using only a regular smartphone / Pad. It is easy to operate and highly accurate, which can effectively alleviate the difficulties in assessing the posture of teenagers and truly benefit the spinal health of teenagers. At the same time, the AI ​​model can be continuously iterated and optimized through data feedback, and the assessment accuracy will continue to improve with the accumulation of data, which has good sustainability. Attached Figure Description

[0035] Figure 1 This is a flowchart of the data processing process for the AI ​​integrated information processing module. Detailed Implementation

[0036] The present invention will now be further described with reference to the accompanying drawings.

[0037] This system is designed for assessing and analyzing the health status of neck and shoulder posture in adolescents. The cloud platform is presented as an app and comprises a registration and interaction module, a consultation information collection module, a visual diagnosis information collection module, and an AI-powered comprehensive information processing module. Consultation information collection primarily involves answering pre-set questions online. Visual diagnosis information collection mainly utilizes the built-in cameras of smartphones or tablets to capture neck and shoulder posture data. The AI-powered comprehensive information processing module integrates computer vision AI technology to achieve quantitative extraction and intelligent analysis of neck and shoulder posture parameters. Its core architecture consists of six layers: a data access layer, an AI preprocessing layer, an AI core analysis layer, a fusion decision layer, a report generation layer, and a data feedback layer. This architecture enables the automatic calculation and quantitative assessment of core posture parameters such as the horizontal offset distance between the ear and shoulder.

[0038] (a) Registration and Interactive Module

[0039] Featuring a simple and user-friendly interface, after downloading and installing the cloud platform app, users can activate the app by clicking the icon and completing real-name registration. Upon completion, the top of the software interface displays a welcome message: "Welcome to the Adolescent Neck and Shoulder Posture Health Status Analysis Cloud Platform." The bottom of the interface shows the entire assessment and analysis process, comprising three steps: Step 1, information collection; Step 2, visual information collection; and Step 3, information summary and assessment. At the very bottom of the interface, a message appears: "Friendly Reminder: To ensure the accuracy of the assessment results, please answer the questions truthfully according to the system's text or illustrative prompts, and present your posture or activities as you normally would." After completing the registration and interaction module on this page, the system will automatically redirect to the information collection module interface.

[0040] (II) Medical Consultation Information Collection Module

[0041] Subjects are required to truthfully select or write down the items for each consultation based on their own actual situation. For selectable items, simply tap the "□" before the desired item. If logically suitable for single selection, the system will prompt for single selection and limit the selection to one option only; that is, after selecting one item, other options under the same item cannot be selected. If logically suitable for multiple selection, the system will prompt for multiple selection and set it to allow multiple options to be selected simultaneously. For writing items, after the subject clicks on such an item, a writing board will immediately appear at the bottom of the interface, along with the prompt "Please write your answer on the writing board that pops up on the interface, and pay attention to the prescribed units." The subject only needs to write their answer on the writing board with their finger according to the required units, and the system will automatically fill the written answer into the parentheses of the item.

[0042] The collected consultation content mainly includes four questions: "General Information", "Neck, Shoulder and Back Symptoms", "Impact on Study or Life", and "Factors that Trigger Neck, Shoulder and Back Pain". Each question is on a separate screen. Each question includes 5-8 short question items. Items marked with "□" are optional and can be selected as a single item (logically suitable for single selection) or multiple items (logically suitable for single or multiple selection). Simply tap the "□" before the item you want to select. Items marked with () need to be filled in truthfully in the parentheses according to the indicated units.

[0043] The "General Information" includes "Gender □ Male □ Female Age □ □ Years Height ( ) cm School Type □ Preschool □ Primary School □ Junior High School □ Senior High School". Except for height, which needs to be written, the rest are single selections. After the "General Information" is entered, the system will automatically jump to the "Neck, Shoulder and Back Symptoms" collection module interface.

[0044] The "neck, shoulder, and back symptoms" include "□No neck, shoulder, or back pain usually, □Neck, shoulder, and back pain usually." These two options are single selections. If "□No neck, shoulder, or back pain usually" is selected, the system will automatically jump to the "Inspection Information Collection Module." If "□Neck, shoulder, and back pain usually" is selected, the system will add the following question items under this question item: In addition to neck, shoulder, and back pain, are there any accompanying symptoms? "□No accompanying symptoms □Dizziness □Headache □Numbness in hands □Chest tightness □Poor sleep □Decreased memory." Accompanying symptoms can be selected individually or in multiples. The duration of neck, shoulder, and back pain is () weeks. Please fill in the actual duration. The score for neck, shoulder, and back pain is () points. Please score your neck, shoulder, and back pain according to the degree of pain. Severe pain is 10 points, and no pain is 0 points.

[0045] After the "neck, shoulder and back symptoms" are entered, the system will automatically redirect to the "impact on study or life" data collection module interface.

[0046] The "Impact on Learning or Life" includes seven items: "Difficulty in Neck Rotation," "Cracking Neck," "Neck, Shoulder, and Back Pain Affecting Learning or Reading," "Neck, Shoulder, and Back Pain Affecting Sports or Reading," "Neck, Shoulder, and Back Pain Affecting Concentration," "Neck, Shoulder, and Back Pain Affecting Normal Sleep," and "Neck, Shoulder, and Back Pain Causing Negative Mood." Each item corresponds to one of four severity levels: "□None □Mild □Moderate □Severe." Users should select the appropriate level by clicking the corresponding "□" after each item. After the "Impact on Learning or Life" data is entered, the system will automatically redirect to the "Factors Inducing Neck, Shoulder, and Back Pain" data collection module interface.

[0047] The "Factors that Trigger Neck, Shoulder, and Back Pain" include seven common influencing factors: "□ Heavy academic workload," "□ Inappropriate height of reading and writing desks and chairs," "□ Computer or mobile phone use," "□ Incorrect reading and writing posture," "□ Neck injury," "□ Musical instrument classes," and "□ Lack of physical exercise." Users can select one or more of these factors in the corresponding "□" boxes based on their actual situation. After the "Factors that Trigger Neck, Shoulder, and Back Pain" are entered, the system will automatically redirect to the "Visual Diagnosis Information" module interface.

[0048] (III) Inspection Information Collection Module

[0049] The system uses the built-in camera and IMU (Integrated Mutual Acuity Unit) of a computer, tablet, or mobile phone to capture images or video movements. When the visual diagnosis information acquisition module is activated, the interface will automatically display a virtual human upper body front and side outline prompt box and text prompts indicating which parts and positions should be included in the front or side image acquisition and precautions (such as a naturally relaxed upper body position, slow movements, wearing a tight-fitting collarless T-shirt, and ensuring that the ear center and acromion point are clearly visible when acquiring side images). It also provides examples through virtual animation. When the subject cooperates with the examination according to the demonstration, the visual diagnosis information acquisition module will record the cervical spine mobility in all directions and the collinearity or coplanarity of surface landmarks, as well as the positional characteristics of the ear center and acromion point in the front and side views of the subject in a natural position, through the camera. It will also store and record image and video information and upload it to the comprehensive information processing module.

[0050] The inspection of the frontal view in a natural body position includes: Are the shoulders of equal height? Are the shoulders not on the same plane front and back? Are the exposed sizes of the left and right auricles symmetrical? Is the line connecting the center points of the left and right ears parallel to the horizontal plane? Is the line connecting the left and right angles of the mandible parallel to the horizontal plane? Are the lateral flexion range of motion of the cervical spine equal on both sides? Are the lateral rotation range of motion of the cervical spine equal on both sides?

[0051] The lower lateral view in the natural body position includes: whether there is a natural forward curve at the back of the neck; whether the upper back is upright and whether there is any kyphosis; whether the plane formed by the back is roughly perpendicular to the horizontal plane; whether the chin is protruding or retracted; whether the flexion and extension range of the cervical spine is approximately 45°; and whether the horizontal offset between the vertical line from the center of the ear and the vertical line from the acromion is within the normal range.

[0052] After the observation and diagnosis information collection module completes the collection, the observation and diagnosis information will be transmitted to the comprehensive information processing module. At the same time, the system interface will prompt the subject to relax the neck and shoulders and rest for 1-2 minutes. The system will automatically generate an assessment report and explain how to obtain the report.

[0053] (iv) AI Integrated Information Processing Module

[0054] This module is the core analysis module, which uses a five-layer architecture to realize the entire process of "data aggregation, AI quantitative analysis, report generation, and model iteration," completing the automatic calculation of the ear-shoulder horizontal offset distance and the quantitative assessment of head protrusion, such as... Figure 1 As shown, the specific architecture and functions are as follows:

[0055] Data access layer: Input structured consultation data, visual diagnosis images / videos, mobile phone IMU gravity sensor data, and user QR codes; perform data verification, anonymization, format normalization, and QR code binding; output standard format AI input dataset, remove blurry images, videos with insufficient duration, and consultation data with null values, and return to the front end to prompt for re-collection if the data is invalid.

[0056] AI Preprocessing Layer: Inputs a standardized dataset; performs image denoising, skeletal point calibration, diagnostic labeling, and invalid frame removal. The core functionality utilizes the MediaPipe Pose lightweight pose model to locate 18 core skeletal points in the neck and shoulder region, including the center of the external auditory canal (ear center) and the acromion apex, outputting two-dimensional coordinates (x, y) and confidence scores. An AI pose compliance pre-detection function is added; if the system detects that the subject's body rotation causes the sagittal plane to be not perpendicular to the camera, or that the ear is obscured by hair resulting in a confidence score <0.6, the system will interrupt the analysis and immediately provide feedback to the front end requesting "Please face the camera sideways and show your ear." It outputs a quantified feature dataset, providing a foundation for calculating body pose parameters.

[0057] AI core analysis layer: Input quantitative feature dataset, use dual-model parallel processing, core realization of automatic calculation of ear-shoulder horizontal offset distance and intelligent analysis of body posture and symptoms;

[0058] Body posture visual AI analysis: Based on skeletal point coordinates, spatial scale calibration is performed by combining preset reference object pixel-to-center ratio or device depth information. An absolute vertical coordinate system is established using IMU gravity data, and the horizontal offset distance between the ear and shoulder is automatically calculated (calculation formula: in the absolute vertical coordinate system, the horizontal offset distance between the ear and shoulder = the vertical distance from the ear center coordinate to the gravity vertical line - the absolute value of the vertical distance from the acromion coordinate on the same side to the gravity vertical line). At the same time, all body posture parameters such as the height difference between the two shoulders, the lateral flexion / rotation angle of the cervical spine, and the kyphosis angle are calculated and compared with the standard threshold in the background. The AI ​​body posture score (0-100 points, above 90 points is healthy), a list of abnormal items (including the abnormal level of the horizontal offset distance between the ear and shoulder), and the degree of abnormality (mild / obvious) are output.

[0059] Symptom risk AI analysis: Using a logistic regression classification model, the consultation data is transformed into quantitative features such as pain score, duration of soreness, and number of accompanying symptoms, and outputs AI risk level (0=no risk, 1=muscle strain, 2=mild abnormality, 3=significant abnormality).

[0060] Report generation layer: Input assessment conclusions, abnormalities (including abnormalities in the horizontal offset distance between the ear and shoulder), and inducing factors; execute personalized maintenance plan generation, voice prompts, and report layout; match exclusive correction plans for abnormalities in the horizontal offset distance between the ear and shoulder (forward head posture); output a comprehensive report on individualized neck and shoulder posture health assessment and analysis, retaining the original platform's report format, voice broadcast, and electronic document storage functions.

[0061] Data feedback layer: Input desensitized assessment data and physician review results; execute incremental training of AI model and database update; output optimized AI model and update standard body posture library. The model is fine-tuned through incremental learning every quarter to improve the recognition and assessment accuracy of parameters such as ear-shoulder horizontal offset distance. The mobile APP is upgraded seamlessly.

[0062] Definition and evaluation criteria of core quantitative indicators:

[0063] Professional term: Ear-shoulder horizontal offset distance (head forward extension distance);

[0064] Standard definition: The horizontal straight-line distance between two vertical lines drawn from the center of the external auditory canal (ear center) and the acromion apex in a natural standing side view.

[0065] Assessment thresholds: Normal ≤1.5cm, mild abnormality 1.5~3.0cm, significant abnormality >3.0cm. The larger the value, the more severe the cervical flexion (head forward extension). (Note: The system converts pixel coordinates into absolute cm values ​​through environmental reference calibration or depth sensor.)

[0066] This module links the information collected from the consultation clinic and the images and videos collected from the observation clinic through the QR code generated by the subject in advance, and transmits them to the APP's backend database. The images and videos of the neck and shoulders collected from the observation clinic use AI technology to perform quantitative analysis of posture parameters, rather than simply comparing templates, thus more accurately judging whether there are any abnormalities. The "neck, shoulder and back symptoms" and "impact on learning or life" items collected from the consultation clinic are analyzed based on the items selected by the subject to determine the presence and severity of neck and shoulder symptoms. Then, the comprehensive information processing module summarizes and analyzes the AI ​​quantitative results of the observation clinic and the judgment results of the consultation clinic, and finally gives the subject's assessment conclusion on the health status of the neck and shoulder posture. At the same time, combined with the selected items of "factors that induce neck, shoulder and back pain" and the subject's own bad habits, targeted neck and shoulder health posture maintenance knowledge is given, and finally, a comprehensive report on the individualized neck and shoulder posture health assessment and analysis for each subject is generated.

[0067] Specific assessment criteria for neck and shoulder posture health:

[0068] If the subject selects "□ No neck pain", and the following criteria are met in a natural frontal view: both shoulders are at the same height, both shoulders are on the same plane, both ears are symmetrical in size, the line connecting the center points of both ears is parallel to the horizontal plane, the line connecting the angles of both jaws is parallel to the horizontal plane, the lateral flexion and rotation of the cervical spine are equal; and in a natural lateral view: the back of the neck has a natural forward curve, the upper back is upright without kyphosis, the plane formed by the back is perpendicular to the horizontal plane, the chin is not protruding or retracted, the horizontal offset between the ear and shoulder is ≤1.5cm, and the flexion and extension of the cervical spine is approximately 45°; then the subject's neck and shoulder posture is healthy.

[0069] If a subject selects "□ has neck pain" and the duration is less than or equal to 2 weeks, selects "□ no other accompanying symptoms", and selects "□ none" for all 7 items in "impact on learning or life", and all items in the frontal and lateral views under natural posture are normal (horizontal deviation of ear and shoulder ≤ 1.5 cm), then although the subject's neck and shoulder posture is normal, there may be neck and shoulder muscle strain or other conditions.

[0070] If a subject selects "□ has neck pain" but only one or two items show abnormalities in the frontal and / or lateral views under natural posture (including mild abnormalities with a horizontal deviation of 1.5 to 3.0 cm between the ear and shoulder), then the subject has mild abnormalities in neck and shoulder posture.

[0071] If a subject selects "□ has neck pain" and selects "one or more other accompanying symptoms", and there are one or more mild or more severe options among the seven items in "impact on learning or life", and there are one or more abnormalities (including obvious abnormalities such as ear-shoulder horizontal deviation distance > 3.0 cm) in the frontal and / or lateral views under natural body position, then the subject's neck and shoulder posture is abnormal.

[0072] For participants whose neck and shoulder posture is assessed as healthy, the report will display a large-font message at the bottom: "Dear child (with a smiley face emoji), congratulations on your healthy posture! (with a clapping emoji), thumbs up for your good habits! (with a thumbs-up emoji) Keep it up! (with a cheering emoji)"; for participants whose neck and shoulder posture is assessed as normal but who have neck and shoulder muscle strain, the report will display a large-font message: "Dear child (with a smiley face emoji), congratulations on your normal posture! (with a clapping emoji), but you may have neck and shoulder overwork. Please remember to rest properly while continuing to study hard!" (with a cheering emoji) (Emoji) For subjects whose neck and shoulder posture is assessed as abnormal, a large-font prompt will appear at the bottom of the report: "Dear child (with smiley face emoji), you have an abnormality in your neck and shoulder posture (with hug emoji). Please go to a regular hospital with your parents for further professional examination and treatment at the relevant specialist department. We wish you a speedy recovery of your neck and shoulder posture! (with oil emoji)." The above comprehensive report on neck and shoulder posture assessment can be presented via a voice interaction system, prompting the subject whether to receive a live voice broadcast of the comprehensive report or to generate an electronic document stored on the subject's mobile phone or computer. The subject can issue corresponding commands through the voice interaction system.

[0073] All personal information, questionnaire data, visual examination images, and movement videos of the participants are stored on the backend server. This information is used solely for this posture assessment and will not be used for any other commercial purposes. Without the participants' authorization, no one may access or download this private information without their permission. The AI ​​analysis process is anonymized locally / in the cloud to ensure no privacy information is leaked.

[0074] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A system for assessing and analyzing the neck and shoulder posture health of adolescents, characterized by: It includes a registration and interaction module, a consultation information collection module, a visual diagnosis information collection module, and an AI comprehensive information processing module; The registration interaction module is used to assist subjects in completing real-name registration and to provide subjects with an introduction to system functions and precautions. The consultation information collection module is used to collect four questions from the subject and upload them to the AI ​​comprehensive information processing module. The four questions are general information, neck, shoulder and back symptoms, impact on learning or life, and factors that induce neck, shoulder and back pain. The diagnostic information collection module assists subjects in capturing and recording the cervical spine mobility in all directions, the collinearity or coplanarity of body surface landmarks, and the positional characteristics of the ear center and acromion point in frontal and lateral views under natural body position through voice prompts and virtual animation examples. After storage, the information is uploaded to the AI ​​integrated information processing module. The AI ​​integrated information processing module includes a data access layer, an AI preprocessing layer, an AI core analysis layer, and a report generation layer. The data access layer inputs structured data collected by the consultation information collection module, image and video information collected by the visual diagnosis information collection module, and gravity sensor data from the electronic product, and associates them with the user QR code generated by the subject in the early stage; it outputs a standard format AI input dataset. The AI ​​preprocessing layer takes into account the standard format AI input dataset output by the data access layer; it uses the MediaPipePose lightweight posture model to locate the core skeletal points of the neck and shoulders and outputs two-dimensional coordinates and confidence scores. The core skeletal points of the neck and shoulders include the center point of the external auditory canal (i.e., the center of the ear) and the acromion apex; it adds an AI posture compliance pre-detection function. If it detects that the subject's body rotation causes the sagittal plane to be not perpendicular to the camera or the ear is obscured by hair, resulting in a confidence score of <0.6, the system will interrupt the analysis and immediately provide feedback to the front end requesting that the subject turn to the side of the camera and expose the ear; it outputs a quantitative feature dataset. The AI ​​core analysis layer takes into account the quantitative feature dataset output by the AI ​​preprocessing layer, and uses dual-model parallel processing to realize body posture visual AI analysis and symptom risk AI analysis. The body posture visual AI analysis is based on skeletal point coordinates, combined with preset reference object pixel size ratios or device depth information to perform spatial scale calibration; An absolute vertical coordinate system is established by combining data from the gravity sensor built into the electronic product. The horizontal offset distance between the ear and shoulder is automatically calculated using the following formula: In the absolute vertical coordinate system, the horizontal offset distance between the ear and shoulder = the vertical distance from the ear center coordinate to the vertical line of gravity - the absolute value of the vertical distance from the acromion coordinate on the same side to the vertical line of gravity. At the same time, posture parameters are calculated and compared with the standard thresholds in the background to output an AI posture score, a list of abnormal items, and the degree of abnormality. The posture parameters include the height difference between the two shoulders, the lateral flexion / rotation angle of the cervical spine, and the kyphosis angle. The symptom risk AI analysis uses a logistic regression classification model to transform the structured data collected by the consultation information collection module into quantitative features and output the AI ​​risk level. The quantitative features include pain score, duration of aches and pains, and number of accompanying symptoms. The report generation layer is used to output the final neck and shoulder posture health assessment and analysis report of the subject. The report includes the horizontal offset distance between the ear and shoulder, and matches a special correction plan for abnormal horizontal offset distance between the ear and shoulder.

2. The adolescent neck and shoulder posture health assessment and analysis system according to claim 1, characterized in that: The AI ​​integrated information processing module also includes a data feedback layer; The data backflow layer desensitizes the evaluation results and periodically backflows them to the background database to incrementally train the MediaPipePose lightweight pose model, outputting the optimized MediaPipe Pose lightweight pose model.

3. The adolescent neck and shoulder posture health assessment and analysis system according to claim 1, characterized in that: The general information includes gender, age, height, and type of school attended; The neck, shoulder, and back symptoms can be selected as either "no neck, shoulder, or back pain" or "neck, shoulder, and back pain". If you select "neck, shoulder, and back pain", you can then select whether there are any accompanying symptoms and provide the duration and score of the neck, shoulder, and back pain. If there are accompanying symptoms, you can then select one or more of the following options: dizziness, headache, numbness in the hands, chest tightness, poor sleep, and memory loss. The impact on learning or life includes seven items, namely, stiff neck, clicking neck, neck, shoulder and back pain affecting learning or reading, neck, shoulder and back pain affecting sports or reading, neck, shoulder and back pain affecting concentration, neck, shoulder and back pain affecting normal sleep, and neck, shoulder and back pain causing bad mood. Each item can be selected as one of the following four severity levels: none, mild, moderate and severe. The factors that induce neck, shoulder and back pain can be selected individually or in combination from the following seven common influencing factors: heavy study load, unsuitable height of reading and writing desks and chairs, computer or mobile phone use, incorrect reading and writing posture, neck injury, musical instrument classes, and lack of physical exercise.

4. The method for assessing and analyzing the neck and shoulder posture health of adolescents implemented by the adolescent neck and shoulder posture health assessment and analysis system according to any one of claims 1-3, characterized in that, The specific steps are as follows: Step 1: Assist participants in completing real-name registration through the registration interaction module, and provide participants with an introduction to the system functions and precautions; Step 2: Collect four questions from the subject through the consultation information collection module and upload them to the AI ​​comprehensive information processing module. The four questions are general information, neck, shoulder and back symptoms, impact on study or life, and factors that induce neck, shoulder and back pain. Step 3: Use the visual diagnosis information collection module to provide voice prompts and virtual animation examples to assist subjects in capturing and recording the cervical spine mobility in all directions, the collinearity or coplanarity of body surface landmarks, and the positional characteristics of the ear center and acromion point in the subject's natural body position from the front and side views using the built-in camera and gravity sensor of the electronic product. After storage, upload the data to the AI ​​integrated information processing module. Step 4: The AI ​​integrated information processing module includes a data access layer, an AI preprocessing layer, an AI core analysis layer, and a report generation layer. The data access layer inputs structured data collected by the consultation information collection module, image and video information collected by the visual diagnosis information collection module, and gravity sensor data from the electronic product, and associates them with the user QR code generated by the subject in the early stage; it outputs a standard format AI input dataset. The AI ​​preprocessing layer takes into account the standard format AI input dataset output by the data access layer; it uses the MediaPipePose lightweight posture model to locate the core skeletal points of the neck and shoulders and outputs two-dimensional coordinates and confidence scores. The core skeletal points of the neck and shoulders include the center point of the external auditory canal (i.e., the center of the ear) and the acromion apex; it adds an AI posture compliance pre-detection function. If it detects that the subject's body rotation causes the sagittal plane to be not perpendicular to the camera or the ear is obscured by hair, resulting in a confidence score of <0.6, the system will interrupt the analysis and immediately provide feedback to the front end requesting that the subject turn to the side of the camera and expose the ear; it outputs a quantitative feature dataset. The AI ​​core analysis layer takes into account the quantitative feature dataset output by the AI ​​preprocessing layer, and uses dual-model parallel processing to realize body posture visual AI analysis and symptom risk AI analysis. The body posture visual AI analysis is based on skeletal point coordinates, combined with preset reference object pixel size ratios or device depth information to perform spatial scale calibration; An absolute vertical coordinate system is established by combining data from the gravity sensor built into the electronic product. The horizontal offset distance between the ear and shoulder is automatically calculated using the following formula: In the absolute vertical coordinate system, the horizontal offset distance between the ear and shoulder = the vertical distance from the ear center coordinate to the vertical line of gravity - the absolute value of the vertical distance from the acromion coordinate on the same side to the vertical line of gravity. At the same time, posture parameters are calculated and compared with the standard thresholds in the background to output an AI posture score, a list of abnormal items, and the degree of abnormality. The posture parameters include the height difference between the two shoulders, the lateral flexion / rotation angle of the cervical spine, and the kyphosis angle. The symptom risk AI analysis uses a logistic regression classification model to transform the structured data collected by the consultation information collection module into quantitative features and output the AI ​​risk level. The quantitative features include pain score, duration of aches and pains, and number of accompanying symptoms. The report generation layer is used to output the final neck and shoulder posture health assessment and analysis report of the subject. The report includes the horizontal offset distance between the ear and shoulder, and matches a special correction plan for abnormal horizontal offset distance between the ear and shoulder.

5. The method for assessing and analyzing the neck and shoulder posture health of adolescents according to claim 4, characterized in that: The AI ​​integrated information processing module also includes a data feedback layer; The data backflow layer desensitizes the evaluation results and periodically backflows them to the background database to incrementally train the MediaPipePose lightweight pose model, outputting the optimized MediaPipe Pose lightweight pose model.

6. The method for assessing and analyzing the neck and shoulder posture health of adolescents according to claim 4, characterized in that: The general information includes gender, age, height, and type of school attended; The neck, shoulder, and back symptoms can be selected as either "no neck, shoulder, or back pain" or "neck, shoulder, and back pain". If you select "neck, shoulder, and back pain", you can then select whether there are any accompanying symptoms and provide the duration and score of the neck, shoulder, and back pain. If there are accompanying symptoms, you can then select one or more of the following options: dizziness, headache, numbness in the hands, chest tightness, poor sleep, and memory loss. The impact on learning or life includes seven items, namely, stiff neck, clicking neck, neck, shoulder and back pain affecting learning or reading, neck, shoulder and back pain affecting sports or reading, neck, shoulder and back pain affecting concentration, neck, shoulder and back pain affecting normal sleep, and neck, shoulder and back pain causing bad mood. Each item can be selected as one of the following four severity levels: none, mild, moderate and severe. The factors that induce neck, shoulder and back pain can be selected individually or in combination from the following seven common influencing factors: heavy study load, unsuitable height of reading and writing desks and chairs, computer or mobile phone use, incorrect reading and writing posture, neck injury, musical instrument classes, and lack of physical exercise.

Images