A data processing method and system

Abstract

The application discloses a data processing method and system, and relates to the technical field of artificial intelligence and biomedical cross. The method comprises the following steps: firstly, grouping patients according to the Hamilton Depression Scale; secondly, collecting the judgment results of visual stimulation patterns of the participants and the presentation duration of the visual stimulation patterns; thirdly, obtaining the multi-dimensional visual motion perception characteristic indexes of the participants according to the collected results; fourthly, obtaining two subgroups of all participants in the patient group through K-Means clustering analysis, and determining the visual motion perception abnormal subgroup in the two subgroups through index comparison and analysis; and finally, applying the positive sample data of the abnormal subgroup and the negative sample data of the non-abnormal subgroup, performing rate verification modeling based on a machine learning algorithm, and obtaining a visual motion perception abnormality recognition model, so that the visual motion perception abnormality recognition model for MDD patient sub-group processing based on visual motion perception data of patients with depressive disorders can be obtained.

Description

Technical Field

[0001] This invention belongs to the field of interdisciplinary technology of artificial intelligence and biomedicine, and specifically relates to a data processing method and system. Background Technology

[0002] Major Depressive Disorder (MDD) is one of the most significant challenges in the global mental health field. Research has revealed that the greatest challenge lies in the diversity of symptoms and pathophysiology among patients with this disorder: MDD patients exhibit significant differences in clinical presentation, disease course, treatment response, genetics, and neurobiology. Therefore, utilizing the diverse characteristics of MDD patients across these aspects to cluster and subgroup them may play a crucial role in future research.

[0003] Current research indicates that visual-motor inhibition is weakened in individuals diagnosed with major depressive disorder (MDD). However, most existing MDD patient subgrouping techniques cluster and subgroup based on cognitive and somatic dimensions (e.g., existing technologies such as CN114966053A, Diagnostic System for Depression Based on Serum Proteins and Brain Function Imaging Indicators, and CN104270942A, Non-human Animal Model of Depression and Its Usage). In other words, the problem of clustering MDD patients based on visual-motor perception data has not yet been solved. Summary of the Invention

[0004] The purpose of this invention is to provide a data processing method and system to solve the problem that existing MDD patient subgrouping techniques fail to address the issue of clustering MDD patients using visual motion perception data.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] Firstly, a data processing method is provided, including:

[0007] Participants were invited to complete the Hamilton Depression Rating Scale, and their scores were quantified based on the scale. The scores were then used to determine whether they exceeded a preset threshold. If so, the participants were assigned to the depressive disorder patient group; otherwise, they were assigned to the healthy group.

[0008] Collect the participants' judgment results on the visual stimulus pattern and the presentation duration of the visual stimulus pattern, and statistically obtain the participants' multidimensional visual motion perception characteristic index based on the collected results. The visual stimulus pattern refers to the pattern of visual stimulus pattern that drifts to the left or right in the central area of ​​the display screen, and the judgment result refers to the judgment of whether the drift direction of the corresponding visual stimulus pattern is to the left or to the right.

[0009] Based on the multidimensional visual-motor perception feature indicators of all participants belonging to the depressive disorder patient group, two subgroups of all participants belonging to the depressive disorder patient group were obtained through K-Means clustering analysis;

[0010] For each subgroup in the two subgroup groups, the multidimensional visual motion perception feature index of all the corresponding participants is compared and analyzed with the multidimensional visual motion perception feature index of all the participants belonging to the healthy group to determine whether the corresponding subgroup group is a visual motion perception abnormal group.

[0011] The multidimensional visual motion perception feature index of the participants who belong to the visual motion perception abnormality group is used as positive sample data, and the multidimensional visual motion perception feature index of the participants who do not belong to the visual motion perception abnormality group is used as negative sample data.

[0012] Based on all the positive sample data and all the negative sample data, a calibration and verification model is performed using a machine learning algorithm to obtain a visual motion perception anomaly recognition model.

[0013] Based on the above-mentioned invention, a novel scheme is provided for using machine learning and cluster analysis techniques to process medical data and construct a visual-motor perception anomaly identification model. First, participants are grouped according to the Hamilton Depression Rating Scale. Then, participants' judgments on visual stimulus patterns and the duration of these patterns are collected. Based on the collected results, multidimensional visual-motor perception characteristic indicators are statistically obtained. Next, K-Means cluster analysis is used to obtain two subgroups for all participants in the patient group. Visual-motor perception anomaly groups within these two subgroups are identified through indicator comparison analysis. Finally, positive sample data from the anomalous groups and negative sample data from the non-abnormal groups are used to calibrate and validate the model based on machine learning algorithms, resulting in a visual-motor perception anomaly identification model. This provides a visual-motor perception anomaly identification model for subgrouping MDD patients based on visual-motor perception data, which can assist in the formulation of personalized diagnosis and treatment plans, facilitating practical application and promotion.

[0014] In one possible design, the display screen, when presenting the visual stimulus pattern, is at the same level as the participant's head and 45-47 cm away from the participant's eyes.

[0015] In one possible design, the visual stimulus pattern includes alternating large visual stimulus patterns and small visual stimulus patterns, wherein the large visual stimulus pattern refers to a pattern in which a large sinusoidal grating pattern with blurred edges drifts to the left / right in the central area of ​​the display screen, and the small visual stimulus pattern refers to a pattern in which a small sinusoidal grating pattern with blurred edges drifts to the left / right in the central area of ​​the display screen.

[0016] In one possible design, the participants' multidimensional visual motion perception feature indicators are statistically obtained based on the collected results, including:

[0017] Based on the collected results, the participants' judgment results on each large visual stimulus pattern and the presentation duration of each large visual stimulus pattern were extracted. The participants' judgment results on each small visual stimulus pattern and the presentation duration of each small visual stimulus pattern were also extracted.

[0018] The presentation time difference of each pair of adjacent large visual stimulus patterns is calculated based on the presentation duration of each large visual stimulus pattern to obtain a large visual stimulus pattern presentation time difference sequence. The presentation time difference of each pair of adjacent small visual stimulus patterns is calculated based on the presentation duration of each small visual stimulus pattern to obtain a small visual stimulus pattern presentation time difference sequence. The adjacent large visual stimulus patterns include a large visual stimulus pattern used in the previous trial and a large visual stimulus pattern used in the subsequent trial that are adjacent in the trial time sequence. The adjacent small visual stimulus patterns include a small visual stimulus pattern used in the previous trial and a small visual stimulus pattern used in the subsequent trial that are adjacent in the trial time sequence.

[0019] Based on the participants' judgments of each large visual stimulus pattern and the presentation duration of each large visual stimulus pattern, the participants' judgments of each small visual stimulus pattern and the presentation duration of each small visual stimulus pattern, the presentation time difference sequence of the large visual stimulus pattern and / or the presentation time difference sequence of the small visual stimulus pattern, a multidimensional visual motion perception feature index is statistically obtained, including any one or any combination of the following indicators (A) to (I):

[0020] (A) The fitting time threshold durationS with 75% accuracy in judging small visual stimulus patterns;

[0021] (B) The fitting time threshold durationL with 75% accuracy in judging large visual stimulus patterns;

[0022] (C) Indicator

[0023] (D) Coefficient of variation (cvS) of small visual stimulus patterns presenting continuous time lag sequences;

[0024] (E) The coefficient of variation (cvL) of the sustained time lag sequence in the large visual stimulus pattern;

[0025] (F) The standard deviation sdS of the continuous time-difference sequence of small visual stimulus patterns;

[0026] (G) The standard deviation sdL of the continuous time-difference sequence of large visual stimulus patterns;

[0027] (H) The mean S of the continuous time difference sequence of small visual stimulus patterns;

[0028] (I) The mean L of the continuous time difference sequence of the large visual stimulus pattern.

[0029] In one possible design, based on the multidimensional visual-motor perception feature indices of all participants belonging to the depressive disorder patient group, K-Means clustering analysis is used to obtain two subgroups of all participants belonging to the depressive disorder patient group, including:

[0030] Principal component analysis was performed on the multidimensional visual-motor perception feature indicators of all participants belonging to the depression disorder patient group to reduce the dimensionality of the multidimensional visual-motor perception feature indicators, resulting in the dimensionality-reduced multidimensional visual-motor perception feature indicators of all participants belonging to the depression disorder patient group.

[0031] Based on the dimensionality-reduced multidimensional visual-motor perception feature indexes of all participants belonging to the depressive disorder patient group, two subgroups of all participants belonging to the depressive disorder patient group were obtained through K-Means clustering analysis.

[0032] In one possible design, based on all the positive sample data and all the negative sample data, a calibration and verification model is performed using a machine learning algorithm to obtain a visual motion perception anomaly recognition model, including:

[0033] Based on all the positive sample data and all the negative sample data, calibration and verification modeling are performed on different machine learning algorithms to obtain multiple visual motion perception anomaly recognition models that correspond one-to-one with the machine learning algorithms.

[0034] The visual motion perception anomaly recognition model with the highest recognition accuracy is selected from the multiple visual motion perception anomaly recognition models as the final visual motion perception anomaly recognition model obtained from the modeling.

[0035] In one possible design, the various machine learning algorithms include support vector machines, K-nearest neighbors, decision trees, random forests, and / or leave-one-out cross-validation.

[0036] In one possible design, after obtaining two subgroups of all participants belonging to the said depressive disorder patient group, the method further includes:

[0037] For each pair of subgroups within the two subgroup groups, the Hamilton Depression Rating Scale was compared for all the corresponding participants to identify the symptom differences among patients with depressive disorders.

[0038] Secondly, a data processing system is provided, including a first human-computer interaction subsystem, a second human-computer interaction subsystem, and computer equipment;

[0039] The first human-computer interaction subsystem is communicatively connected to the computer device and is used to invite participants to complete the Hamilton Depression Rating Scale and transmit the Hamilton Depression Rating Scale to the computer device.

[0040] The second human-computer interaction subsystem is communicatively connected to the computer device and is used to collect the participant's judgment results on the visual stimulus pattern and the presentation duration of the visual stimulus pattern, and transmit the collection results to the computer device. The visual stimulus pattern refers to a pattern in which a visual stimulus pattern that drifts to the left or right is presented in the central area of ​​the display screen, and the judgment result refers to judging whether the drift direction of the corresponding visual stimulus pattern is to the left or to the right.

[0041] The computer device is used to perform the following data processing steps:

[0042] The participant's score was quantified using the Hamilton Depression Rating Scale. Then, it was determined whether the score exceeded a preset threshold. If so, the participant was assigned to the depressive disorder patient group; otherwise, the participant was assigned to the healthy group.

[0043] Based on the collected results, the multidimensional visual motion perception characteristic index of the participants was statistically obtained.

[0044] Based on the multidimensional visual-motor perception feature indicators of all participants belonging to the depressive disorder patient group, two subgroups of all participants belonging to the depressive disorder patient group were obtained through K-Means clustering analysis;

[0045] For each subgroup in the two subgroup groups, the multidimensional visual motion perception feature index of all the corresponding participants is compared and analyzed with the multidimensional visual motion perception feature index of all the participants belonging to the healthy group to determine whether the corresponding subgroup group is a visual motion perception abnormal group.

[0046] The multidimensional visual motion perception feature index of the participants who belong to the visual motion perception abnormality group is used as positive sample data, and the multidimensional visual motion perception feature index of the participants who do not belong to the visual motion perception abnormality group is used as negative sample data.

[0047] Based on all the positive sample data and all the negative sample data, a calibration and verification model is performed using a machine learning algorithm to obtain a visual motion perception anomaly recognition model.

[0048] In one possible design, the second human-computer interaction subsystem includes the display screen, keyboard, voice speaker, and control device;

[0049] The display screen is communicatively connected to the control device and is used to present the visual stimulus pattern under the control of the control device.

[0050] The keyboard is communicatively connected to the control device and is used to input the participant's judgment result on the visual stimulus pattern and transmit the judgment result to the control device.

[0051] The loudspeaker is communicatively connected to the control device and is used to emit sound under the control of the control device.

[0052] The control device is communicatively connected to the computer device and is used to collect the participant's judgment results on the visual stimulus pattern and the presentation duration of the visual stimulus pattern, and transmit the collected results to the computer device. When the judgment result is found to be incorrect, the control device transmits a sound control command to the speaker to control the speaker to emit an error prompt sound.

[0053] The beneficial effects of the above scheme are:

[0054] (1) This invention provides a new scheme for medical data processing based on machine learning and cluster analysis to construct a visual motion perception abnormality identification model. First, participants are grouped according to the Hamilton Depression Rating Scale. Then, the participants' judgment results on visual stimulus patterns and the duration of presentation of visual stimulus patterns are collected. Based on the collected results, the participants' multidimensional visual motion perception characteristic indicators are statistically obtained. Then, K-Means cluster analysis is used to obtain two subgroups of all participants in the patient group. The abnormal visual motion perception groups in the two subgroups are determined by index comparison analysis. Finally, positive sample data of the abnormal group and negative sample data of the non-abnormal group are used to perform calibration and verification modeling based on machine learning algorithm to obtain the visual motion perception abnormality identification model. In this way, a visual motion perception abnormality identification model can be obtained for MDD patients with depressive disorders based on visual motion perception data to subgroup them. This can help assist in the formulation of personalized diagnosis and treatment plans and facilitate practical application and promotion.

[0055] (2) Personalized treatment can be carried out by using cluster analysis to divide MDD patients into different subgroups, which helps to develop personalized treatment plans based on the different characteristics of each subgroup and improve the treatment effect.

[0056] (3) It helps to discover the cause of the disease. That is, cluster analysis can help reveal the characteristic differences between different subgroups, which helps to understand the causes and mechanisms of depression more deeply.

[0057] (4) It helps to improve the accuracy of diagnosis, that is, dividing patients with depression into different subgroups helps to improve the accuracy of diagnosis and provides doctors with more reliable diagnostic basis. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 This is a flowchart illustrating the data processing method provided in an embodiment of this application.

[0060] Figure 2 Example diagram of visual stimulus patterns provided in embodiments of this application, wherein, Figure 2 (a) shows an example diagram of a large sinusoidal grating pattern. Figure 2 (b) shows an example diagram of a small sinusoidal grating pattern.

[0061] Figure 3 An example diagram showing the subgrouping results of MDD patients obtained from K-Means clustering analysis, provided for embodiments of this application.

[0062] Figure 4 An example diagram showing the index analysis results based on principal component analysis provided in this application embodiment.

[0063] Figure 5 This is an example figure showing the comparative analysis results of indicators between two MDD patient subgroups and a healthy group provided in the embodiments of this application.

[0064] Figure 6 This is a schematic diagram of the structure of the data processing system provided in the embodiments of this application. Detailed Implementation

[0065] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in conjunction with the accompanying drawings and descriptions of the embodiments or the prior art. Obviously, the following description of the structure of the accompanying drawings is only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained based on these embodiments without creative effort. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention.

[0066] It should be understood that although the terms "first" and "second", etc., may be used herein to describe various objects, these objects should not be limited by these terms. These terms are only used to distinguish one object from another. For example, the first object may be referred to as the second object, and similarly, the second object may be referred to as the first object, without departing from the scope of the exemplary embodiments of the invention.

[0067] It should be understood that the term "and / or" that may appear in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, B exists alone, or A and B exist simultaneously. Another example is A, B and / or C, which can mean that any one of A, B, and C or any combination thereof exists. The term " / and" that may appear in this document describes another relationship between related objects, indicating that two relationships can exist. For example, A / and B can mean: A exists alone or A and B exist simultaneously. In addition, the character " / " that may appear in this document generally indicates that the related objects before and after it are in an "or" relationship.

[0068] Example

[0069] like Figure 1As shown, the data processing method provided in the first aspect of this embodiment may include, but is not limited to, the following steps S1 to S6.

[0070] S1. Invite participants to complete the Hamilton Depression Rating Scale, and quantify the participants' scores based on the Hamilton Depression Rating Scale. Then determine whether the scores exceed a preset threshold. If so, classify the participants into the depressive disorder patient group; otherwise, classify the participants into the healthy group.

[0071] In step S1, the Hamilton Depression Scale (HAMD), developed by Hamilton in 1960, is the most widely used scale in clinical practice for assessing depressive states. Specifically, it can be completed and quantified using conventional question-and-answer methods. Furthermore, the preset threshold is used as a basis for determining whether a participant has a depressive disorder; a specific example, but not limited to, is 17 points.

[0072] S2. Collect the participants' judgment results on the visual stimulus pattern and the presentation duration of the visual stimulus pattern, and statistically obtain the participants' multidimensional visual motion perception characteristic index based on the collected results. The visual stimulus pattern refers to the pattern of visual stimulus pattern that drifts to the left or right in the central area of ​​the display screen, and the judgment result refers to judging whether the drift direction of the corresponding visual stimulus pattern is to the left or to the right.

[0073] In step S2, the multidimensional visual motion perception feature index is used as visual motion perception data (i.e., psychophysical experimental data). The specific collection process includes, but is not limited to, the following details: (1) Before the experiment begins, the participants are required to sit upright in the room, ensuring their heads are at the same level as the display screen; (2) Throughout the experiment, the participants' eyes must be 45-47 cm away from the display screen; (3) The experimental process is explained to the participants, and tests are conducted to confirm their understanding; (4) After preparation, all lights are turned off, and the test officially begins; (5) During the experiment, the participants are asked to stare at a black dot in the center of the screen. After the black dot disappears, the participants observe the visual stimulus pattern presented in the center of the screen and determine the drift direction of the visual stimulus pattern; (6) The participants' judgment of the visual stimulus pattern and the duration of its presentation are recorded. Based on the aforementioned details, it is known that when presenting the visual stimulus pattern, the display screen must be at the same level as the participants' heads and 45-47 cm away from their eyes to ensure the reliability of data collection.

[0074] In step S2, specifically, the visual stimulus pattern includes, but is not limited to, alternating large visual stimulus patterns and small visual stimulus patterns. The large visual stimulus pattern refers to a pattern in which a large sinusoidal grating pattern with blurred edges and drifting left / right is presented in the central area of ​​the display screen. The small visual stimulus pattern refers to a pattern in which a small sinusoidal grating pattern with blurred edges and drifting left / right is presented in the central area of ​​the display screen. Examples of the aforementioned large and small sinusoidal grating patterns are as follows: Figure 2 As shown. In detail, the specific parameters of the aforementioned experiment can be designed, but are not limited to, as follows: (11) Screen linear calibration: The background brightness of the screen is set to 56 cd / m². 2 (12) The visual stimulus pattern is composed of a sinusoidal grating pattern with the following parameters: contrast of 50%, spatial frequency of 1 cycle / degree, drift speed of 4° / s, direction of motion to the left or right, diameter of large sinusoidal grating pattern of 5°, diameter of small sinusoidal grating pattern of 1°, the edge of sinusoidal grating pattern is blurred and implemented using a Gaussian function, and the blur width is 30%; (13) The experimental procedure for presenting the visual stimulus pattern is implemented using the Psychtoolbox in MATLAB software; (14) The duration of the visual stimulus pattern presentation is adaptively adjusted using a three-step-up ladder procedure; (15) Participants use the keyboard to input whether the drift direction of the visual stimulus pattern is to the left or right, and a "beep" sound is emitted when the judgment is wrong, and no sound is emitted when the judgment is correct.

[0075] In step S2, more specifically, the multidimensional visual motion perception feature index of the participants is statistically obtained based on the collected results, including but not limited to the following steps S21 to S23.

[0076] S21. Based on the collected results, extract the participants' judgment results on each large visual stimulus pattern and the presentation duration of each large visual stimulus pattern, and also extract the participants' judgment results on each small visual stimulus pattern and the presentation duration of each small visual stimulus pattern.

[0077] In step S21, for example, if the participant performs 160 visual stimulus tests, the following results will be collected due to the alternating large and small visual stimulus patterns: the participant's judgment results on 80 large visual stimulus patterns, the presentation duration of the 80 large visual stimulus patterns, the participant's judgment results on 80 small visual stimulus patterns, and the presentation duration of the 80 small visual stimulus patterns.

[0078] S22. Calculate the presentation time difference of each pair of adjacent large visual stimulus patterns based on the presentation duration of each large visual stimulus pattern to obtain a large visual stimulus pattern presentation time difference sequence; and calculate the presentation time difference of each pair of adjacent small visual stimulus patterns based on the presentation duration of each small visual stimulus pattern to obtain a small visual stimulus pattern presentation time difference sequence, wherein the adjacent large visual stimulus patterns include a large visual stimulus pattern used in the previous trial and a large visual stimulus pattern used in the subsequent trial that are adjacent in the trial time sequence, and the adjacent small visual stimulus patterns include a small visual stimulus pattern used in the previous trial and a small visual stimulus pattern used in the subsequent trial that are adjacent in the trial time sequence.

[0079] In step S22, based on the example in step S21, the presentation time difference of 79 pairs of adjacent large visual stimulus patterns can be calculated according to the presentation duration of the 80 large visual stimulus patterns, resulting in a large visual stimulus pattern presentation time difference sequence containing the 79 large visual stimulus pattern presentation time differences; and the presentation time difference of 79 pairs of adjacent small visual stimulus patterns can also be calculated according to the presentation duration of the 80 small visual stimulus patterns, resulting in a small visual stimulus pattern presentation time difference sequence containing the 79 small visual stimulus pattern presentation time differences.

[0080] S23. Based on the participants' judgment results on each large visual stimulus pattern and the presentation duration of each large visual stimulus pattern, the participants' judgment results on each small visual stimulus pattern and the presentation duration of each small visual stimulus pattern, the presentation time difference sequence of the large visual stimulus pattern and / or the presentation time difference sequence of the small visual stimulus pattern, a multidimensional visual motion perception feature index is statistically obtained, including but not limited to any one or any combination of the following indicators (A) to (I):

[0081] (A) The fitting time threshold durationS with 75% accuracy in judging small visual stimulus patterns;

[0082] (B) The fitting time threshold durationL with 75% accuracy in judging large visual stimulus patterns;

[0083] (C) Indicator

[0084] (D) Coefficient of variation (cvS) of small visual stimulus patterns presenting continuous time lag sequences;

[0085] (E) The coefficient of variation (cvL) of the sustained time lag sequence in the large visual stimulus pattern;

[0086] (F) The standard deviation sdS of the continuous time-difference sequence of small visual stimulus patterns;

[0087] (G) The standard deviation sdL of the continuous time-difference sequence of large visual stimulus patterns;

[0088] (H) The mean S of the continuous time difference sequence of small visual stimulus patterns;

[0089] (I) The mean L of the continuous time difference sequence of the large visual stimulus pattern.

[0090] In step S23, the fitting time threshold durationS with 75% accuracy for the judgment result of the small visual stimulus pattern can be conventionally obtained by using existing S-curve fitting methods based on the participant's judgment result of each small visual stimulus pattern and the presentation duration of each small visual stimulus pattern. Similarly, the fitting time threshold durationL with 75% accuracy for the judgment result of the large visual stimulus pattern can also be obtained by using existing S-curve fitting methods based on the participant's judgment result of each large visual stimulus pattern and the presentation duration of each large visual stimulus pattern. The coefficient of variation (cvS), standard deviation (sdS), and mean (meanS) of the small visual stimulus pattern presentation duration sequence are obtained through conventional fitting using curve fitting methods. Similarly, the coefficient of variation (cvL), standard deviation (sdL), and mean (meanL) of the large visual stimulus pattern presentation duration sequence are obtained through conventional statistics based on the large visual stimulus pattern presentation duration sequence.

[0091] S3. Based on the multidimensional visual-motor perception feature indicators of all participants belonging to the depressive disorder patient group, K-Means clustering analysis is used to obtain two subgroups of all participants belonging to the depressive disorder patient group.

[0092] In step S3, the K-Means clustering analysis is a common existing clustering algorithm that divides the dataset into K predefined clusters, making the points within each cluster as close together as possible, while the points between different clusters are as far apart as possible. Figure 3As shown, all participants belonging to the depressive disorder patient group can be clustered into two subgroups: MDD patient subgroup 1, represented by green dots, and MDD patient subgroup 2, represented by red dots. To accelerate cluster analysis and reduce the computational resources required for the analysis, preferably, based on the multidimensional visual-motor perception feature indicators of all participants belonging to the depressive disorder patient group, K-Means cluster analysis is used to obtain the two subgroups of all participants belonging to the depressive disorder patient group, including but not limited to the following steps S31-S32.

[0093] S31. Perform principal component analysis on the multidimensional visual-motor perception feature indicators of all participants belonging to the depression disorder patient group to reduce the dimensionality of the multidimensional visual-motor perception feature indicators, and obtain the dimensionality-reduced multidimensional visual-motor perception feature indicators of all participants belonging to the depression disorder patient group.

[0094] In step S31, principal component analysis is a statistical method used to reduce the dimensionality of data by linearly combining the original variables while retaining as much information as possible. For example, when the multidimensional visual motion perception feature indicators include indicators (A) to (I) from step S22, the corresponding principal component analysis results can be as follows: Figure 4 As shown ( Figure 4 The nine horizontal indicators are nine principal components, rather than directly corresponding to indicators (A) to (I). The principal components can be understood as projections of the original data. Thus, the first five principal components from the left can be selected as the multidimensional visual motion perception feature indicators after dimensionality reduction (i.e., reduced from 9 dimensions to five dimensions).

[0095] S32. Based on the dimensionality-reduced multidimensional visual-motor perception feature index of all participants belonging to the depressive disorder patient group, K-Means clustering analysis is used to obtain two subgroups of all participants belonging to the depressive disorder patient group.

[0096] S4. For each subgroup in the two subgroup groups, compare and analyze the multidimensional visual motion perception feature indicators of all the participants in the corresponding subgroup with the multidimensional visual motion perception feature indicators of all the participants in the healthy group to determine whether the corresponding subgroup group is a visual motion perception abnormal group.

[0097] In step S4, the comparative analysis results of the corresponding indicators for the MDD patient subgroup 1 and MDD patient subgroup 2 are as follows: Figure 5As shown, it can be seen that the multidimensional visual-motor perception feature indicators of all participants belonging to the MDD patient subgroup 1 are significantly different from those of all participants belonging to the healthy group, while the multidimensional visual-motor perception feature indicators of all participants belonging to the MDD patient subgroup 2 are basically no different from those of all participants belonging to the healthy group. Therefore, the MDD patient subgroup 1 can be identified as a visual-motor perception abnormal group, while the MDD patient subgroup 2 can be identified as not being a visual-motor perception abnormal group.

[0098] S5. The multidimensional visual motion perception feature index of the participants who belong to the visual motion perception abnormality group is taken as positive sample data, and the multidimensional visual motion perception feature index of the participants who do not belong to the visual motion perception abnormality group is taken as negative sample data.

[0099] In step S5, since the subsequent step is to obtain a visual motion perception anomaly recognition model, it is necessary to use the multidimensional visual motion perception feature indicators of the participants belonging to the visual motion perception anomaly group as positive sample data: that is, the multidimensional visual motion perception feature indicators of the participants belonging to the visual motion perception anomaly group (i.e., the MDD patient subgroup 1) as the model input items for positive samples, and assign the model output item of the positive samples a value of 1; and it is necessary to use the multidimensional visual motion perception feature indicators of the participants not belonging to the visual motion perception anomaly group as negative sample data: that is, the multidimensional visual motion perception feature indicators of the participants not belonging to the visual motion perception anomaly group (i.e., the MDD patient subgroup 2 and the healthy group) as the model input items for negative samples, and assign the model output item of the negative samples a value of 0.

[0100] S6. Based on all the positive sample data and all the negative sample data, perform calibration and verification modeling based on machine learning algorithm to obtain the visual motion perception anomaly recognition model.

[0101] In step S6, the machine learning algorithm is a core artificial intelligence algorithm that specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills, reorganize existing knowledge structures to continuously improve their performance, and is the fundamental way to make computers intelligent. Specifically, the machine learning algorithm preferably uses a linear regression algorithm based on the Python sklearn library to quickly and accurately find patterns in the data. The specific process of calibration and verification modeling includes a model calibration process and a verification process, that is, first comparing the model simulation results with the measured data, and then adjusting the model parameters according to the comparison results to make the simulation results match the actual results. Therefore, the visual motion perception anomaly recognition model can be obtained through conventional calibration and verification modeling methods. Preferably, in the calibration and verification modeling process of the visual motion perception anomaly recognition model, a tree-based Bayesian optimization algorithm is used to fine-tune the model parameters. In addition, the machine learning algorithm can also use, but is not limited to, machine learning algorithms based on support vector machines, K-nearest neighbor method, stochastic gradient descent method, multivariate linear regression, multilayer perceptron, decision tree, backpropagation neural network, or radial basis function network, etc.

[0102] In step S6, in order to obtain the optimal visual motion perception anomaly recognition model, preferably, a calibration and verification model is performed based on a machine learning algorithm according to all the positive sample data and all the negative sample data to obtain the visual motion perception anomaly recognition model, including but not limited to the following steps S61 to S62.

[0103] S61. Based on all the positive sample data and all the negative sample data, calibration and verification modeling are performed on different machine learning algorithms to obtain multiple visual motion perception anomaly recognition models that correspond one-to-one with the machine learning algorithms.

[0104] In step S61, specifically, the various machine learning algorithms include, but are not limited to, support vector machines, K-Nearest Neighbor (KNN) algorithm, decision trees, random forests and / or leave-one-out cross-validation (LOOCV).

[0105] S62. Select the visual motion perception anomaly recognition model with the highest recognition accuracy from the plurality of visual motion perception anomaly recognition models as the final visual motion perception anomaly recognition model obtained from modeling.

[0106] In step S62, when the various machine learning algorithms include support vector machine, K-nearest neighbor algorithm, decision tree, random forest and leave-one-out cross-validation, the comparison results of the recognition accuracy are shown in Table 1 below:

[0107] Table 1. Comparison of recognition accuracy of multiple visual motion perception anomaly recognition models based on different machine learning algorithms.

[0108]

[0109]

[0110] As shown in Table 1 above, the visual-motor perception anomaly identification model based on support vector machines has the highest identification accuracy: an average accuracy of 95.74%, precision of 95.87%, recall of 95.74%, and an F1 score of 95.68%. This not only confirms the hypothesis that MDD patient subgroup 1 exhibits visual-motor perception anomalies within the MDD patient subgroup, but also enables this visual-motor perception anomaly identification model to be used for MDD patient subgrouping based on visual-motor perception data for participants belonging to the depressive disorder patient group. This facilitates the development of personalized diagnosis and treatment plans and is convenient for practical application and promotion. Furthermore, after obtaining two subgroup groups for all participants belonging to the depressive disorder patient group, the method also includes, but is not limited to: comparing the Hamilton Depression Rating Scale scores of all participants for each pair of subgroup groups in the two subgroup groups to find the symptom differences of the corresponding depressive disorder patients.

[0111] Therefore, based on the data processing method described in steps S1 to S6 above, a new scheme is provided for medical data processing based on machine learning and cluster analysis techniques to construct a visual-motor perception anomaly identification model. First, participants are grouped according to the Hamilton Depression Rating Scale. Then, participants' judgments on visual stimulus patterns and the duration of visual stimulus presentation are collected. Based on the collected results, multidimensional visual-motor perception characteristic indicators of the participants are statistically obtained. Next, K-Means cluster analysis is used to obtain two subgroups for all participants in the patient group. Visual-motor perception anomaly groups within these two subgroups are determined through indicator comparison analysis. Finally, positive sample data from the abnormal groups and negative sample data from the non-abnormal groups are used to calibrate and validate the model based on machine learning algorithms, resulting in a visual-motor perception anomaly identification model. This provides a visual-motor perception anomaly identification model for MDD patients with depressive disorders based on visual-motor perception data for subgroup processing, which can assist in the formulation of personalized diagnosis and treatment plans, facilitating practical application and promotion.

[0112] like Figure 6 As shown, the second aspect of this embodiment provides a physical system for implementing the data processing method described in the first aspect, including but not limited to a first human-computer interaction subsystem, a second human-computer interaction subsystem, and computer equipment.

[0113] The first human-computer interaction subsystem, communicatively connected to the computer device, is used to invite participants to complete the Hamilton Depression Rating Scale and transmit the Hamilton Depression Rating Scale to the computer device. The aforementioned first human-computer interaction subsystem can be implemented, but is not limited to, through a tablet computer or smartphone.

[0114] The second human-computer interaction subsystem, communicatively connected to the computer device, is used to collect the participant's judgment results on the visual stimulus pattern and the presentation duration of the visual stimulus pattern, and transmit the collection results to the computer device. The visual stimulus pattern refers to a pattern in which a visual stimulus pattern drifts left or right in the central area of ​​the display screen, and the judgment result refers to determining whether the drift direction of the corresponding visual stimulus pattern is left or right. Specifically, as... Figure 6 As shown, the second human-computer interaction subsystem includes, but is not limited to, the display screen, keyboard, speaker, and control device; the display screen is communicatively connected to the control device and is used to present the visual stimulus pattern under the control of the control device; the keyboard is communicatively connected to the control device and is used to input the participant's judgment result on the visual stimulus pattern and transmit the judgment result to the control device; the speaker is communicatively connected to the control device and is used to emit sound under the control of the control device; the control device is communicatively connected to the computer device and is used to collect the participant's judgment result on the visual stimulus pattern and the presentation duration of the visual stimulus pattern, transmit the collected results to the computer device, and when the judgment result is found to be incorrect, transmit a sound control command to the speaker to control the speaker to emit an error prompt sound.

[0115] The computer device is used to perform the following data processing steps:

[0116] The participant's score was quantified using the Hamilton Depression Rating Scale. Then, it was determined whether the score exceeded a preset threshold. If so, the participant was assigned to the depressive disorder patient group; otherwise, the participant was assigned to the healthy group.

[0117] Based on the collected results, the multidimensional visual motion perception characteristic index of the participants was statistically obtained.

[0118] Based on the multidimensional visual-motor perception feature indicators of all participants belonging to the depressive disorder patient group, two subgroups of all participants belonging to the depressive disorder patient group were obtained through K-Means clustering analysis;

[0119] For each subgroup in the two subgroup groups, the multidimensional visual motion perception feature index of all the corresponding participants is compared and analyzed with the multidimensional visual motion perception feature index of all the participants belonging to the healthy group to determine whether the corresponding subgroup group is a visual motion perception abnormal group.

[0120] The multidimensional visual motion perception feature index of the participants who belong to the visual motion perception abnormality group is used as positive sample data, and the multidimensional visual motion perception feature index of the participants who do not belong to the visual motion perception abnormality group is used as negative sample data.

[0121] Based on all the positive sample data and all the negative sample data, a calibration and verification model is performed using a machine learning algorithm to obtain a visual motion perception anomaly recognition model.

[0122] In one possible design, the computer device is further configured to, after obtaining two subgroups of all said participants belonging to the group of patients with depressive disorders, compare the Hamilton Depression Rating Scale of the corresponding participants for each pair of subgroups in the two subgroups to find the symptom differences of the corresponding patients with depressive disorders.

[0123] The working process, working details and technical effects of the aforementioned system provided in the second aspect of this embodiment can be found in the data processing method described in the first aspect, and will not be repeated here.

[0124] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A data processing method, characterized in that, include: Participants were invited to complete the Hamilton Depression Rating Scale, and their scores were quantified based on the scale. The scores were then used to determine whether they exceeded a preset threshold. If so, the participants were assigned to the depressive disorder patient group; otherwise, they were assigned to the healthy group. The participants' judgments on visual stimulus patterns and the duration of their presentations were collected. Based on the collected results, multidimensional visual-motor perception characteristic indicators were statistically derived. The visual stimulus pattern refers to a pattern in which a visual stimulus pattern drifts left / right in the central area of ​​the display screen. The visual stimulus pattern includes alternating large and small visual stimulus patterns. The large visual stimulus pattern refers to a large sinusoidal grating pattern with blurred edges that drifts left / right in the central area of ​​the display screen. The small visual stimulus pattern refers to a small sinusoidal grating pattern with blurred edges that drifts left / right in the central area of ​​the display screen. The judgment result refers to determining whether the drift direction of the corresponding visual stimulus pattern is left or right. Based on the collected results, the multidimensional visual-motor perception characteristic indicators were statistically derived, specifically including: extracting the participants' judgments on each large visual stimulus pattern and the duration of their presentations, and also extracting the participants' judgments on each small visual stimulus pattern. The judgment results and the presentation duration of each small visual stimulus pattern are used to determine the presentation time difference of each pair of adjacent large visual stimulus patterns. Similarly, the presentation time difference of each pair of adjacent small visual stimulus patterns is calculated based on the presentation duration of each large visual stimulus pattern to obtain a large visual stimulus pattern presentation time difference sequence. The adjacent large visual stimulus patterns include a large visual stimulus pattern used in the earlier trial and a large visual stimulus pattern used in the later trial that are adjacent in the trial time sequence. The adjacent small visual stimulus patterns also include a small visual stimulus pattern used in the earlier trial and a small visual stimulus pattern used in the later trial that are adjacent in the trial time sequence. Based on the participants' judgment results on each large visual stimulus pattern and the presentation duration of each large visual stimulus pattern, the participants' judgment results on each small visual stimulus pattern and the presentation duration of each small visual stimulus pattern, the large visual stimulus pattern presentation time difference sequence, and / or the small visual stimulus pattern presentation time difference sequence, multidimensional visual motion perception feature indicators are statistically obtained. Based on the multidimensional visual-motor perception feature indicators of all participants belonging to the depressive disorder patient group, two subgroups of all participants belonging to the depressive disorder patient group were obtained through K-Means clustering analysis; For each subgroup in the two subgroup groups, the multidimensional visual motion perception feature index of all the corresponding participants is compared and analyzed with the multidimensional visual motion perception feature index of all the participants belonging to the healthy group to determine whether the corresponding subgroup group is a visual motion perception abnormal group. The multidimensional visual motion perception feature index of the participants who belong to the visual motion perception abnormality group is used as positive sample data, and the multidimensional visual motion perception feature index of the participants who do not belong to the visual motion perception abnormality group is used as negative sample data. Based on all the positive and negative sample data, a calibration and verification model is performed using machine learning algorithms to obtain a visual motion perception anomaly recognition model. Specifically, this includes: calibrating and verifying the model using different machine learning algorithms based on all the positive and negative sample data to obtain multiple visual motion perception anomaly recognition models corresponding one-to-one with the machine learning algorithms. These machine learning algorithms include support vector machines, K-nearest neighbors, decision trees, random forests, and / or leave-one-out cross-validation. Finally, the visual motion perception anomaly recognition model with the highest recognition accuracy is selected from these multiple models as the final model.

2. The data processing method according to claim 1, characterized in that, When displaying the visual stimulus pattern, the display screen is at the same level as the participant's head and 45-47 centimeters away from the participant's eyes.

3. The data processing method according to claim 1, characterized in that, The multidimensional visual motion perception feature index includes any one of the following indices (A) to (I) or any combination thereof: (A) The fitting time threshold durationS with 75% accuracy in judging small visual stimulus patterns; (B) The fitting time threshold durationL with 75% accuracy in judging large visual stimulus patterns; (C) Indicator (D) Coefficient of variation (cvS) of small visual stimulus patterns presenting continuous time lag sequences; (E) The coefficient of variation (cvL) of the sustained time-difference sequence of large visual stimulus patterns; (F) The standard deviation sdS of the continuous time-difference sequence of small visual stimulus patterns; (G) The standard deviation sdL of the continuous time-difference sequence of large visual stimulus patterns; (H) The mean S of the continuous time difference sequence of small visual stimulus patterns; (I) The mean L of the continuous time difference sequence of the large visual stimulus pattern.

4. The data processing method according to claim 1, characterized in that, Based on the multidimensional visual-motor perception feature indices of all participants belonging to the aforementioned depressive disorder patient group, K-Means clustering analysis was used to obtain two subgroups of all participants belonging to the aforementioned depressive disorder patient group, including: Principal component analysis was performed on the multidimensional visual-motor perception feature indicators of all participants belonging to the depression disorder patient group to reduce the dimensionality of the multidimensional visual-motor perception feature indicators, resulting in the dimensionality-reduced multidimensional visual-motor perception feature indicators of all participants belonging to the depression disorder patient group. Based on the dimensionality-reduced multidimensional visual-motor perception feature indexes of all participants belonging to the depressive disorder patient group, two subgroups of all participants belonging to the depressive disorder patient group were obtained through K-Means clustering analysis.

5. The data processing method according to claim 1, characterized in that, After obtaining two subgroups of all participants belonging to the said depressive disorder patient group, the method further includes: For each pair of subgroups within the two subgroup groups, the Hamilton Depression Rating Scale was compared for all the corresponding participants to identify the symptom differences among patients with depressive disorders.

6. A data processing system, characterized in that, It includes a first human-computer interaction subsystem, a second human-computer interaction subsystem, and computer equipment; The first human-computer interaction subsystem is communicatively connected to the computer device and is used to invite participants to complete the Hamilton Depression Rating Scale and transmit the Hamilton Depression Rating Scale to the computer device. The second human-computer interaction subsystem, communicatively connected to the computer device, is used to collect the participant's judgment results on the visual stimulus pattern and the presentation duration of the visual stimulus pattern, and transmit the collection results to the computer device. The visual stimulus pattern refers to a pattern in which a visual stimulus pattern drifts left / right in the central area of ​​the display screen. The visual stimulus pattern includes alternating large and small visual stimulus patterns. The large visual stimulus pattern refers to a pattern in which a large sinusoidal grating pattern with blurred edges drifts left / right in the central area of ​​the display screen. The small visual stimulus pattern refers to a pattern in which a small sinusoidal grating pattern with blurred edges drifts left / right in the central area of ​​the display screen. The judgment result refers to determining whether the drift direction of the corresponding visual stimulus pattern is left or right. The computer device is used to perform the following data processing steps: The participant's score was quantified using the Hamilton Depression Rating Scale. Then, it was determined whether the score exceeded a preset threshold. If so, the participant was assigned to the depressive disorder patient group; otherwise, the participant was assigned to the healthy group. Based on the collected results, multidimensional visual motion perception feature indicators of the participants were statistically obtained, specifically including: extracting the participants' judgment results on each large visual stimulus pattern and the presentation duration of each large visual stimulus pattern based on the collected results; also extracting the participants' judgment results on each small visual stimulus pattern and the presentation duration of each small visual stimulus pattern; calculating the presentation time difference of each pair of adjacent large visual stimulus patterns based on the presentation duration of each large visual stimulus pattern to obtain the large visual stimulus pattern presentation time difference sequence; and calculating the presentation time difference of each pair of adjacent small visual stimulus patterns based on the presentation duration of each small visual stimulus pattern to obtain the small visual stimulus pattern presentation time difference sequence. The time difference sequence, wherein the adjacent large visual stimulus patterns include a large visual stimulus pattern used in the previous trial and a large visual stimulus pattern used in the subsequent trial that are adjacent in the trial time sequence, and the adjacent small visual stimulus patterns include a small visual stimulus pattern used in the previous trial and a small visual stimulus pattern used in the subsequent trial that are adjacent in the trial time sequence; based on the participants' judgment results on each large visual stimulus pattern and the presentation duration of each large visual stimulus pattern, the participants' judgment results on each small visual stimulus pattern and the presentation duration of each small visual stimulus pattern, the time difference sequence of the presentation of the large visual stimulus patterns and / or the time difference sequence of the presentation of the small visual stimulus patterns, multidimensional visual motion perception feature indicators are statistically obtained. Based on the multidimensional visual-motor perception feature indicators of all participants belonging to the depressive disorder patient group, two subgroups of all participants belonging to the depressive disorder patient group were obtained through K-Means clustering analysis; For each subgroup in the two subgroup groups, the multidimensional visual motion perception feature index of all the corresponding participants is compared and analyzed with the multidimensional visual motion perception feature index of all the participants belonging to the healthy group to determine whether the corresponding subgroup group is a visual motion perception abnormal group. The multidimensional visual motion perception feature index of the participants who belong to the visual motion perception abnormality group is used as positive sample data, and the multidimensional visual motion perception feature index of the participants who do not belong to the visual motion perception abnormality group is used as negative sample data. Based on all the positive and negative sample data, a calibration and verification model is performed using machine learning algorithms to obtain a visual motion perception anomaly recognition model. Specifically, this includes: calibrating and verifying the model using different machine learning algorithms based on all the positive and negative sample data to obtain multiple visual motion perception anomaly recognition models corresponding one-to-one with the machine learning algorithms. These machine learning algorithms include support vector machines, K-nearest neighbors, decision trees, random forests, and / or leave-one-out cross-validation. Finally, the visual motion perception anomaly recognition model with the highest recognition accuracy is selected from these multiple models as the final model.

7. The data processing system as described in claim 6, characterized in that, The second human-computer interaction subsystem includes the display screen, keyboard, speaker, and control device; The display screen is communicatively connected to the control device and is used to present the visual stimulus pattern under the control of the control device. The keyboard is communicatively connected to the control device and is used to input the participant's judgment result on the visual stimulus pattern and transmit the judgment result to the control device. The loudspeaker is communicatively connected to the control device and is used to emit sound under the control of the control device. The control device is communicatively connected to the computer device and is used to collect the participant's judgment results on the visual stimulus pattern and the presentation duration of the visual stimulus pattern, and transmit the collected results to the computer device. When the judgment result is found to be incorrect, the control device transmits a sound control command to the speaker to control the speaker to emit an error prompt sound.

Images