Epilepsy prediction method based on comprehensive analysis of multiple brain regions

Abstract

The application belongs to the technical field of epilepsy detection, and particularly relates to an epilepsy prediction method based on comprehensive analysis of multiple brain regions. The application provides an epilepsy prediction method based on an epilepsy prediction system based on comprehensive analysis of multiple brain regions, and the method can accurately predict whether a subject has epilepsy.

Description

[0001] Divisional application

[0002] This application is a divisional application of Chinese invention patent application No. 202411376519.X, filed on September 30, 2024, entitled "A Method and System for Epilepsy Prediction Based on Multi-Brain Region Comprehensive Analysis". Technical Field

[0003] This invention belongs to the field of epilepsy detection technology, specifically relating to an epilepsy prediction method based on comprehensive analysis of multiple brain regions. Background Technology

[0004] Epilepsy is a neurological disorder caused by abnormal brain activity, characterized by recurrent and persistent seizures. It affects individuals of all ages, and the number of cases has been steadily increasing in recent years. Predicting seizures can reduce the risk of injury and improve patient prognosis.

[0005] Currently, the clinical diagnosis of epilepsy primarily relies on doctors recording patients' seizure symptoms through long-term video and simultaneous EEG monitoring to determine the type of epilepsy. Because EEG is non-invasive and provides comprehensive information about the brain, it is widely used in epilepsy seizure prediction research. However, in practice, simultaneously viewing patient seizures and monitoring EEG status via video and software is often time-consuming, and the accuracy and efficiency of manual monitoring gradually decline over time. Furthermore, identifying abnormal EEG patterns to determine the brain's electrical activity and abnormalities for diagnosis and treatment requires a high level of expertise from the physician.

[0006] To address existing shortcomings, some studies have proposed using artificial intelligence (AI) to predict the probability of epileptic seizures. For example, the literature review "A Review of Research on Epilepsy Seizure Prediction Based on Artificial Intelligence" (202404) suggests that deep learning technology has greater value in epileptic seizure prediction compared to classic machine learning models that rely on manual feature extraction. However, this technology relies on analyzing multiple epileptic EEG datasets. Specifically, it points out that unprocessed EEG signals contain various artifacts, categorized as physiological artifacts (physiological electrical signals generated by eye movements, heartbeats, etc.) and non-physiological artifacts (interference from the external environment). Therefore, preprocessing of the acquired EEG is necessary to minimize or eliminate the influence of these artifacts. Thus, while this patented technology utilizes AI, the analyzed data is still EEG. Therefore, improvements based on the existing framework are necessary. Summary of the Invention

[0007] The purpose of this invention is to provide a comprehensive method for predicting epilepsy, which partially solves or alleviates the above-mentioned deficiencies in the prior art. The invention specifically adopts the following technical solution.

[0008] A method for predicting epilepsy based on multi-brain region comprehensive analysis, characterized in that the method is implemented based on an epilepsy prediction system based on multi-brain region comprehensive analysis, the epilepsy prediction system comprising:

[0009] The signal acquisition module is configured to acquire characteristic signals of the monitored subjects under different saccade task paradigms, the characteristic signals including saccade task parameters; the subjects include epilepsy patients and healthy volunteers.

[0010] The information preprocessing module, connected to the information acquisition module, is configured to organize, classify, and preliminarily screen multiple feature signals acquired by the signal acquisition module.

[0011] The information filtering and analysis module, connected to the preprocessing module, is configured to establish a logistic regression model and further filter and exclude the initially selected saccade task parameters; it is also configured to establish a second logistic regression model to integrate the selected saccade task parameters with the results output by the first logistic regression model.

[0012] The data output module, connected to the information filtering and analysis module, is set to output the results of the second logistic regression model, which are the probability of epilepsy based on multi-brain region comprehensive analysis.

[0013] The steps of the epilepsy prediction method are as follows:

[0014] S01: Collect saccade task parameters of the subject under different saccade task paradigms in the information acquisition module. The saccade task includes forward saccade task, reverse saccade task, memory saccade task and bistep saccade task.

[0015] The positive saccade task is used to assess cognitive function;

[0016] The reverse saccade task is used to study cognitive control and executive functions;

[0017] The memory saccade task is used to test an individual's memory and visuospatial processing abilities.

[0018] The two-step saccade task was used to study saccade control and cognitive function.

[0019] S02: In the information preprocessing module, the collected saccade task parameters of the subjects under different saccade tasks are sorted, classified and preliminarily screened to identify saccade task parameters that differ among the subjects; the subjects include epilepsy patients and healthy volunteers;

[0020] S03: In the information filtering and analysis module, a logistic regression model for each saccade task paradigm is established based on each saccade task paradigm in S01.

[0021] S04: In the information filtering and analysis module, multiple saccade task parameters initially filtered in S02 are combined to obtain multiple datasets containing different saccade task parameters. The datasets contain both velocity-type parameters and spatial-type parameters.

[0022] S05: In the information filtering and analysis module, multiple datasets are input into the logistic regression model of a single saccade task paradigm, and the first result of each model is output. The result is the epilepsy probability value based on the single saccade task paradigm.

[0023] S06: In the information filtering and analysis module, the AUC of the epilepsy probability value based on a single saccade task paradigm is calculated for each dataset, and datasets with an AUC less than 0.6 are excluded; further, static class parameters in the datasets are excluded to obtain a specific dataset containing specific saccade task parameters; the specific saccade task parameters include the accuracy of the forward saccade paradigm, the average speed of the reverse saccade paradigm, the spatial error of the reverse saccade paradigm, the spatial error of the memory saccade paradigm, the corrected saccade latency of the memory saccade paradigm, the average speed of the two-step saccade, and the spatial error of the two-step saccade.

[0024] S07: In the information filtering and analysis module, the saccade task parameters from the specific dataset are input into a logistic regression model of a single saccade task paradigm to obtain four specific saccade task paradigm models. The second result of each model is output, which is the epilepsy probability value based on a single specific saccade task paradigm. Then, a comprehensive logistic regression model is established based on the second result of each specific saccade task paradigm model.

[0025] The comprehensive logistic regression model is as follows:

[0026] ;

[0027] Among them, P pro P represents the probability value for judging a subject as having epilepsy based on a positive saccade task. anti P represents the probability value for determining that a subject has epilepsy based on the inverse saccade task. memory P represents the probability value for judging a subject to have epilepsy based on a memory-saccade task. double This represents the probability value for determining whether a subject has epilepsy based on a bistep saccade task;

[0028] S08: The data output module outputs the third result of the comprehensive logistic regression model, which is the probability of the subject having epilepsy based on multi-brain region comprehensive analysis.

[0029] Furthermore, the logistic regression model for the saccade task paradigm includes a positive saccade paradigm model.

[0030] Furthermore, the logistic regression model of the saccade paradigm includes the inverse saccade paradigm model.

[0031] Furthermore, the logistic regression model for the saccade paradigm includes the memory saccade paradigm model.

[0032] Furthermore, the logistic regression model of the saccade paradigm includes the two-step saccade paradigm model.

[0033] Furthermore, in step S06, static class parameters in the dataset are further excluded, so that there are no more than two static class parameters in the dataset.

[0034] Furthermore, the information preprocessing module is also used to eliminate interference signals, including signals that cause deviations and influences on the saccade results.

[0035] In some specific implementations, the interference signal includes the offset and influence of gender, age, and / or educational characteristics of epilepsy patients and healthy volunteers on the saccade results.

[0036] Furthermore, the information preprocessing module is also configured to eliminate interference signals.

[0037] This invention can also provide an epilepsy prediction system based on multi-brain region comprehensive analysis, the epilepsy prediction system comprising:

[0038] The signal acquisition module is configured to acquire characteristic signals of the monitored subjects under different saccade task paradigms, the signal characteristics including saccade task parameters;

[0039] The information preprocessing module, connected to the information acquisition module, is configured to sort, classify and preliminarily screen multiple feature signals acquired by the signal acquisition module, and is also configured to eliminate interference signals.

[0040] The information filtering and analysis module, connected to the preprocessing module, is configured to establish a first logistic regression model and filter and exclude the preliminarily selected saccade task parameters based on the established first logistic regression model; it is also configured to establish a second logistic regression model to integrate the selected saccade task parameters with the results output by the first logistic regression model.

[0041] The second logistic regression model is as follows:

[0042] ;

[0043] Among them, P pro P represents the probability value for identifying a subject as having epilepsy based on a positive eye-sac task. anti P represents the probability value for identifying a subject as having epilepsy based on the inverse saccade task. memory P represents the probability value for identifying a subject as having epilepsy based on a memory-based saccade task. double This represents the posterior probability of identifying a subject as having epilepsy based on a bistep saccade task;

[0044] The data output module, connected to the data analysis module, is configured to output the results of the second logistic regression model, which is a determination of the probability that the subject has epilepsy based on multi-brain region comprehensive analysis.

[0045] Beneficial technical effects:

[0046] This invention proposes an epilepsy prediction method based on multi-brain region comprehensive analysis, which differs from current mainstream methods that rely on EEG information for epilepsy prediction. Instead, it employs saccade artifacts, which are even considered by mainstream research to be negligible, to address the problem of predicting whether a subject has epilepsy. The method proposed in this invention involves relatively complex parameter selection and model training, aiming to obtain more accurate and objective prediction results.

[0047] First, since different saccade parameters correspond to the functional analysis of different brain regions, the method of this invention performs multiple rounds of elimination and screening of saccade task parameters, forming a dataset containing specific saccade parameters. Second, this invention also constructs multiple logistic regression models to achieve comprehensive prediction of epileptic seizures in multiple brain regions. Because the brain regions analyzed by different saccade paradigms overlap, the comprehensive system and method proposed in this invention have higher prediction accuracy than predictions based on a single saccade paradigm, and have guiding significance for the missed and misdiagnosis of epilepsy. In addition, the acquisition of saccade signals / parameters is more convenient, has a shorter operation cycle, and requires relatively lower skill from experimental personnel.

[0048] Because different diseases may cause overlapping abnormalities in brain regions, different diseases may involve examining the same saccade paradigm. For example, existing technology discloses a biomarker set and method for detecting migraines, which uses forward saccade paradigm, reverse saccade paradigm, memory saccade paradigm, and bistep saccade paradigm to analyze the probability of migraine occurrence. However, different saccade parameters within the same saccade paradigm reflect different brain region functions. Therefore, screening accurate and specific saccade parameters is necessary to predict different diseases. The saccade parameters disclosed in the existing technology include the latency of forward saccades, the rate of interrupted fixation of forward saccades, the latency of reverse saccades, the accuracy of memory saccades, the latency of memory saccades, the final saccade landing point error of bistep saccades, and the rate of interrupted fixation of bistep saccades. These saccade parameters are all static parameters. In contrast, the epilepsy prediction method and system based on multi-brain region comprehensive analysis proposed in this invention includes multiple velocity and spatial parameters, which is groundbreaking in the field of epilepsy prediction. Attached Figure Description

[0049] 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. In all the drawings, similar elements or parts are generally identified by similar reference numerals. The elements or parts in the drawings are not necessarily drawn to scale. Obviously, the drawings described below are some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0050] Figure 1 This is a schematic diagram of an epilepsy prediction system based on multi-brain region comprehensive analysis as described in this invention;

[0051] Figure 2 This is a schematic diagram of an epilepsy prediction method based on multi-brain region comprehensive analysis as described in this invention;

[0052] Figure 3 This is the significance detection result of multiple saccade task parameters under the positive saccade task paradigm in one embodiment of the present invention;

[0053] Figure 4 This is the significance detection result of multiple saccade task parameters under the reverse saccade task paradigm in one embodiment of the present invention;

[0054] Figure 5 This is the significance detection result of multiple saccade task parameters under the memory saccade task paradigm in one embodiment of the present invention;

[0055] Figure 6This is the significance detection result of multiple saccade task parameters under the two-step saccade task paradigm in one embodiment of the present invention;

[0056] Figure 7 This is the ROC curve for determining whether a patient has epilepsy using a comprehensive logistic regression model in one embodiment of the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0058] In this document, "and / or" includes any and all combinations of one or more of the listed related items.

[0059] In this article, "multiple" means two or more, that is, it includes two, three, four, five, etc.

[0060] As used in this specification, the term "about" typically means + / -5% of the value, more typically + / -4% of the value, more typically + / -3% of the value, more typically + / -2% of the value, even more typically + / -1% of the value, and even more typically + / -0.5% of the value.

[0061] In this specification, certain embodiments may be disclosed in a range-bound format. It should be understood that this "range-bound" description is merely for convenience and brevity and should not be construed as a rigid limitation on the disclosed range. Therefore, the description of a range should be considered as having specifically disclosed all possible subranges and the individual numerical values ​​within those ranges. For example, a description of the range 1-6 should be considered as having specifically disclosed subranges such as from 1 to 3, from 1 to 4, from 1 to 5, from 2 to 4, from 2 to 6, from 3 to 6, etc., and the individual numbers within those ranges, such as 1, 2, 3, 4, 5, and 6. This rule applies regardless of the breadth of the range.

[0062] Definition of noun:

[0063] The "static parameters" described in this invention include accuracy, latency, average number of saccades, rate of unresponsive saccades, blink rate, incidence of erroneous reflex saccades accompanied by corrective saccades, and latency of corrective saccades under each eye movement paradigm.

[0064] The "velocity parameters" mentioned in this invention include average velocity, maximum velocity, etc., under various eye-tracking paradigms.

[0065] The "spatial parameters" mentioned in this invention include spatial errors under each eye movement paradigm, first-step saccade amplitude, corrected saccade spatial error, landing point error, etc.

[0066] Example 1

[0067] This embodiment provides a system and method for determining whether a subject has epilepsy.

[0068] An epilepsy prediction system based on multi-brain region comprehensive analysis is shown in the schematic diagram below. Figure 1 It includes a signal acquisition module, which is configured to acquire characteristic signals of the monitored subject under different saccade task paradigms, the signal characteristics including saccade task parameters;

[0069] The information preprocessing module, connected to the information acquisition module, is configured to sort, classify and preliminarily screen multiple feature signals acquired by the signal acquisition module, and is also configured to eliminate interference signals.

[0070] The information filtering and analysis module, connected to the preprocessing module, is configured to establish a first logistic regression model and filter and exclude the initially selected saccade task parameters based on the established first logistic regression model; it is also configured to establish a second logistic regression model to integrate the selected saccade task parameters with the output of the first logistic regression model, wherein the second logistic regression model is:

[0071] ;

[0072] Among them, P pro P represents the probability value for identifying a subject as having epilepsy based on a positive eye-sac task. anti P represents the probability value for identifying a subject as having epilepsy based on the inverse saccade task. memory P represents the probability value for identifying a subject as having epilepsy based on a memory-based saccade task. double This represents the posterior probability of identifying a subject as having epilepsy based on a bistep saccade task;

[0073] The data output module, connected to the data analysis module, is configured to output the results of the second logistic regression model.

[0074] A method for epilepsy prediction based on the above system is illustrated in the diagram below. Figure 2 .

[0075] S01: Collect saccade task parameters of subjects under different saccade task paradigms.

[0076] S02: Initially screen out eye saccade task parameters that differ between epilepsy patients and healthy subjects, such as parameters that show significant differences.

[0077] S03: Establish a logistic regression model for a single saccade task paradigm.

[0078] S04: Combine the multiple eye saccade task parameters that have been initially selected to obtain multiple datasets containing eye saccade task parameters; each dataset must contain both velocity-type parameters and spatial-type parameters.

[0079] S05: Input the dataset containing saccade task parameters into a logistic regression model of a single saccade task paradigm and output the first result, which is the probability value of epilepsy.

[0080] S06: Based on the results in the first step, further filter the dataset to obtain a specific dataset containing specific saccade task parameters; the filtering process includes: 1) calculating the AUC of the epilepsy probability value of a single saccade task paradigm and excluding datasets with an AUC less than 0.6; 2) further excluding static class parameters in the dataset.

[0081] S07: Input the specific saccade parameters into the logistic regression model of a single saccade task paradigm to obtain the specific saccade task paradigm model, output the second result, and establish a comprehensive logistic regression model based on the second result.

[0082] S08: Output the third result of the comprehensive logistic regression model.

[0083] The following embodiments provide a specific implementation process example based on the system and method proposed in this embodiment.

[0084] Example 2

[0085] 1. Study Subjects: This study recruited epilepsy patients aged 18-67 years who voluntarily participated and signed informed consent forms at the outpatient clinic of West China Hospital, Sichuan University. Simultaneously, healthy controls matched for gender, age, and education were recruited from Chengdu and surrounding communities. Data on age, gender, vital signs, marital status, education level, lifestyle, dietary habits, age of onset of epilepsy, location, nature, triggers, presence or absence of aura, disease course, treatment, and family history were collected. Eye sac behavior testing was also conducted. Exclusion Criteria: Ophthalmic diseases affecting testing (e.g., blindness, significant atrophy in one or both eyes, glaucoma, severe inflammation of eye-related tissues, retinal detachment), organic neurological and psychiatric diseases (e.g., cerebral infarction, cerebral hemorrhage, brain tumor, dementia, claustrophobia, schizophrenia, and moderate to severe anxiety and depression), and headaches caused by medication overuse were included. This study was approved by the Ethics Committee of West China Hospital, Sichuan University.

[0086] 2. Saccade Behavioral Examination: All participants voluntarily signed informed consent forms. Visual acuity, visual field, and data from four different saccade task paradigms (positive saccades, P...) were collected in a relatively quiet, dark environment. pro Reverse eye twitching P anti Memory eye sac P memory and double-step eye twitching P double Data. The main observations included the subjects' accuracy, latency, saccade amplitude, average saccade rate, maximum saccade velocity, spatial error, multistep saccade incidence, precision and reversal, and the corrected saccade latency and corrected saccade amplitude gain for each paradigm. Calibration and validation were performed before each saccade behavior test. Each subject underwent 12 learning trials before the formal start of the task; each subject performed 40 formal trials for each task.

[0087] 3. Preliminary Data Screening and Removal: The saccade data processing workflow focuses on improving data quality and analytical accuracy. For blinking data, a strategic approach is adopted: in stages where blinking is frequent or has a significant impact, data is selectively skipped or simply marked, and not directly used for core analysis to avoid blinking interference. This is because blinking creates two saccade artifacts in the trajectory; these artifacts differ from actual saccades in speed, acceleration, and duration, and therefore need to be removed. Simultaneously, other noisy data is corrected or removed using interpolation, filtering, and other methods to ensure the continuity and accuracy of the remaining data.

[0088] 4. Statistical analysis: (1) Test whether the matching elements and other basic information of the two groups of subjects (for categorical variables such as gender and education level, the chi-square test is used) and parameters such as accuracy, latency, and final landing error of the four saccade task paradigms satisfy normality and homogeneity of variance. If both are satisfied, the T test is used; if normality is satisfied but homogeneity of variance is not satisfied, the Welch test is used; if the requirements of the parametric test are not met, the non-parametric U test is used. (2) In order to compare whether there are differences in the different parameters of the four saccade task paradigms of the two groups, the bias and influence of gender, age, and education level on the saccade results are removed. There are no statistically significant differences in saccades between the two groups in terms of gender, age, and education level.

[0089] Example 3

[0090] 1. Comparison of clinical data of subjects: A total of 100 epilepsy subjects were recruited. Some subjects were excluded because they did not complete the eye twitching test. Finally, 92 subjects were used for analysis. See Table 1 for details.

[0091] Table 1. Comparison of general characteristics between the epilepsy group and the control group.

[0092]

[0093] 2. Comparison of saccade parameters between the two groups: Overall, the latency of both forward and reverse saccades was longer in the epilepsy group than in the control group, with statistically significant differences (P<0.05). Furthermore, the final landing point error of bistep saccades was significantly larger in the epilepsy group than in the control group (p<0.05). Additionally, regarding stimulus location, in relatively complex saccade paradigms (memorized saccades and bistep saccades), the final landing point error differed in different locations for epilepsy patients, suggesting that the differences may be related to the site of brain damage. There were no statistically significant differences between the two groups in saccade amplitude, average saccade rate, maximum saccade rate, multistep saccade incidence, accuracy, landing point error, corrected saccade latency, and corrected saccade amplitude gain.

[0094] Overall, the accuracy of the four task paradigms in the epilepsy group was lower than that in the control group, and the spatial error was larger in the epilepsy group. The results were statistically significant (p<0.05), as shown in Table 2-5.

[0095] Table 2. Detailed list of significant parameters between the positive oculomotor paradigm epilepsy group and the control group.

[0096]

[0097] Table 3. Detailed list of significant parameters between the reverse saccade paradigm epilepsy group and the control group.

[0098]

[0099] Table 4. Detailed list of significant parameters between the memory-saccade paradigm epilepsy group and the control group.

[0100]

[0101] Table 5. Detailed list of significant parameters between the bistep saccade paradigm epilepsy group and the control group.

[0102]

[0103] The different saccade tasks involved in Tables 2-5 are described below.

[0104] Prosaccade: A tool used to assess cognitive function. This task assesses the brain's processing speed and accuracy by measuring rapid eye movements (i.e., saccades), thus indirectly reflecting the state of cognitive function.

[0105] The Antisaccade task is a psychological experiment used to study cognitive control and executive function. In this task, participants are asked to ignore a sudden stimulus and saccade in the opposite direction to the stimulus. This task tests participants' ability to suppress pre-set responses and execute controlled responses.

[0106] The memory-guided saccade is a psychological experiment that tests an individual's memory and visuospatial processing abilities. In this task, participants must first memorize the location of a target that appears randomly in their visual field. After the target disappears, they must maintain their attention for a delay and then, without visual cues, accurately move their gaze to the previously located target. This task primarily assesses the integration of memory, attention, and spatial orientation abilities.

[0107] The double-step saccade is an experimental paradigm for studying eye saccade control and related cognitive functions. It is often used to study how the brain plans and executes rapid target localization activities.

[0108] The following shows an example of multiple saccade task parameter combinations obtained in the preliminary screening stage by the method of the present invention.

[0109] Table 6 Combinations of parameters for the saccade task

[0110]

[0111] It is understood that the content shown in Table 6 is merely an example and not a limitation.

[0112] Example 4

[0113] A logistic regression model was established using data collected and organized in Examples 2 and 3 to predict the incidence of epilepsy in the subjects.

[0114] (1) Collect saccade task parameters of volunteers and epilepsy patients under four saccade task paradigms in the information collection module.

[0115] (2) In the information preprocessing module, the collected saccade task parameters are sorted, classified and preliminarily screened to identify saccade task parameters with significant differences among subjects (p < 0.05).

[0116] (3) In the information filtering and analysis module, logistic regression models for the four saccade task paradigms are established based on the four saccade task paradigms, denoted as P. pro P anti P memory and P double .

[0117] (4) In the information filtering and analysis module, multiple saccade task parameters with significant differences initially selected by the system are combined to obtain multiple datasets containing different saccade task parameters.

[0118] (5) In the information filtering and analysis module, the above multiple datasets are input into the logistic regression model of the single eye saccade task paradigm, and the first result of each model is output. The result is the epilepsy probability value based on the single eye saccade task paradigm.

[0119] (6) Calculate the AUC of the epilepsy probability value based on a single saccade task paradigm for each dataset, and exclude datasets with an AUC less than 0.6; further exclude static class parameters in the dataset to ensure that there are no more than 2 static class parameters in the entire dataset. Finally, obtain a specific dataset containing specific saccade task parameters, which are pro accuracy, anti average speed, anti spatial error, double average speed, double spatial error, memory spatial error, and memory corrected saccade latency.

[0120] (7) In the information filtering and analysis module, the eye saccade task parameters in the above-mentioned specific dataset are input into the logistic regression model of a single eye saccade task paradigm to obtain 4 specific eye saccade task paradigm models. The second result of each model is output. The second result is the epilepsy probability value based on a single specific eye saccade task paradigm, as follows.

[0121] Positive eye saccade paradigm model:

[0122] ;

[0123] P pro Let P represent the posterior probability of identifying a subject as having epilepsy based on positive eye saccades, and X1 represent the accuracy of positive eye saccades. After calculation, P... pro The AUC value is 0.628.

[0124] Reverse saccade paradigm model:

[0125] ;

[0126] P anti Let P represent the posterior probability of identifying a subject as having epilepsy based on retrograde eye saccades, where X1 represents the average velocity of the retrograde eye saccades, and X2 represents the spatial error of the retrograde eye saccades. After calculation, P... anti The AUC value is 0.703.

[0127] Memory eye saccade paradigm model:

[0128] ;

[0129] P memory Let P represent the posterior probability of a subject being diagnosed with epilepsy based on memory eye sacs, where X1 represents the spatial error of memory eye sacs and X2 represents the corrected sacacic latency of memory eye sacs. After calculation, P... memoryThe AUC value is 0.750.

[0130] Two-step saccade paradigm model:

[0131] ;

[0132] P double Let P represent the posterior probability of identifying a subject as having epilepsy based on bistep saccades, where X1 represents the average velocity of the bistep saccades, and X2 represents the spatial error of the bistep saccades. After calculation, P... double The AUC value is 0.703.

[0133] To further quantify the probability assessment of the four saccade paradigms, a comprehensive logistic regression model is then established based on the second result of the model for each specific saccade task paradigm. The comprehensive logistic regression model is as follows:

[0134] .

[0135] (8) Output the third result of the comprehensive logistic regression model, which is the probability of the subject having epilepsy based on multi-brain region comprehensive analysis.

[0136] After calculation, the constructed comprehensive logistic regression model had a sensitivity of 77.4%, a specificity of 73.1%, and an area under the curve (AUC) of 0.82 in predicting the probability that the subject had epilepsy.

[0137] Sensitivity and specificity are important metrics for evaluating the performance of epileptic seizure prediction models. Sensitivity measures the true positive probability (TPR), while specificity measures the true negative probability (TNR). ROC evaluates the TPR and FPR during the interictal and preictal periods. AUC ranks the classification algorithms based on TPR and FPR. AUC values ​​range from 0 to 1; a higher AUC value indicates better classification algorithm performance.

[0138] The following shows the predicted probabilities of disease in 3 healthy volunteers and 3 patients with epilepsy based on the above models.

[0139] Table 7. Predicted probability of different subjects having epilepsy

[0140]

[0141] As shown in Table 7, the probabilities of whether subjects have epilepsy vary considerably among different eye saccade paradigm models. The comprehensive logistic regression model can quantify these differences to some extent and has guiding significance for the missed diagnosis and misdiagnosis of epilepsy.

[0142] Example 5

[0143] The seven saccade parameters finally selected in Example 4 were quantitatively analyzed to examine the differences between the control group and the epilepsy group, and the effect size was output. The results are shown in Table 8.

[0144] Table 8 Quantitative Analysis of Saccade Parameters

[0145]

[0146] Accuracy primarily assesses cognitive function; average speed primarily assesses executive and information processing functions; spatial error primarily assesses motor control and spatial calculation functions; and corrected saccade latency primarily assesses attention, inhibitory function, and executive-related functions. These results indicate that epileptic patients primarily differ significantly from normal individuals in motor and executive functions. Analyzing brain regions controlling motor and executive functions can be used to better predict whether a subject has epilepsy.

[0147] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. An information processing method based on multi-brain region comprehensive analysis, characterized in that, The information processing method is implemented based on an information processing system for multi-brain region comprehensive analysis, the information processing system comprising: The signal acquisition module is configured to acquire characteristic signals of the monitored subject under different saccade task paradigms, the characteristic signals including saccade task parameters; The information preprocessing module, connected to the information acquisition module, is configured to organize, classify, and preliminarily screen multiple feature signals acquired by the signal acquisition module. The information filtering and analysis module, connected to the preprocessing module, is configured to establish a logistic regression model and further filter and exclude the initially selected saccade task parameters; it is also configured to establish a second logistic regression model to integrate the selected saccade task parameters with the results output by the first logistic regression model. The data output module, connected to the information filtering and analysis module, is set to output the results of the second logistic regression model, which are the probability of epilepsy based on multi-brain region comprehensive analysis. The steps of the information processing method are as follows: S01: Collect saccade task parameters of the subject under different saccade task paradigms in the information acquisition module. The saccade task includes forward saccade task, reverse saccade task, memory saccade task and bistep saccade task. The positive saccade task is used to assess cognitive function; The reverse saccade task is used to study cognitive control and executive functions; The memory saccade task is used to test an individual's memory and visuospatial processing abilities. The two-step saccade task was used to study saccade control and cognitive function. S02: In the information preprocessing module, the collected saccade task parameters of the subjects under different saccade tasks are sorted, classified and preliminarily screened to identify saccade task parameters that differ among the subjects; the subjects include epilepsy patients and healthy volunteers; S03: In the information filtering and analysis module, a logistic regression model for each saccade task paradigm is established based on each saccade task paradigm in S01. S04: In the information filtering and analysis module, multiple saccade task parameters initially filtered in S02 are combined to obtain multiple datasets containing different saccade task parameters. The datasets contain both velocity-type parameters and spatial-type parameters. S05: In the information filtering and analysis module, multiple datasets are input into the logistic regression model of a single saccade task paradigm, and the first result of each model is output. The result is the epilepsy probability value based on the single saccade task paradigm. S06: In the information filtering and analysis module, the AUC of the epilepsy probability value based on a single saccade task paradigm is calculated for each dataset, and datasets with an AUC less than 0.6 are excluded; further, static class parameters in the datasets are excluded to obtain a specific dataset containing specific saccade task parameters; the specific saccade task parameters include the accuracy of the forward saccade paradigm, the average speed of the reverse saccade paradigm, the spatial error of the reverse saccade paradigm, the spatial error of the memory saccade paradigm, the corrected saccade latency of the memory saccade paradigm, the average speed of the two-step saccade, and the spatial error of the two-step saccade. S07: In the information filtering and analysis module, the saccade task parameters from the specific dataset are input into a logistic regression model of a single saccade task paradigm to obtain four specific saccade task paradigm models. The second result of each model is output, which is the epilepsy probability value based on a single specific saccade task paradigm. Then, a comprehensive logistic regression model is established based on the second result of each specific saccade task paradigm model. The comprehensive logistic regression model is as follows: ； Among them, P pro P represents the probability value for judging a subject as having epilepsy based on a positive saccade task. anti P represents the probability value for determining that a subject has epilepsy based on the inverse saccade task. memory P represents the probability value for judging a subject to have epilepsy based on a memory-saccade task. double This represents the probability value for determining whether a subject has epilepsy based on a bistep saccade task; S08: The data output module outputs the third result of the comprehensive logistic regression model, which is the probability of the subject having epilepsy based on multi-brain region comprehensive analysis.

2. The information processing method based on multi-brain region comprehensive analysis as described in claim 1, characterized in that, The logistic regression model for the saccade task paradigm includes the positive saccade paradigm model: ； X1 represents the accuracy of positive eye saccades.

3. The information processing method based on multi-brain region comprehensive analysis as described in claim 1, characterized in that, The logistic regression model for the saccade paradigm includes the inverse saccade paradigm model: ； Where X1 represents the average velocity of the reverse eye saccade, and X2 represents the spatial error of the reverse eye saccade.

4. The information processing method based on multi-brain region comprehensive analysis as described in claim 1, characterized in that, The logistic regression model for the saccade paradigm includes the memory saccade paradigm model: ； Where X1 represents the spatial error of memory saccades, and X2 represents the corrected saccade latency of memory saccades.

5. The information processing method based on multi-brain region comprehensive analysis as described in claim 1, characterized in that, The logistic regression model for the saccade paradigm includes a two-step saccade paradigm model: ； Where X1 represents the average velocity of the double-step saccade, and X2 represents the spatial error of the double-step saccade.

6. The information processing method based on multi-brain region comprehensive analysis as described in claim 1, characterized in that, In step S06, static class parameters in the dataset are further excluded, so that there are no more than two static class parameters in the dataset.

7. The information processing method based on multi-brain region comprehensive analysis as described in claim 1, characterized in that, The information preprocessing module is also used to eliminate interference signals, including signals that cause deviations and influences on the saccade results.

8. The information processing method based on multi-brain region comprehensive analysis as described in claim 7, characterized in that, The interference signals include the offset and influence of gender, age, and / or educational characteristics of epilepsy patients and healthy volunteers on saccade results.

9. The information processing method based on multi-brain region comprehensive analysis as described in claim 1, characterized in that, The information preprocessing module is also configured to eliminate interference signals.

Images