Hemodynamics big data percentile three-dimensional visualization system based on artificial intelligence, operation method and application

Abstract

The invention relates to a haemodynamics big data percentile three-dimensional visualization system based on artificial intelligence, an operation method and application, and the system comprises a data set construction unit which is configured to obtain haemodynamics sample data and construct a data set; a first fitting unit configured to: construct a data subset of the data set, and fit an LMS curve coefficient; a mapping data set construction unit configured to: construct a mapping data set of the data subsets; a second fitting unit configured to: construct a data subset of the mapping data set, and fit an LMS curve coefficient; and the percentile plane acquisition unit is configured to calculate a percentile curve value of standard two-dimensional normal distribution to obtain a percentile plane. According to the method, two dimensions are expanded to three dimensions, more change rules are displayed through a curved surface, and research and analysis of a user aiming at problems in actual work are facilitated.

Description

Technical Field

[0001] This invention relates to an artificial intelligence-based hemodynamic big data percentile three-dimensional visualization system, its operation method, and its application, belonging to the biomedical field. Background Technology

[0002] The LMS method is a method for fitting percentile curves based on time-series data samples. The principle of this method is to convert skewed data into a normal distribution using the Box-Cox method, and then estimate the percentile of the dependent variable by fitting the skewness coefficient curve L (Lambda), median curve M (Median), and coefficient of variation curve S (Sigma) of different dependent variables over time, thereby generating a smooth curve.

[0003] As a tool for statistical data analysis, the Landslide Metrics (LMS) method plays a vital role in the biomedical field. LMS can be used to establish models of disease incidence and mortality rates as they change with age, statistically compare physiological indicators across different diseases, and analyze the impact of various indicators such as gender and genetic factors on disease development. This helps researchers understand disease progression and predict disease risk. LMS can also assist decision-makers in establishing scientifically sound reference standards, such as for various physical health indicators in adolescents, including height, weight, head circumference, and waist circumference. By statistically analyzing and fitting curves showing the changes of these indicators with age, appropriate standard values ​​can be provided. By analyzing clinical data and patient characteristics, LMS can help doctors classify and diagnose diseases, enabling them to more accurately grasp the type of disease and the severity of symptoms. In clinical trials, medical personnel need to consider the impact of different treatment plans on patients; LMS can analyze and model these complex relationships, thereby helping doctors determine the optimal medical strategy.

[0004] Hemodynamics is the science that studies the flow patterns of blood in the circulatory system and its relationship with cardiovascular function. Key parameters include cardiac output, vascular resistance, vascular elasticity, and fluid volume. It has significant clinical importance for the accurate diagnosis and treatment of cardiovascular and cerebrovascular diseases such as hypertension and heart failure. However, traditional hemodynamic examinations lack precise data standards. The reference ranges currently used are "one-size-fits-all" and do not adjust for individual factors such as patient age, gender, and BMI. As a result, most doctors cannot accurately interpret the test reports, which lacks guidance for clinical diagnosis and treatment.

[0005] In recent years, with the development of computer technology and precision sensor technology, non-invasive hemodynamic testing based on impedance cardiography (ICG) has gradually become mainstream, while expensive and risky invasive hemodynamic testing is used less and less in clinical practice. Non-invasive ICG testing is non-invasive, safe, and convenient, similar to electrocardiogram (ECG). Electrodes are placed in the base of the neck and on both sides of the abdomen within the chest cavity. The testing process takes only 3-5 minutes, making it possible to collect large-scale hemodynamic data.

[0006] With access to large-scale hemodynamic data, hemodynamic modeling (LMS) can help researchers construct and optimize dynamic hemodynamic models, enabling them to more accurately reflect actual physiological processes. This contributes to understanding the body's blood pressure regulation mechanisms, and allows doctors to better diagnose and treat various cardiovascular diseases. While LMS is highly effective in demonstrating patterns of change and holds significant importance in the medical field, it still has limitations. The method can only fit percentile curves of one-dimensional random variables over time, resulting in two-dimensional curves. In practice, professionals can only observe the changes in a single indicator over time and fit multiple curves for individual observation; however, independently observing the curves of each variable lacks the correlation between two indicators. Furthermore, the relatively complex calculation process of LMS typically requires specialized software for calculation and graphing, making it currently impossible to establish percentile curves that correlate two variables. Hemodynamics is a complex dynamic system, and two-dimensional curves alone are insufficient to extract comprehensive patterns from large datasets; therefore, a new visualization method is needed to address this issue. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides an AI-based three-dimensional visualization system for hemodynamic big data percentiles, its operating method, and applications. It extends the random variables expressed by the LMS method from one dimension to two dimensions, thus expanding the percentile line changing over time from a curve to a surface. Simultaneously, based on self-collected hemodynamic big data, a smooth surface is fitted to comprehensively demonstrate the relationship between male and female cardiac indices and age and body mass index. By increasing the dimension of random variables, the observability of important indicators is enhanced for medical personnel in practical use, more directly revealing the correlations between various indicators, thereby helping medical personnel to solve biomedical problems more intuitively and summarize more practical experience and conclusions.

[0008] Terminology Explanation: 1. LMS Method: This is a method for fitting percentile curves based on time-series data samples. The principle is to transform skewed data into a normal distribution using the Box-Cox method, and then estimate the percentiles of the dependent variable by fitting the skewness coefficient curves (L, Lambda), median curve (M), and coefficient of variation curve (S, Sigma) of different dependent variables over time, thus generating a smooth curve. The LMS method typically represents the percentiles of a single dependent variable, and is therefore represented as a curve changing over time.

[0009] 2. Impedance cardiometry (ICG): This is a non-invasive medical instrument for monitoring cardiac hemodynamics. It assesses cardiac function and circulatory status by measuring changes in thoracic impedance.

[0010] 3. Unity3D: Developed by Unity Technologies, Unity3D is a cross-platform real-time 3D interactive content creation engine that supports various scenarios such as games, architectural visualization, and VR / AR development. The Unity platform provides a complete set of software solutions for creating, operating, and monetizing any real-time interactive 2D and 3D content, supporting platforms including mobile phones, tablets, PCs, game consoles, augmented reality, and virtual reality devices.

[0011] The technical solution of the present invention is as follows: An AI-based 3D visualization system for hemodynamic big data percentiles includes: The dataset construction unit is configured to: acquire hemodynamic sample data and construct a dataset; The first fitting unit is configured to: construct a subset of the dataset and fit the coefficients of the LMS curve; The mapping dataset building unit is configured to: build a mapping dataset of a subset of data; The second fitting unit is configured to: construct a subset of the data in the mapping dataset and fit the coefficients of the LMS curve; The percentile surface acquisition unit is configured to calculate the percentile curve values ​​of a standard two-dimensional normal distribution to obtain the percentile surface.

[0012] The method for three-dimensional visualization of hemodynamic big data percentiles based on artificial intelligence, implemented through the aforementioned three-dimensional visualization system for hemodynamic big data percentiles, includes: Step 1: Obtain hemodynamic sample data and construct the dataset by using dataset construction units; Step 2: Construct a subset of the dataset using the first fitting unit and fit the LMS curve coefficients; Step 3: Construct a mapped dataset of the data subset by using the mapped dataset building unit; Step 4: Construct a subset of the mapped dataset using the second fitting unit and fit the LMS curve coefficients; Step 5: Obtain the unit through the percentile surface, calculate the percentile curve value of the standard two-dimensional normal distribution, and obtain the percentile surface.

[0013] According to a preferred embodiment of the present invention, step 1 involves acquiring hemodynamic sample data and constructing a dataset; including: Obtain a hemodynamic sample data set, including collection time, age, gender, height, weight, hypertension, hypotension, body mass index, and cardiac index; Based on the gender in the dataset, the dataset is divided into two groups: male and female. From the dataset, randomly select two variables from height, weight, high blood pressure, low blood pressure, body mass index, and heart rate index, and choose the corresponding age to obtain the complete dataset. , , t ;in , These represent two randomly selected variables. t Indicate age; , and t Build as a dataset .

[0014] According to a preferred embodiment of the present invention, step 2 involves constructing a subset of the dataset and fitting LMS curve coefficients; including: Dataset according to Sort the values ​​in ascending order, and based on... The value is used to divide the window, where the window size is... Step size is Generate a subset of data ; Fitting data subset The LMS curve coefficients, i.e., for each window's data subset. Uniform sampling T 1 Each sampling interval is used to calculate the value in each sampling interval. median And using the LMS method according to and t The skewness coefficient for each subset of data is calculated. , median Coefficient of variation .

[0015] According to a preferred embodiment of the present invention, step 3 involves constructing a mapping dataset of a data subset; including: Based on the calculated skewness coefficient of each data subset , median and coefficient of variation Calculate the value of each data subset. value, The calculation formula is as follows: ; in, Indicates according to Calculated value; Based on the calculated data subset Values, and combined with subsets of data Build and generate mapping dataset ( ).

[0016] According to a preferred embodiment of the present invention, step 4 involves constructing a subset of the mapping dataset and fitting LMS curve coefficients; including: Map the dataset of Three attributes, according to Sort the values ​​in ascending order, and according to The value is used to divide the window, where the window size is... The step size is Construct a subset of the mapping dataset ; For each subset of data ( ), obtained by uniform sampling Each sampling interval is used to calculate the value in each sampling interval. median And using the LMS method according to and t The skewness coefficient for each subset of data is calculated. , median Coefficient of variation ; Calculate all The values ​​are shown below: ; in, Indicates a given Under the conditions, the calculated The value; The calculated The value, combined with By combining the t-value, we obtain the complete dataset. All mapping values .

[0017] According to a preferred embodiment of the present invention, step 5 involves calculating the percentile curve values ​​of a standard two-dimensional normal distribution to obtain the percentile surface; this includes: For any given percentile There exists a set of percentile curve points, as shown below: ; in, express t 2D standard normal distribution at time 1 Coordinates of a point on the percentile plane and These represent the component coordinates of a point on the percentile plane. and Describes a random variable that follows a 2D standard normal distribution. Represents a 2D standard normal distribution; These represent the maximum and minimum values ​​of age, respectively; K percentiles were obtained by solving the above formula. , Each age group received One point; calculate The corresponding values ​​of the original random variable space and Get coordinates As shown below: ; ; in, Indicated on the c-percentage surface t Moment The component values, express t time hour, The corresponding median, express t time hour, The corresponding skewness coefficient, express t time hour, The corresponding coefficient of variation; Represents the percentage surface of c t Moment Component value, express t time The corresponding median, express t time The corresponding skewness coefficient, express t time The corresponding coefficient of variation; The coordinates are obtained as of By connecting all adjacent points, the percentile plane is obtained.

[0018] An extended reality glasses device, equipped with an AI-based hemodynamic big data percentile 3D visualization system, includes: Input all the point data of the obtained percentile plane into the Unity3D tool. In the Unity3D tool, connect the points of the same percentile plane in spatial order to form a mesh plane. All the percentile planes constitute multiple mesh planes. Scale transformation of the mesh surfaces, that is, adjusting the size of the bounding box of the 3D structure composed of all mesh surface data; Color the mesh surface and set its transparency; In Unity3D, create an empty object as the root node, and make all mesh face objects the child objects of the root node; Place the root node in front of the camera in Unity3D, set it to no rotation and keep its size at the scale after the transformation; After the setup is complete, an executable application is generated and deployed to an augmented reality glasses device. The percentile plane 3D structure is observed in the augmented reality glasses device, completing the 3D visualization of the percentile.

[0019] The beneficial effects of this invention are as follows: 1. This invention extends two-dimensional to three-dimensional, and displays more variation patterns through curved surfaces, which is more beneficial for users to conduct research and analysis on problems in practical work.

[0020] 2. This invention uses the LMS method as the basis for drawing, and the fitted surface also has LMS properties, ensuring the smoothness of the surface, having high accuracy at the boundary points, and avoiding surface intersections and overlaps, thus serving as a true reflection of the data.

[0021] 3. This invention relies on real-world medical big data, and the generated surface has medical reference value and can be provided to hospital doctors for reference; it provides a new guidance method for precision medication of hypertension, filling the gap in my country's precision medication of hypertension field.

[0022] 4. The entire process of blood flow imaging is non-invasive and simple, similar to an electrocardiogram (ECG). Results are available in just 3-5 minutes, allowing clinicians to develop standardized and reasonable individualized hypertension medication plans based on the blood flow imaging report. Even non-hypertension doctors and primary care physicians can diagnose and treat hypertension, thereby changing the current situation where clinicians rely solely on clinical experience in treatment and truly realizing individualized, intelligent, and visualized hypertension treatment in the era of big data precision medicine. Attached Figure Description

[0023] Figure 1 This is an LMS percentile surface plot of male body mass index and cardiac index as a function of age in this invention. Figure 1 (a) is the 10th percentile surface plot; Figure 1 (b) is the 25th percentile plot; Figure 1 (c) is the 50th percentile surface plot; Figure 1 The middle (d) plot is the 75th percentile surface plot; Figure 1 (e) is the 90th percentile plot; Figure 2 This is an LMS percentile surface plot of female body mass index and cardiac index as a function of age, according to the present invention. Figure 2 (a) is the 10th percentile surface plot; Figure 2 (b) is the 25th percentile plot; Figure 2 (c) is the 50th percentile surface plot; Figure 2 The middle (d) plot is the 75th percentile surface plot; Figure 2 (e) is the 90th percentile plot; Figure 3 This is a stereoscopic dual-view diagram of the male CI-BMI percentile surface of the present invention; Figure 3 (a) is the left-side three-dimensional view of the male CI-BMI percentile surface; Figure 3 (b) is a three-dimensional right view of the male CI-BMI percentile surface; Figure 4 This is a three-dimensional dual-view diagram of the female CI-BMI percentile surface of the present invention; Figure 4 (a) is a three-dimensional left view of the female CI-BMI percentile surface; Figure 4 (b) is a three-dimensional right view of the female CI-BMI percentile surface. Detailed Implementation

[0024] The present invention will be further described below with reference to the embodiments and accompanying drawings, but is not limited thereto.

[0025] Example 1 A three-dimensional visualization system for hemodynamic big data percentiles based on artificial intelligence, such as Figure 1-4 As shown, it includes: The dataset construction unit is configured to: acquire hemodynamic sample data and construct a dataset; The first fitting unit is configured to: construct a subset of the dataset and fit the coefficients of the LMS curve; The mapping dataset building unit is configured to: build a mapping dataset of a subset of data; The second fitting unit is configured to: construct a subset of the data in the mapping dataset and fit the coefficients of the LMS curve; The percentile surface acquisition unit is configured to calculate the percentile curve values ​​of a standard two-dimensional normal distribution to obtain the percentile surface.

[0026] Example 2 The method for three-dimensional visualization of hemodynamic big data percentiles based on artificial intelligence is implemented through the three-dimensional visualization system for hemodynamic big data percentiles described in Example 1, including: Step 1: Obtain hemodynamic sample data and construct the dataset by using dataset construction units; Step 2: Construct a subset of the dataset using the first fitting unit and fit the LMS curve coefficients; Step 3: Construct a mapped dataset of the data subset by using the mapped dataset building unit; Step 4: Construct a subset of the mapped dataset using the second fitting unit and fit the LMS curve coefficients; Step 5: Obtain the unit through the percentile surface, calculate the percentile curve value of the standard two-dimensional normal distribution, and obtain the percentile surface.

[0027] Example 3 The difference between the three-dimensional visualization method for hemodynamic big data percentiles based on artificial intelligence described in Example 2 and the method described in Example 2 is as follows: In step 1, hemodynamic sample data are acquired using an impedance cardiography device to construct a dataset; including: Using an impedance cardiograph (ICG) device, electrodes are placed on both sides of the subject's neck and abdomen to collect hemodynamic data and obtain subject information, resulting in a hemodynamic sample data set, including collection time, age, sex, height, weight, hypertension, hypotension, body mass index (BMI), and cardiac index (CI). Based on the gender in the dataset, the dataset is divided into two groups: male and female. Two variables are randomly selected from the dataset of height, weight, high blood pressure, low blood pressure, body mass index (BMI), and cardiac index (CI), along with their corresponding ages, to obtain the complete dataset. , , t ;in , These represent two randomly selected variables. t Indicate age; , and t Build as a dataset .

[0028] Step 2 involves constructing a subset of the dataset and fitting LMS curve coefficients; this includes: Dataset according to Sort the values ​​in ascending order, and based on... The value divides the window to, in Construct a subset of data above, where the window size is Step size is Generate a subset of data (m=1, ..., M); Data overlap is allowed between subsets of data to ensure that the constructed subsets have enough sampling points so that the final drawn surface is smooth; Fitting data subset The LMS curve coefficients, i.e., for each window's data subset. Uniform sampling T 1 Each sampling interval is used to calculate the value in each sampling interval. median And using the LMS method according to and t The skewness coefficient for each subset of data is calculated. , median Coefficient of variation .

[0029] Step 3 involves constructing a mapping dataset for a subset of the data; this includes: Based on the calculated skewness coefficient of each data subset , median and coefficient of variation Calculate the value of each data subset. value, The calculation formula is as follows: ; in, Indicates according to Calculated value; Based on the calculated data subset Values, and combined with subsets of data Build and generate mapping dataset ( ).

[0030] In step 4, a subset of the mapped dataset is constructed, and LMS curve coefficients are fitted; this includes: Map the dataset of Three attributes, according to Sort the values ​​in ascending order, and according to The value is used to divide the window, where the window size is... The step size is Construct a subset of the mapping dataset Data overlap between subsets is allowed, which ensures that the constructed subsets have enough sampling points to make the final drawn surface smooth. For each subset of data ( ), obtained by uniform sampling Each sampling interval is used to calculate the value in each sampling interval. median And using the LMS method according to and t The skewness coefficient for each subset of data is calculated. , median Coefficient of variation ; Calculate all The values ​​are shown below: ; in, Indicates a given Under the conditions, the calculated The value; The calculated The value, combined with By combining the t-value, we obtain the complete dataset. All mapping values .

[0031] In step 5, the percentile curve values ​​of the standard two-dimensional normal distribution are calculated to obtain the percentile surface; this includes: For any given percentile There exists a set of percentile curve points, and all points in this set can be calculated and solved, as shown below: ; in, express t 2D standard normal distribution at time 1 Coordinates of a point on the percentile plane and These represent the component coordinates of a point on the percentile plane. and Describes a random variable that follows a 2D standard normal distribution. Represents a 2D standard normal distribution; These represent the maximum and minimum values ​​of age, respectively; K percentiles were obtained by solving the above formula. , Each age group received One point; calculate The corresponding values ​​of the original random variable space and Get coordinates As shown below: ; ; in, This indicates that on the c-percentage plane (e.g., c=75, the c-percentage plane represents the 75% plane, meaning that 75% of the values ​​are smaller than the values ​​of points on this plane). t Moment The component values, express t time hour, The corresponding median, express t time hour, The corresponding skewness coefficient, express t time hour, The corresponding coefficient of variation; Represents the percentage surface of c t Moment Component value, express t time The corresponding median, express t time The corresponding skewness coefficient, express t time The corresponding coefficient of variation; The coordinates are obtained as of By connecting all adjacent points, a percentile plane can be obtained, which can be presented using graphic rendering.

[0032] Example 4 An extended reality glasses device, deployed with the AI-based hemodynamic big data percentile three-dimensional visualization system described in Example 1, includes: Input all the point data of the obtained percentile plane into the Unity3D tool. In the Unity3D tool, connect the points of the same percentile plane in spatial order to form a mesh plane. All the percentile planes constitute multiple mesh planes. Scale the mesh surfaces by adjusting the size of the bounding box of the 3D structure formed by all mesh surface data (the initial diameter of the bounding box is adjusted to 0.5 meters). Select the root node object PercentileSurfacesRoot containing all percentile surfaces, view the bounding box range of all sub-models in the Inspector panel, and record the maximum length in the X, Y, and Z directions. Calculate the uniform scaling ratio based on this value, using the formula 0.5 / maximum size. Then, enter the corresponding scaling value (e.g., 0.25, 0.25, 0.25) in the Transform → Scale property to uniformly adjust the diameter of the bounding box of the entire percentile surface structure to approximately 0.5 meters. Color the mesh surfaces and set their transparency (ensuring that surfaces of the same percentile have the same color); this makes it easier to observe all percentile surfaces (the transparency makes even occluded surfaces visible), enhancing the perception effect; In the Assets panel, right-click and select Create → Material to create separate material files for different percentile surfaces (such as Material_25, Material_50, etc.). In the Inspector panel, set the Shader to Universal Render Pipeline / Lit or Standard, and adjust the Rendering Mode to Transparent. To distinguish different percentile layers, you can set different colors, and set the alpha value of the colors to about 0.5 to obtain about 50% transparency. Make sure Surface Type: Transparent is checked and Blend Mode is enabled to correctly display the semi-transparent effect. Finally, drag and drop each material onto the MeshRenderer component of the corresponding percentile surface in the scene, and preview the scene to confirm that the colors are clearly distinguishable, the transparency is appropriate, and the layering is normal. In Unity3D, set all mesh faces as virtual objects and place them directly in front of the augmented reality glasses in the world coordinate system, presenting them in 3D in front of the observer; the specific steps are as follows: In Unity3D, create an empty object as the root node, name it PercentileSurfacesRoot, and make all mesh face objects children of the root node. Place the root node approximately two meters in front of the camera in Unity3D (Position set to (0, 0, 2)), set it to no rotation (Rotation set to (0, 0, 0)), and maintain the size at the scaled ratio (set via the Scale property); this way, when running the scene, the user can see the complete 3D percentile structure directly in front of their field of view using an extended reality device; After setup, generate an executable application and deploy it to an augmented reality glasses device (such as PICO glasses, AR, MR, VR glasses). Observe the stereoscopic percentile plane three-dimensional structure with transparency and a sense of hierarchy in the augmented reality glasses device to complete the three-dimensional visualization of the percentile.

[0033] Open Window → Package Manager → XR Plugin Management, install and enable the OpenXR plugin, and install the corresponding XR support module according to the device you are using; then, delete the default camera, create an XR Origin object in the Hierarchy, and ensure that the model is in a visible position in front of it; in File → Build Settings, select the target platform (such as Windows, Android, etc.), and click "Switch Platform" to complete the platform switch; then adjust the relevant parameters in Player Settings, and click "Build and Run" to generate the executable application; after deploying the generated program to the XR glasses device, users can observe a stereoscopic 3D structure with transparency and layering in the extended reality environment; Once the application is processed in Unity3D, it can be published on extended reality glasses such as PICO. Observers wearing glasses can see the percentile composed of 3D data points of all the percentiles suspended in the air. In virtual reality mode, only the percentile plane can be observed through the glasses; in mixed reality mode, the surrounding environment, such as the patient, can also be observed. Observers can view the percentile plane from different angles (up, down, left, right, front, back), and the percentile plane maintains its inherent 3D shape, providing a stereoscopic visual effect. When the observer uses gestures, they can zoom in, zoom out, and rotate the percentile plane as a whole; thus, it ensures that doctors can accurately understand all the shapes and interrelationships of the percentile plane.

Claims

1. An artificial intelligence based blood flow dynamics big data percentile three dimensional visualization system characterized by, include: The dataset construction unit is configured to: acquire hemodynamic sample data and construct a dataset; The first fitting unit is configured to: construct a subset of the dataset and fit the coefficients of the LMS curve; The mapping dataset building unit is configured to: build a mapping dataset of a subset of data; The second fitting unit is configured to: construct a subset of the data in the mapping dataset and fit the coefficients of the LMS curve; The percentile surface acquisition unit is configured to calculate the percentile curve values ​​of a standard two-dimensional normal distribution to obtain the percentile surface.

2. The method for running the blood flow dynamics big data percentile three-dimensional visualization based on artificial intelligence, is realized through the blood flow dynamics big data percentile three-dimensional visualization system of claim 1, it is characterized in being, include: Step 1: Obtain hemodynamic sample data and construct the dataset by using dataset construction units; Step 2: Construct a subset of the dataset using the first fitting unit and fit the LMS curve coefficients; Step 3: Construct a mapped dataset of the data subset by using the mapped dataset building unit; Step 4: Construct a subset of the mapped dataset using the second fitting unit and fit the LMS curve coefficients; Step 5: Obtain the unit through the percentile surface, calculate the percentile curve value of the standard two-dimensional normal distribution, and obtain the percentile surface.

3. The artificial intelligence-based hemodynamic big data percentile three- dimensional visualization run method of claim 2, wherein In step 1, hemodynamic sample data is obtained, and a dataset is constructed; this includes: Obtain a hemodynamic sample data set, including collection time, age, gender, height, weight, hypertension, hypotension, body mass index, and cardiac index; Based on the gender in the dataset, the dataset is divided into two groups: male and female. From the dataset, randomly select two variables from height, weight, high blood pressure, low blood pressure, body mass index, and heart rate index, and choose the corresponding age to obtain the complete dataset. , , t ;in , These represent two randomly selected variables. t Indicate age; , and t Build as a dataset .

4. The method for three-dimensional visualization of hemodynamic big data percentiles based on artificial intelligence according to claim 3, characterized in that, In step 2, a subset of the dataset is constructed, and the coefficients of the LMS curve are fitted. include: Dataset according to Sort the values ​​in ascending order, and based on... The value is used to divide the window, where the window size is... Step size is Generate a subset of data ; Fitting data subset The LMS curve coefficients, i.e., for each window's data subset. Uniform sampling T 1 Each sampling interval is used to calculate the value in each sampling interval. median And using the LMS method according to and t The skewness coefficient for each subset of data is calculated. , median Coefficient of variation .

5. The method for three-dimensional visualization of hemodynamic big data percentiles based on artificial intelligence according to claim 4, characterized in that, Step 3 involves constructing a mapping dataset for a subset of the data; this includes: Based on the calculated skewness coefficient of each data subset , median and coefficient of variation Calculate the value of each data subset. value, The calculation formula is as follows: ； in, Indicates according to Calculated value; Based on the calculated data subset Values, and combined with subsets of data Build and generate mapping dataset ( ).

6. The method for three-dimensional visualization of hemodynamic big data percentiles based on artificial intelligence according to claim 5, characterized in that, In step 4, a subset of the mapping dataset is constructed, and the LMS curve coefficients are fitted. include: Map the dataset of Three attributes, according to Sort the values ​​in ascending order, and according to The value is used to divide the window, where the window size is... The step size is Construct a subset of the mapping dataset ; For each subset of data ( ), obtained by uniform sampling Each sampling interval is used to calculate the value in each sampling interval. median And using the LMS method according to and t The skewness coefficient for each subset of data is calculated. , median Coefficient of variation ; Calculate all The values ​​are shown below: ； in, Indicates a given Under the conditions, the calculated The value; The calculated The value, combined with By combining the t-value, we obtain the complete dataset. All mapping values .

7. The method for three-dimensional visualization of hemodynamic big data percentiles based on artificial intelligence according to claim 6, characterized in that, In step 5, the percentile curve values ​​of the standard two-dimensional normal distribution are calculated to obtain the percentile surface; this includes: For any given percentile There exists a set of percentile curve points, as shown below: ； in, express t 2D standard normal distribution at time 1 Coordinates of a point on the percentile plane and These represent the component coordinates of a point on the percentile plane. and Describes a random variable that follows a 2D standard normal distribution. Represents a 2D standard normal distribution; These represent the maximum and minimum values ​​of age, respectively. K percentiles were obtained by solving the above formula. , Each age group received One point; calculate The corresponding values ​​of the original random variable space and Get coordinates As shown below: ； ； in, Indicates on the c-percentage plane t Moment The component values, express t time hour, The corresponding median, express t time hour, The corresponding skewness coefficient, express t time hour, The corresponding coefficient of variation; Represents the percentage surface of c t Moment Component value, express t time The corresponding median, express t time The corresponding skewness coefficient, express t time The corresponding coefficient of variation; The coordinates are obtained as of By connecting all adjacent points, the percentile plane is obtained.

8. An extended reality glasses device, equipped with an artificial intelligence-based hemodynamic big data percentile three-dimensional visualization system, characterized in that, include: Input all the point data of the obtained percentile plane into the Unity3D tool. In the Unity3D tool, connect the points of the same percentile plane in spatial order to form a mesh plane. All the percentile planes constitute multiple mesh planes. Scale transformation of the mesh surfaces, that is, adjusting the size of the bounding box of the 3D structure composed of all mesh surface data; Color the mesh surface and set its transparency; In Unity3D, create an empty object as the root node, and make all mesh face objects the child objects of the root node; Place the root node in front of the camera in Unity3D, set it to no rotation and keep its size at the scale after the transformation; After the setup is complete, an executable application is generated and deployed to an augmented reality glasses device. The percentile plane 3D structure is observed in the augmented reality glasses device, completing the 3D visualization of the percentile.

Images