A cognitive function testing device based on WebGL technology
Through the cognitive function testing device based on WebGL technology, the machine learning model is used to dynamically adjust the test type and difficulty, and the existing system needs to be processed manually and lacks intelligence, improving the accuracy and simplicity of the test, and suitable for daily life Clinical environment.
Patent Information
- Application Number
- CN202411245047.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-06
- Publication Date
- 2024-12-10
- Estimated Expiration
- 2044-09-06
AI Technical Summary
Existing cognitive function testing systems require assistance in manual experience processing and the testing tasks lack intelligence, so they cannot achieve true intelligent testing. The traditional methods take time and require professionals to score and explain, making them difficult to apply in daily clinical environments.
Using a cognitive function test device based on WebGL technology, the WebGL unit initializes the environment to generate test points, the main processing unit records interactive events, the adjustment unit dynamically adjusts the test type and difficulty based on the machine learning model, and the output unit combines medical analysis rules to output test results. The regression model is trained using the model training unit for data processing and feature extraction.
It realizes dynamic adjustment of test difficulty and content based on user real-time performance, improves the accuracy and pertinence of tests, simplifies the testing process, and reduces dependence on professionals.
Smart Images
Figure CN118749921B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of cognitive function testing, and in particular to a cognitive function testing device based on WebGL technology. Background Art
[0002] Hepatic encephalopathy causes cognitive dysfunction and emotional and behavioral abnormalities, and its symptoms can range from mild cognitive impairment to severe coma. Medical workers need to monitor and observe patients for a long time. Traditional cognitive function assessment methods often rely on complex psychometric tools or professional neuropsychological tests, which can be time-consuming and require professionally trained personnel to score and interpret, which is not always feasible in daily clinical settings. In order to overcome these limitations, researchers in recent years are developing some simpler, faster and easier-to-operate cognitive function test systems.
[0003] Although there are some tablet or smartphone-based applications that assess patients' cognitive functions through simple games or tasks, some existing solutions still require the assistance of manual experience processing. At the same time, the existing simple game tasks are not intelligent and are limited to copying paper operations to smart devices, rather than achieving true intelligent testing. Summary of the invention
[0004] In view of at least one shortcoming of the prior art, the present invention provides a cognitive function testing device based on WebGL technology, which effectively solves the problem of difficulty in existing cognitive function testing and analysis.
[0005] In order to solve the above technical problems, the present invention is solved by the following technical solutions:
[0006] A cognitive function testing device based on WebGL technology, comprising:
[0007] The WebGL unit initializes the WebGL environment in response to a received test instruction, and generates a plurality of test points within a preset coordinate range; the test instruction includes a test type, and the test type includes completing a plurality of consecutive interaction events according to a preset rule;
[0008] The main processing unit, in response to the interaction signal triggered by the graphical user interface, records the interaction event corresponding to each interaction signal, the interaction event includes coordinates and time; after completing multiple consecutive interaction events according to the preset rules, calculates the test data, the test data includes: total completion time, time and coordinates of each interaction event, accuracy, error rate and error type;
[0009] The adjustment unit, based on the trained machine learning model, inputs the test data of the current round, outputs the test type of the next round, the number, size and color of the test points; and renders the test points to the graphical user interface, repeating the implementation of the main processing unit;
[0010] The output unit outputs the cognitive function test result based on more than two test data or the last test data in combination with medical analysis rules.
[0011] It also includes a model training unit, which is used to train the regression model. The historical interaction data is cleaned, standardized and feature extracted before training the regression model.
[0012] Preferably, the model training unit trains the regression model, specifically including:
[0013] The number, color, size, coordinates, total completion time, time of each interaction event, accuracy, error rate, time of each error correction and error type of the current test points are input into the trained machine learning model, and the next test type and the number, color, size and coordinates of the test points corresponding to the test type are output. Based on more than two test data or the last test data, combined with medical analysis rules, the cognitive function test results are output.
[0014] Preferably, the WebGL unit implements a method for generating a plurality of test points within a preset coordinate range, comprising:
[0015] Predefine a test point data set, the test point data set includes the test point shape, size and color, randomly generate test point coordinates within a preset coordinate range, and draw test points on the randomly generated test point coordinates;
[0016] The number, size, shape and color of the test points are determined according to the output parameters of the machine learning model.
[0017] Preferably, completing multiple consecutive interaction events according to preset rules includes:
[0018] The test type includes a sequential connection digital test, which determines whether the interaction is correct by receiving the interaction signal. If the interaction is correct, the interaction signal is recorded and waits for the next trigger. If it is judged to be an interaction error, a correction instruction is issued and the correction time is recorded;
[0019] The test data of the completed sequential connection digital test is recorded, and the test data includes the total completion time, the time of each interactive event, the accuracy rate, the error rate, the time of each error correction, and the error type.
[0020] The WebGL unit includes a display unit, and the test point is displayed on the display unit. Optionally, the display unit includes a display screen or a touch screen, and the interactive signal includes a signal obtained by clicking the display screen or touching the touch screen with a mouse.
[0021] Preferably, the output unit implements a method for outputting cognitive function test results based on the last test data in combination with medical analysis rules, including:
[0022] The total completion time is substituted into the outlier analysis formula, which includes the age parameter and the outlier value of the corresponding age parameter. If it falls into the outlier range, it is judged as cognitive impairment.
[0023] The beneficial effects of the present invention are as follows: dynamically adjusting the test difficulty and content according to the user's real-time performance, integrating more than two test data or only using the adjusted last test data, especially combining the test data adjusted according to the previous test data, is conducive to improving the accuracy and pertinence of the test. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0025] Figure 1 It is a flow chart of cognitive function testing method based on WebGL technology.
[0026] Figure 2 It is a schematic diagram of the test points of test type NCT-A.
[0027] Figure 3 It is a schematic diagram of the test points of test type NCT-B. DETAILED DESCRIPTION
[0028] The present invention is further described in detail below in conjunction with embodiments. The following embodiments are for explanation of the present invention but the present invention is not limited to the following embodiments.
[0029] (I) A cognitive function testing device based on WebGL technology, comprising:
[0030] The WebGL unit initializes the WebGL environment in response to a received test instruction, and generates a plurality of test points within a preset coordinate range; the test instruction includes a test type, and the test type includes completing a plurality of consecutive interaction events according to a preset rule;
[0031] The main processing unit, in response to the interaction signal triggered by the graphical user interface, records the interaction event corresponding to each interaction signal, the interaction event includes coordinates and time; after completing multiple consecutive interaction events according to the preset rules, calculates the test data, the test data includes: total completion time, time and coordinates of each interaction event, accuracy, error rate and error type;
[0032] The adjustment unit, based on the trained machine learning model, inputs the test data of the current round and outputs the test type of the next round, which includes the number, size, color and coordinates of the test points; and renders the test points to the graphical user interface, repeating the implementation of the main processing unit;
[0033] The output unit outputs the cognitive function test result based on more than two test data or the last test data in combination with medical analysis rules.
[0034] A cognitive function testing device based on WebGL technology, further comprising a model training unit,
[0035] Used to train regression models. After cleaning, standardizing and feature extracting historical interaction data, the regression model is trained.
[0036] Among them, the model training unit trains the regression model, specifically including:
[0037] The number, color, size, coordinates, total completion time, time of each interaction event, accuracy, error rate, time of each error correction and error type of the current test points are input into the trained machine learning model, and the next test type and the number, color, size and coordinates of the test points corresponding to the test type are output. Based on more than two test data or the last test data, combined with medical analysis rules, the cognitive function test results are output.
[0038] The WebGL unit implements a method for generating a number of test points within a preset coordinate range, including:
[0039] Predefine a test point data set, the test point data set includes the test point shape, size and color, randomly generate test point coordinates within a preset coordinate range, and draw test points on the randomly generated test point coordinates;
[0040] The number, size, shape and color of the test points are determined according to the output parameters of the machine learning model.
[0041] Among them, completing multiple consecutive interaction events according to preset rules includes:
[0042] The test type includes a sequential connection digital test, which determines whether the interaction is correct by receiving the interaction signal. If the interaction is correct, the interaction signal is recorded and waits for the next trigger. If it is judged to be an interaction error, a correction instruction is issued and the correction time is recorded;
[0043] The test data of the completed sequential connection digital test is recorded, and the test data includes the total completion time, the time of each interactive event, the accuracy rate, the error rate, the time of each error correction, and the error type.
[0044] The WebGL unit includes a display unit, and the test point is displayed on the display unit. Specifically, the display unit includes a display screen or a touch screen, and the received interactive signal includes a signal obtained by clicking the display screen or touching the touch screen with a mouse.
[0045] The output unit implements a method for outputting cognitive function test results based on the last test data and in combination with medical analysis rules, including:
[0046] The total completion time is substituted into the outlier analysis formula, which includes the age parameter and the outlier value of the corresponding age parameter. If it falls into the outlier range, it is judged as cognitive impairment.
[0047] Alternatively, the average value of more than two test data or the weighted value (the time weight of the adjusted test is greater) is substituted into the outlier analysis formula.
[0048] (II) Correspondingly, a cognitive function testing method based on WebGL technology is disclosed, referring to Figure 1 As shown, including:
[0049] Step 1: In response to a received test instruction, the WebGL environment is initialized and a number of test points are generated within a preset coordinate range; the test instruction includes a test type, and the test type includes completing multiple consecutive interaction events according to preset rules.
[0050] Step two, in response to the interaction signal triggered by the graphical user interface, record the interaction event corresponding to each interaction signal, the interaction event includes coordinates and time; after completing multiple consecutive interaction events according to preset rules, calculate the test data, the test data includes: total completion time, time and coordinates of each interaction event, accuracy, error rate and error type.
[0051] Step three: Based on the machine learning model trained with historical test data, input the test data of the current round and output the test type of the next round, which includes the number, size, color and coordinates of the test points; render the test points to the graphical user interface and repeat step two.
[0052] Step 4: Output the cognitive function test results based on two or more test data or the last test data in combination with medical analysis rules.
[0053] For example, Figure 1 This embodiment sets two tests, NCT-A and NCT-B. The user selects and initiates NCT-A (digital connection test A) or NCT-B (digital connection test B) test command through the graphical user interface. In response to the received test instruction, the system initializes the WebGL environment accordingly.
[0054] The method of initializing the WebGL environment includes calling the getContext method to obtain the WebGL context to create a rendering environment.
[0055] Initialize WebGL settings, including clearing colors, depth testing, viewport settings, etc.
[0056] The method for generating a plurality of test points within a preset coordinate range includes:
[0057] Depending on the test type (NCT-A or NCT-B), a random series of test points are generated within the specified coordinate range.
[0058] The coordinates (x, y) of each test point are preset in advance and stored in an array for later use. Other solutions also include randomly generating test point coordinates according to preset rules, such as a certain distance between test points, a coordinate range, etc.
[0059] NCT-A is a random number from 1 to 25, such as Figure 2 ;
[0060] NCT-B arbitrarily arranges 1 to 13 and 1 to 12, such as Figure 3 .
[0061] Then draw the test points: use WebGL to draw the generated test points, and each test point is displayed on the canvas as a dot or other shapes, such as a square or irregular shape.
[0062] Call the vertex shader and fragment shader to render the test points to the screen. For example, in the default scene, NCT-A is a circle with white background and NCT-B is a circle with black background.
[0063] Then the user starts the test operation: the user starts the test by clicking the display screen or the touch screen with the mouse. The received interactive signals include the signals triggered by clicking the display screen and the touch screen with the mouse. The system monitors the user's interactive events in real time, responds to the interactive signals triggered by the graphical user interface, and records the interactive events corresponding to each interactive signal. The interactive events include the coordinates and click time of the user's click, where each click time (trigger time) can be used to calculate the user's reaction time, and the coordinates are used to calculate the accuracy and error rate data. This is to facilitate the subsequent analysis of the user's reaction speed and accuracy.
[0064] As a preferred solution, a method for completing multiple consecutive interaction events according to preset rules includes:
[0065] The test type includes a sequential connection digital test, which determines whether the interaction is correct by receiving the interaction signal. If the interaction is correct, the interaction signal is recorded and waits for the next trigger. If it is judged to be an interaction error, a correction instruction is issued and the correction time is recorded.
[0066] The test data of the completed sequential connection digital test is recorded, and the test data includes the total completion time, the time of each interactive event, the accuracy rate, the error rate, the time of each error correction, and the error type.
[0067] Complete multiple consecutive interactive events according to preset rules. Specifically,
[0068] Test type NCT-A Judgment requirements: Connect the numbers 1 to 25 in any order in sequence. If an error occurs during the connection process, correct it immediately and continue from the corrected point. Record the time required, including the time spent correcting the error.
[0069] Test type NCT-B judgment requirements: Connect 1 to 13 and 1 to 12 in the order of 1-1, 2-2. If an error occurs during the connection process, correct it immediately and continue from the corrected point. Record the time required, including the time spent correcting the error.
[0070] After completing multiple consecutive interaction events according to the preset rules, the test data is calculated. The test data includes: total completion time, time and coordinates of each interaction event, accuracy, error rate and error type. The test results of the current round are output based on the test data and the preset analysis rules.
[0071] Considering that different patients have different conditions, it is easy to make misjudgments if a fixed test type is used, so different test point generation rules need to be designed.
[0072] As a preferred solution, the method for generating a plurality of test points within a preset coordinate range includes:
[0073] Predefine a test point data set, the test point data set includes the test point shape, size and color, randomly generate test point coordinates within a preset coordinate range, and draw test points on the randomly generated test point coordinates;
[0074] The number, size, shape and color of the test points are determined according to the output parameters of the machine learning model.
[0075] The above method is to randomly generate the coordinates of the test points. A more accurate design also includes presetting coordinate parameters of different complexities. For example, for test points with higher complexity, the distance between the coordinates is smaller, and for test points with lower complexity, the distance between the coordinates is larger.
[0076] The number, color, size, coordinates, total completion time, time of each interaction event, accuracy, error rate, time of each error correction and error type of the current test points are input into the trained machine learning model, and the next test type and the number, color, size and coordinates of the test points corresponding to the test type are output. Based on more than two test data or the last test data, combined with medical analysis rules, the cognitive function test results are output.
[0077] Training parameters may also include whether to display in Chinese or in numbers.
[0078] The training parameters may also include complexity parameters. For example, the complexity can be calculated based on parameters such as the number, color, size and coordinates of the test points. For example, 1-10 is used as the complexity parameter. The complexity parameter calculated based on the number, whether the color is easy to recognize, the size of the test points, etc. is between 1-10.
[0079] The total completion time includes error correction time.
[0080] The training method of the machine learning model includes: based on the regression model, collecting historical interaction data, cleaning, standardizing and extracting features from the historical interaction data, and then using it to train the regression model. The regression model is used to perform machine learning training on the feature data. The model is used to predict the user's performance in real time and dynamically adjust the test difficulty and content.
[0081] Model adjustments include: adjusting the size and coordinates of test points according to user operations, adjusting the number of test points according to the user's accuracy and error rate, and dynamically adjusting the order of test points according to the user's error type and distribution. For example, when the user's accuracy rate is low, cross-validation combining multiple methods such as changing the size of graphics, text size, and graphics order can reduce the error results caused by noise.
[0082] Furthermore, in order to further increase the diversity and adaptability of the test, the shape complexity and color contrast of the test points can be adjusted according to the user's performance, and the size, shape and / or color of the test points can be adjusted. For example, if the user performs well, complex polygonal or irregular shaped test points are generated, and if the performance is poor, simple circular or square test points are generated; if the user performs well under high contrast, the contrast can be reduced, and if the performance is poor under low contrast, the contrast can be increased.
[0083] In addition to the recorded historical data, the collected real-time data and the adjusted effects are used to further train and optimize the machine learning model. By continuously updating the model, the prediction accuracy and adjustment effect of the model are improved, ensuring that the test system can continue to adapt to the needs of different users and provide more accurate testing and evaluation. This dynamic adjustment mechanism helps to comprehensively evaluate the user's cognitive function, especially in the cognitive evaluation of patients with hepatic encephalopathy, improving the effectiveness and reliability of the test.
[0084] Among them, the medical analysis rules involve clinical experience rules, which need to be collected and sorted in advance from researchers, medical staff, etc. in related fields, which will not be elaborated here. At the same time, different medical institutions have different medical analysis rules. One of the models is disclosed as an example: the total completion time (including error correction time) is substituted into the outlier analysis formula. The outlier analysis formula includes age parameters and outliers corresponding to age parameters. If it falls into the outlier range, it is judged as impaired cognitive function. For example, substitute x seconds into the following formula:
[0085] Abnormal values: age <35 years, time >34.3 s; 35-44 years, time >45.7 s; 45-54 years, time >52.8 s; 55-64 years, time >61.9 s;
[0086] This embodiment only lists a simple formula for explanation, and the actual analysis may be more complicated. At the same time, different institutions may adopt different test type templates, so the corresponding parameters also need to be adaptively modified.
[0087] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. The units may or may not be physically separated, and the components displayed as units may be one physical unit or multiple physical units, that is, they may be located in one place, or they may be distributed in multiple different places. Some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.
Claims
1. A cognitive function testing device based on WebGL technology, It is characterized in that include: The WebGL unit, in response to the received test instruction, initializes the WebGL environment and generates a number of test points within a preset coordinate range; The test instructions include a test type, and the test type includes completing a plurality of consecutive interaction events according to a preset rule; The main processing unit records the interaction event corresponding to each interaction signal in response to the interaction signal triggered by the graphical user interface, wherein the interaction event includes coordinates and time; After completing multiple consecutive interaction events according to preset rules, calculate the test data; Among them, the test type includes a sequential connection digital test, which determines whether the interaction is correct through the received interaction signal. If the interaction is correct, the interaction signal is recorded and waits for the next trigger. If it is judged to be an interaction error, a correction instruction is issued and the correction time is recorded; Record the test data of the completed sequential connection digital test, including the total completion time, the time and coordinates of each interaction event, the accuracy rate, the error rate, the time of each error correction and the error type; The adjustment unit, based on the trained machine learning model, inputs the test data of the current round, outputs the test type of the next round, the number, size and color of the test points; and renders the test points to the graphical user interface, repeating the implementation of the main processing unit; The output unit outputs the cognitive function test result based on more than two test data or the last test data in combination with medical analysis rules.
2. A cognitive function testing device based on WebGL technology according to claim 1, It is characterized in that The WebGL unit implements a method to generate several test points within a preset coordinate range, including: Predefine a test point data set, the test point data set includes the test point shape, size and color, randomly generate test point coordinates within a preset coordinate range, and draw test points on the randomly generated test point coordinates; The number, size, shape and color of the test points are determined according to the output parameters of the machine learning model.
3. A cognitive function testing device based on WebGL technology according to claim 1, It is characterized in that It also includes a model training unit, where the machine learning model is implemented as a regression model. The model training unit is used to train the regression model. The historical interaction data is cleaned, standardized, and feature extracted before training the regression model.
4. A cognitive function testing device based on WebGL technology according to claim 3, It is characterized in that The model training unit trains the regression model, including: The number, color, size, coordinates, total completion time, time of each interaction event, accuracy, error rate, time of each error correction and error type of the current test points are input into the trained machine learning model, and the next test type and the number, color, size and coordinates of the test points corresponding to the test type are output. Based on more than two test data or the last test data, combined with medical analysis rules, the cognitive function test results are output.
5. The cognitive function testing device based on WebGL technology according to claim 1, It is characterized in that The WebGL unit includes the display unit, and the test points are displayed in the display unit.
6. The cognitive function testing device based on WebGL technology according to claim 1, It is characterized in that The output unit implements a method for outputting cognitive function test results based on the last test data and in combination with medical analysis rules, including: The total completion time is substituted into the outlier analysis formula, which includes the age parameter and the outlier value of the corresponding age parameter. If it falls into the outlier range, it is judged as cognitive impairment.
7. The cognitive function testing device based on WebGL technology according to claim 5, It is characterized in that The display unit includes a display screen or a touch screen, and the interactive signal includes a signal obtained by clicking the display screen with a mouse or touching the touch screen.