A physical fitness visual detection platform

Abstract

The application discloses a physical health visual detection platform, comprising: a monitoring data demand matching unit, which matches corresponding monitoring data demand; a health index data acquisition unit, which constructs a patient health index dataset; a historical health event data acquisition unit, which constructs a health event related dataset; a measured three-dimensional visual model generation unit, which generates a measured three-dimensional visual model associated with a public time axis; a dynamic three-dimensional visual model generation unit, which generates a dynamic three-dimensional visual model; a physical health trend analysis unit, which associates a physical health time series dataset with health event data and health index data, constructs a multi-dimensional health space-time dataset, and adopts a time series analysis algorithm to model a physical health trend to generate a physical health trend curve and extract corresponding trend characteristics; a fusion prediction unit, which generates a corrected prediction result; and a visual interaction unit, which establishes a visual component library and provides a visual configuration interface.

Description

Technical Field

[0001] This invention relates to the field of physical health testing technology, and in particular to a visual physical health testing platform. Background Technology

[0002] This project aims to construct a visualized physical health monitoring platform. By integrating biosensors, mobile terminals, cloud computing, and data visualization technologies, it seeks to achieve real-time, dynamic monitoring, intelligent analysis, and intuitive graphical presentation of multi-dimensional human physiological indicators such as heart rate, blood pressure, blood sugar, body fat percentage, muscle mass, and athletic ability. This overcomes the limitations of traditional health monitoring methods, including spatial and temporal constraints, fragmented data, and insufficient professional interpretation of results. Its core objective is to provide users with convenient and efficient self-health management tools to monitor changes in physical condition, identify potential health risks, and develop scientific exercise and nutrition plans. Simultaneously, it provides structured, continuous, and traceable health data support for medical institutions, health management organizations, and research units to optimize treatment pathways, conduct precise health interventions, and promote innovation in physical health research. Ultimately, this will improve health literacy and quality of life at the individual level, reduce medical costs and disease burden, and promote the popularization of public health awareness, support the implementation of the Healthy China strategy, and drive the digital transformation and high-quality development of the health industry. Finally, it aims to achieve a paradigm shift from disease treatment to health management, constructing a closed-loop health management system encompassing prevention, monitoring, intervention, and improvement. This approach has significant public health value, social benefits, and economic significance.

[0003] Existing predictive models are based only on short-term indicators and ignore the differences in the response of patients with different physical conditions to the same treatment. Therefore, a visual detection platform for physical health is proposed. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a visual detection platform for physical health.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A visual health monitoring platform, comprising: Monitoring data requirement matching unit: acquires clinical trial information and inputs it into a pre-set clinical trial requirement database to match the corresponding monitoring data requirements; Health indicator data acquisition unit: Based on the monitoring data requirements, acquire health indicator data of all patients in the clinical trial, and preprocess and organize the health indicator data to construct a patient health indicator dataset; Historical health event data acquisition unit: acquires the patient's historical health event data, and associates the health indicator data with the health event data based on a pre-set public timeline to construct a health event-related dataset; The measured 3D visual model generation unit performs comprehensive analysis on the corresponding health indicator data in the health event-related dataset to generate a measured 3D visual model associated with the common time axis. Dynamic 3D Visual Model Generation Unit: Based on a common time axis, the measured 3D visual models within a continuous time period are continuously fused to generate a dynamic 3D visual model; Clinical trial visualization report generation unit: Generates a visualization report of the clinical trial based on the dynamic three-dimensional visual model; Physical health trend analysis unit: Collects patients' physical health data according to a preset period to form a physical health time series dataset. Based on the public time axis, the physical health time series dataset is associated with the health event data and health indicator data to construct a multidimensional health spatiotemporal dataset. The time series analysis algorithm is used to model the physical health trend to generate a physical health trend curve and extract the corresponding trend features. Fusion prediction unit: Constructs a prediction feature library containing physical characteristics, wherein the physical characteristics include at least the dynamic trend features extracted by the physical health trend analysis unit, and uses machine learning algorithms to train a prediction model that integrates physical factors to generate corrected prediction results; Visualization Interaction Unit: Establishes a visualization component library and provides a visualization configuration interface, allowing users to dynamically generate customized views through interactive operations, and integrates and displays the trend analysis results generated by the physical health trend analysis unit and the prediction results generated by the fusion prediction unit in the dynamic three-dimensional visual model in an overlay form.

[0006] The above technical solution further includes: Furthermore, the specific steps for the measured 3D visual model generation unit to generate the measured 3D visual model are as follows: From the constructed health event-related dataset, extract health indicator data corresponding to each time point on the public timeline. The health indicator data includes at least laboratory test results data, vital sign data, and imaging test data. Align the health indicator data with patient identifiers and timestamps to ensure that different indicators at the same time point belong to the same patient. The extracted health indicator data is standardized to eliminate the dimensional differences between different indicators. At the same time, a mapping rule in three-dimensional space is assigned to each indicator. The mapping rule includes mapping the common time axis to the X-axis, mapping the indicator type (such as laboratory indicators and vital signs) to the Y-axis classification, mapping the specific value of the indicator to the Z-axis height or color depth, and encoding the individual patient or event type with different shapes or colors. Based on standardized data points, a 3D surface reconstruction or scattered cloud rendering algorithm is used to generate a measured 3D visual model corresponding to a specific time point on a common time axis. This model can intuitively display the distribution status and abnormal fluctuation areas of all patients on multiple health indicators at that moment.

[0007] Furthermore, the specific steps for the dynamic 3D visual model generation unit to generate the dynamic 3D visual model are as follows: The measured 3D visual models, corresponding to multiple consecutive time points, are obtained from the measured 3D visual model generation unit in chronological order to form a model sequence. For the measured 3D visual models at adjacent time points, linear interpolation or spline interpolation algorithms are used to generate intermediate frame models in the time dimension to eliminate the sense of jump when switching models and achieve a smooth transition of changes in health indicators. The original model and the generated intermediate frame model are concatenated in sequence along a common timeline to form a coherent dynamic 3D visual model. At the same time, the model is bound to a timeline slider control, allowing users to observe the continuous change trend of health indicators over time by dragging the timeline.

[0008] Furthermore, in the physical health trend analysis unit, the physical health data collected according to a preset period includes one or more of body mass index, body fat percentage, muscle mass, basal metabolic rate, cardiorespiratory endurance, and bone density. The modeling of physical health trends using time series analysis algorithms includes using long short-term memory networks or dynamic time warping algorithms to generate physical health trend curves and extracting trend features including slope, fluctuation amplitude, and seasonality patterns.

[0009] Furthermore, the physical health trend analysis unit is also used to perform correlation analysis between the trend characteristics and the probability of health events, identify physical change patterns associated with specific health events, and integrate the identified physical change patterns into the measured three-dimensional visual model or dynamic three-dimensional visual model in the form of superimposed curves or heat maps.

[0010] Furthermore, in the fusion prediction unit, the prediction feature library includes the patient's static physical characteristics, which include age, gender, and genetic background data. The prediction model that integrates physical factors using machine learning algorithms includes training with ensemble learning algorithms or deep learning models, using the probability of health events or the quantitative value of nursing effects as labels, and performing hierarchical training for different physical types to obtain a physical-specific prediction sub-model.

[0011] Furthermore, during real-time prediction, the fusion prediction unit matches the corresponding constitution-specific prediction sub-model based on the patient's current physical health data, or performs personalized fine-tuning of the general prediction model through transfer learning to generate prediction results that include constitution weights.

[0012] Furthermore, the fusion prediction unit compares the generated corrected prediction results with the measured data and continuously updates the parameters of the prediction model using an online learning mechanism.

[0013] The present invention has the following beneficial effects: In this invention, by using a fusion prediction unit, the dynamic trend features extracted by the physical health trend analysis unit are used as input variables to construct a prediction model that integrates physical factors. This solves the technical problem that changes in the same indicator indicate different risks in patients with different physical conditions. At the same time, by introducing physical characteristics, the prediction results are made more consistent with the actual situation of individual patients, reducing the false alarm and false negative rates. Attached Figure Description

[0014] Figure 1 This is a system block diagram of a physical health visualization detection platform proposed in this invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Please see Figure 1 As shown, the present invention is a visual detection platform for physical health, comprising: Monitoring data requirement matching unit: acquires clinical trial information and inputs it into a pre-set clinical trial requirement database to match the corresponding monitoring data requirements; First, the monitoring data requirement matching unit receives the input clinical trial information, which includes at least one or more of the following: trial protocol name, investigational drug code, indication type, or trial phase. The monitoring data requirement matching unit uses this information as a query index and inputs it into a pre-set clinical trial requirement database containing various clinical trial templates for precise matching. The matching process aims to retrieve the monitoring data requirements corresponding to the current clinical trial information from the database. The monitoring data requirements are a structured data list used to clearly indicate the specific health indicator data categories that need to be collected and monitored regularly for all enrolled patients during this clinical trial. For example, for a cardiovascular drug trial, its monitoring data requirements may explicitly require the collection of four indicators such as blood pressure, heart rate, electrocardiogram, and blood lipids. Health indicator data acquisition unit: Based on the monitoring data requirements, acquire health indicator data of all patients in the clinical trial, and preprocess and organize the health indicator data to construct a patient health indicator dataset; After the monitoring data requirements are determined, the health indicator data acquisition unit starts the data collection process. This unit connects to the hospital's laboratory information system, electronic medical record system, and image archiving and communication system through an interface, and automatically captures the original health indicator data of all patients in the clinical trial according to the monitoring data requirement list. The health indicator data obtained should include at least laboratory test results (such as complete blood count, urinalysis, liver function), vital signs data (such as body temperature, pulse, respiration, blood pressure) and imaging test data (such as CT and MRI measurements or report conclusions). After acquiring the raw data, the unit immediately performs a preprocessing operation, which includes: Imputing or marking missing values, identifying and correcting outliers, standardizing and converting data formats from different sources, and removing duplicate data; After the above processing, all the scattered and heterogeneous raw data were organized into a structured patient health indicator dataset that is easy to analyze later. Historical health event data acquisition unit: acquires the patient's historical health event data, and associates the health indicator data with the health event data based on a pre-set public timeline to construct a health event-related dataset; The historical health event data acquisition unit acquires all patients' historical health event data recorded during the trial and in their past medical history from the electronic data acquisition system or adverse event reporting system of the clinical trial. The historical health events include, but are not limited to, disease diagnosis, disease relapse, and adverse drug reactions, which are used to indicate clinical conditions that require special attention. Then, the historical health event data acquisition unit calls a pre-set public timeline, which is scaled with absolute dates and has a unified time base; The historical health event data acquisition unit anchors each data point in the generated patient health indicator dataset and each historical health event data point to the corresponding position on the public time axis according to the actual date of occurrence or recording, thereby constructing a multi-dimensional data set that integrates indicator monitoring values ​​and event records with time as the index, namely, a health event related dataset. The measured 3D visual model generation unit performs comprehensive analysis on the corresponding health indicator data in the health event-related dataset to generate a measured 3D visual model associated with the common time axis. Dynamic 3D Visual Model Generation Unit: Based on a common time axis, the measured 3D visual models within a continuous time period are continuously fused to generate a dynamic 3D visual model; Clinical trial visualization report generation unit: Generates a visualization report of the clinical trial based on the dynamic three-dimensional visual model; Physical health trend analysis unit: Collects patients' physical health data according to a preset period to form a physical health time series dataset. Based on the public time axis, the physical health time series dataset is associated with the health event data and health indicator data to construct a multidimensional health spatiotemporal dataset. The time series analysis algorithm is used to model the physical health trend to generate a physical health trend curve and extract the corresponding trend features. Fusion prediction unit: Constructs a prediction feature library containing physical characteristics, wherein the physical characteristics include at least the dynamic trend features extracted by the physical health trend analysis unit, and uses machine learning algorithms to train a prediction model that integrates physical factors to generate corrected prediction results; Visualization Interaction Unit: Establishes a visualization component library and provides a visualization configuration interface, allowing users to dynamically generate customized views through interactive operations, and integrates and displays the trend analysis results generated by the physical health trend analysis unit and the prediction results generated by the fusion prediction unit in the dynamic three-dimensional visual model in an overlay form.

[0017] In one embodiment, the specific steps for the measured 3D visual model generation unit to generate the measured 3D visual model are as follows: From the constructed health event-related dataset, extract health indicator data corresponding to each time point on the public timeline. The health indicator data includes at least laboratory test results data, vital sign data, and imaging test data. Align the health indicator data with patient identifiers and timestamps to ensure that different indicators at the same time point belong to the same patient. The extracted health indicator data is standardized to eliminate the dimensional differences between different indicators. At the same time, a mapping rule in three-dimensional space is assigned to each indicator. The mapping rule includes mapping the common time axis to the X-axis, mapping the indicator type (such as laboratory indicators and vital signs) to the Y-axis classification, mapping the specific value of the indicator to the Z-axis height or color depth, and encoding the individual patient or event type with different shapes or colors. Specifically: X-axis: Represents the common time axis T, with units such as days / weeks / months, and the direction from left to right indicates the passage of time; Y-axis: Represents standardized health indicator data values. This reflects the changes in the level of the indicator; Z-axis: Represents different categories of health indicators or individual patient identifiers. Depending on the specific needs, one of the following two mapping methods can be selected: Multiple indicators in parallel: different health indicators are mapped to different discrete positions on the Z-axis to form multiple parallel indicator planes; Multiple patient comparison: Mapping data from different patients under the same indicator to different positions on the Z-axis for cross-sectional comparison; For each data point This generates the corresponding point cloud in three-dimensional space. Based on standardized data points, a 3D surface reconstruction or scattered cloud rendering algorithm is used to generate a measured 3D visual model corresponding to a specific time point on a common time axis. This model can intuitively display the distribution status and abnormal fluctuation areas of all patients on multiple health indicators at that moment.

[0018] Specifically: Based on point cloud data, a continuous 3D visualization model is generated using B-spline surface fitting technology. For each fixed Z value (i.e., each indicator or patient), a 2D B-spline curve is constructed using the X-axis time and the Y-axis indicator value. ,in, As control points, Given a p-th order B-spline basis function, a trajectory of the index is generated in the measured three-dimensional visual model representing the change of the index over time through curve fitting. The trajectories of all the indices are arranged along the Z-axis, thus forming an index curtain in three-dimensional space. To demonstrate the relationship between multiple indicators, a connecting surface can be constructed between any two indicator planes. For example, Coons surface patches can be used to connect the points of adjacent indicators at the same time to form a three-dimensional solid feel. The dataset related to health events records the time points of each health event. And event type e, in the generated 3D model, at the corresponding time coordinate Special 3D markers, such as spheres, cubes, or specific icons, are embedded at locations (usually the average of all indicators or the location of a specific indicator). The color or shape of the marker represents different event types (e.g., a red sphere indicates a serious adverse event, and a yellow square indicates a minor discomfort). At the same time, the markers can be accompanied by interactive information, displaying event details when the mouse hovers over them.

[0019] In one embodiment, the specific steps for the dynamic 3D visual model generation unit to generate the dynamic 3D visual model are as follows: The measured 3D visual models, corresponding to multiple consecutive time points, are obtained from the measured 3D visual model generation unit in chronological order to form a model sequence. For the measured 3D visual models at adjacent time points, linear interpolation or spline interpolation algorithms are used to generate intermediate frame models in the time dimension to eliminate the sense of jump when switching models and achieve a smooth transition of changes in health indicators. The original model and the generated intermediate frame model are concatenated in sequence along a common timeline to form a coherent dynamic 3D visual model. At the same time, the model is bound to a timeline slider control, allowing users to observe the continuous change trend of health indicators over time by dragging the timeline.

[0020] In one embodiment, the physical health trend analysis unit includes one or more of the following physical health data collected at a preset period: body mass index, body fat percentage, muscle mass, basal metabolic rate, cardiorespiratory endurance, and bone density. The step of modeling the physical health trend using a time series analysis algorithm includes generating a physical health trend curve using a long short-term memory network or a dynamic time warping algorithm, and extracting the trend features including slope, fluctuation amplitude, and seasonality.

[0021] In one embodiment, the physical health trend analysis unit is further configured to perform correlation analysis between the trend characteristics and the probability of health events, identify physical change patterns associated with specific health events, and integrate the identified physical change patterns into the measured three-dimensional visual model or dynamic three-dimensional visual model in the form of overlay curves or heat maps.

[0022] In one embodiment, in the fusion prediction unit, the prediction feature library includes the patient's static physical characteristics, which include age, gender, and genetic background data. The step of training the prediction model that integrates physical factors using machine learning algorithms includes using an ensemble learning algorithm or a deep learning model, training it with the probability of health events or the quantitative value of care effects as labels, and performing stratified training for different physical types to obtain a physical-specific prediction sub-model.

[0023] In one embodiment, during real-time prediction, the fusion prediction unit matches the corresponding constitution-specific prediction sub-model based on the patient's current physical health data, or performs personalized fine-tuning of the general prediction model through transfer learning to generate prediction results that include constitution weights.

[0024] In one embodiment, the fusion prediction unit compares the generated corrected prediction results with the measured data and continuously updates the parameters of the prediction model using an online learning mechanism. The visualization interaction unit is also used to add display elements of physical constitution influencing factors to the visualization interface to intuitively show the contribution of physical constitution factors to the prediction results.

[0025] In the visualization interaction unit, the visualization component library includes basic charts and composite view templates. The basic charts include line charts, scatter plots, heat maps, and 3D surface charts. The composite view templates include time axis comparison charts, risk matrix charts, and constitution radar charts.

[0026] In the visualization interaction unit, the visualization configuration interface allows users to select health indicators, time ranges, patient groups or event types of interest by dragging and dropping, and set comparison dimensions. The system automatically generates customized views based on the configuration and supports multi-view linkage analysis.

[0027] The visualization interaction unit also includes a natural language interaction interface, which is used to receive the user's voice or text query intent and automatically generate corresponding visualization charts through semantic parsing technology, as well as embed "point of interest" marking function in the dynamic three-dimensional visual model, allowing users to add annotations to specific data points, save view status and generate personalized analysis reports.

[0028] The visualization interaction unit is also used to record the user's historical interaction behavior and intelligently recommend analysis dimensions based on the historical interaction behavior. At the same time, the high-risk areas automatically identified and highlighted by the physical health trend analysis unit based on the algorithm are integrated as preset view components into the user-customized visualization interface.

[0029] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A visual health monitoring platform, characterized in that, include: Monitoring data requirement matching unit: acquires clinical trial information and inputs it into a pre-set clinical trial requirement database to match the corresponding monitoring data requirements; Health indicator data acquisition unit: acquires health indicator data of all patients in the clinical trial, and preprocesses and organizes the health indicator data to construct a patient health indicator dataset; Historical health event data acquisition unit: acquires the patient's historical health event data, and associates the health indicator data with the health event data based on a pre-set public timeline to construct a health event-related dataset; The measured 3D visual model generation unit performs comprehensive analysis on the corresponding health indicator data in the health event-related dataset to generate a measured 3D visual model associated with the common time axis. Dynamic 3D Visual Model Generation Unit: Based on a common time axis, the measured 3D visual models within a continuous time period are continuously fused to generate a dynamic 3D visual model; Physical health trend analysis unit: Collects patients' physical health data according to a preset period to form a physical health time series dataset. Based on the public time axis, the physical health time series dataset is associated with the health event data and health indicator data to construct a multidimensional health spatiotemporal dataset. The time series analysis algorithm is used to model the physical health trend to generate a physical health trend curve and extract the corresponding trend features. Fusion prediction unit: Construct a prediction feature library containing physical characteristics, and use machine learning algorithms to train a prediction model that integrates physical factors to generate corrected prediction results; Visualization and Interaction Unit: Establishes a visualization component library and provides a visualization configuration interface, and integrates and displays the trend analysis results generated by the physical health trend analysis unit and the prediction results generated by the fusion prediction unit in the dynamic three-dimensional visual model in an overlay form.

2. The physical health visualization detection platform according to claim 1, characterized in that: The specific steps for the measured 3D visual model generation unit to generate the measured 3D visual model are as follows: From the existing health event-related dataset, extract health indicator data corresponding to each time point on the public timeline, and align the health indicator data using patient identifiers and timestamps; The extracted health indicator data is standardized to eliminate the dimensional differences between different indicators. At the same time, a mapping rule in three-dimensional space is assigned to each indicator. The mapping rule includes mapping the common time axis to the X-axis, mapping the indicator type to the Y-axis classification, mapping the specific value of the indicator to the Z-axis height or color depth, and encoding individual patients or event types with different shapes or colors. Based on the standardized data points, a 3D surface reconstruction or scattered cloud rendering algorithm is used to generate a measured 3D visual model corresponding to a specific time point on a common time axis.

3. The physical health visualization detection platform according to claim 1, characterized in that: The specific steps for the dynamic 3D visual model generation unit to generate the dynamic 3D visual model are as follows: The measured 3D visual models output by the measured 3D visual model generation unit are obtained in chronological order to form a model sequence; For the measured 3D visual models at adjacent time points, linear interpolation or spline interpolation algorithms are used to generate intermediate frame models in the time dimension; The original model and the generated intermediate frame model are concatenated in sequence along a common timeline to form a coherent dynamic 3D visual model.

4. The physical health visualization detection platform according to claim 1, characterized in that: In the physical health trend analysis unit, the physical health data collected according to a preset period includes body mass index, body fat percentage, muscle mass, basal metabolic rate, cardiorespiratory endurance and bone density. The modeling of physical health trends using time series analysis algorithms includes using long short-term memory networks or dynamic time warping algorithms to generate physical health trend curves and extracting trend features including slope, fluctuation amplitude and seasonality patterns.

5. The physical health visualization detection platform according to claim 1, characterized in that, The physical health trend analysis unit is also used to perform correlation analysis between the trend characteristics and the probability of health events, identify physical change patterns associated with specific health events, and integrate the identified physical change patterns into the measured three-dimensional visual model or dynamic three-dimensional visual model in the form of superimposed curves or heat maps.

6. The physical health visualization detection platform according to claim 1, characterized in that, In the fusion prediction unit, the prediction feature library includes the patient's static physical characteristics, which include age, gender, and genetic background data. The prediction model that integrates physical factors using machine learning algorithms includes training with ensemble learning algorithms or deep learning models, using the probability of health events or the quantitative value of nursing effects as labels, and performing stratified training for different physical types to obtain a physical-specific prediction sub-model.

7. The physical health visualization detection platform according to claim 6, characterized in that, During real-time prediction, the fusion prediction unit matches the corresponding constitution-specific prediction sub-model based on the patient's current physical health data to generate a prediction result that includes constitution weights.

8. The physical health visualization detection platform according to claim 1, characterized in that, The fusion prediction unit compares the generated corrected prediction results with the measured data and continuously updates the parameters of the prediction model using an online learning mechanism.

Images