Personnel cognitive competence assessment and recovery method based on big data

Abstract

The invention provides a personnel cognitive competence assessment and recovery method based on big data, and the assessment method comprises the steps: collecting basic data of a to-be-assessed target personnel, and sequentially carrying out a plurality of assessment tests under each cognitive competence assessment classification on the target personnel based on a plurality of preset cognitive competence assessment classifications; obtaining test data, a single capability evaluation value and completion time of each evaluation test, inputting the basic data and each test data into a prediction model, and calculating a big data analysis index of each evaluation test based on a big data analysis capability evaluation value output by the prediction model; and calculating a final cognitive competence assessment result of the target person according to the single competence assessment value of each assessment test, the completion time and the big data index analysis index. According to the method, a computer technology and a big data analysis technology are utilized to carry out multi-dimensional comprehensive assessment on the cognitive competence of the operator, and the accuracy, intelligence and comprehensiveness of assessment are improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method for evaluating and restoring human cognitive abilities based on big data. Background Art

[0002] At present, the relatively closed working environment and high-intensity and stressful work tasks may have adverse effects on the cognitive abilities of workers and even affect their work efficiency. Therefore, it is necessary to evaluate the cognitive abilities of workers so that subsequent recovery training can be carried out based on the evaluation results, thereby ensuring the normal progress of the work and improving the work efficiency of workers.

[0003] In related technologies, the traditional cognitive ability assessment method is generally a scale assessment, and usually the tester selects relevant cognitive measurement tools to conduct scale tests on the operators. For example, commonly used measurement tools include the Wechsler Intelligence Scale (WAIS) and the Liebert Memory Test. With the development of computer technology, the popularity of computerized cognitive ability assessment has gradually increased. Compared with the traditional scale assessment method, computerized cognitive ability assessment has a greater advantage as a modern and intelligent testing method, and it can obtain more accurate assessment results.

[0004] However, the evaluation schemes in the above-mentioned related technologies, for the scale evaluation method, rely heavily on the professionalism and personal experience of the testers, so the accuracy and credibility of the evaluation results are low, and it is not convenient for data processing and large-scale implementation. For the computerized cognitive ability evaluation method, there is a lack of intelligent and comprehensive evaluation of the cognitive ability of operators in special working environments, such as relatively closed environments. In addition, the evaluation method and evaluation content of the current computerized cognitive ability evaluation scheme are relatively simple. The standardized scores obtained by the scheme can only reflect the performance of the subjects in a certain cognitive ability relative to the peer group. The stereotyped evaluation method ignores the complexity and diversity of the individual, and cannot comprehensively and comprehensively evaluate the individual's cognitive ability level, and there is a strong one-sidedness. Furthermore, because the evaluation results cannot meet the needs, the formulation of the recovery training program lacks data basis and support, and it is impossible to formulate a reasonable recovery training program for specific personnel. In addition, the cognitive recovery programs in the related technologies are mostly trained in a single way, lacking a variety of training recovery methods, resulting in unsatisfactory recovery effects. Summary of the invention

[0005] The present application aims to solve one of the technical problems in the related art at least to some extent.

[0006] To this end, the first purpose of this application is to propose a personnel cognitive ability assessment method based on big data. This method uses computer technology and big data analysis technology to conduct a multi-dimensional comprehensive assessment of the cognitive ability of operators. It can solve the current problems of single cognitive ability assessment method, low credibility and lack of comprehensive assessment, and improve the accuracy, intelligence and comprehensiveness of the assessment.

[0007] The second purpose of this application is to propose a method for restoring the cognitive ability of personnel based on assessment data;

[0008] The third object of this application is to propose a device for evaluating personnel cognitive ability based on big data;

[0009] The fourth object of the present application is to provide a device for restoring a person's cognitive ability based on evaluation data;

[0010] The fifth object of the present application is to provide an electronic device;

[0011] A sixth object of the present application is to provide a computer-readable storage medium.

[0012] To achieve the above objectives, the first aspect of the present application is to propose a method for evaluating personnel cognitive ability based on big data, the method comprising the following steps:

[0013] Collecting basic data of the target person to be evaluated, and based on a plurality of preset cognitive ability evaluation categories, sequentially performing a plurality of evaluation tests under each of the cognitive ability evaluation categories on the target person;

[0014] Obtain the test data, single capability evaluation value and completion time of each of the evaluation tests, and input the basic data and each test data into a prediction model, and calculate the big data analysis index of each evaluation test based on the big data analysis capability evaluation value output by the prediction model, wherein the prediction model is pre-trained based on big data analysis technology, and the big data analysis index represents the error between the big data analysis capability evaluation value and the single capability evaluation value;

[0015] The final cognitive ability assessment result of the target person is calculated based on the single ability assessment value of each assessment test, the completion time and the big data indicator analysis index.

[0016] Optionally, according to one embodiment of the present application, the calculation of the big data indicator analysis index for each assessment test includes: for each assessment test, calculating the absolute value of the difference between the single ability assessment value and the big data analysis ability assessment value; dividing the absolute value by the single ability assessment value to obtain the big data indicator analysis index for the corresponding single assessment test, wherein the big data analysis ability assessment value is used as the true result of the target person's cognitive ability, and the big data indicator analysis index is negatively correlated with the final cognitive ability assessment result.

[0017] Optionally, according to one embodiment of the present application, the prediction model is constructed and trained in advance based on big data analysis technology, including: constructing a functional connection network based on the blood oxygen level-dependent BOLD signal in the functional magnetic resonance imaging data; extracting features from the functional connection network, and associating the extracted features with preset cognitive performance indicators through a machine learning algorithm to construct a prediction model based on functional connectivity; obtaining cognitive assessment data of a large number of participants through a big data platform, dividing the cognitive assessment data into a training data set and a test set, and optimizing the training data set through a big data optimization algorithm; using the optimized training data set, training the prediction model based on a deep learning algorithm to predict the fluid intelligence quotient FIQ, crystallized intelligence quotient CIQ and cognitive comprehensive measurement value IC-Cognition for each assessment test; calculating multiple evaluation indicators of the model, evaluating the prediction accuracy and generalization ability of the trained prediction model through the multiple evaluation indicators, and correcting the trained prediction model according to the evaluation results.

[0018] Optionally, according to one embodiment of the present application, the multiple evaluation indicators include: out-of-sample test-retest reliability, and calculating the out-of-sample test-retest reliability includes: inputting the test set into the trained prediction model for testing; obtaining two sets of feature weights of the prediction model during the training process and the testing process; calculating the intra-group correlation coefficient ICC between the two sets of feature weights to quantify the consistency of the feature weights between the two segmentations.

[0019] Optionally, according to one embodiment of the present application, the big data optimization algorithm includes: a branch and bound algorithm, and the optimizing the training data set by the big data optimization algorithm includes: seeking the global optimal solution for each set of cognitive assessment data by the branch and bound algorithm to simplify the large-scale training data set.

[0020] Optionally, according to one embodiment of the present application, the calculation of the final cognitive ability assessment result of the target person includes: multiplying the single ability assessment value, the completion time and the big data indicator analysis index of each assessment test by the corresponding first weight coefficient, respectively, to obtain a single-phase score for each assessment test, and the big data indicator analysis index is negatively correlated with the single-phase score; generating a comparative judgment matrix based on the hierarchical analysis method, calculating the second weight coefficient of each assessment test through the comparative judgment matrix, and multiplying the single-phase score of each assessment test by the corresponding second weight coefficient to obtain the final cognitive ability assessment result.

[0021] Optionally, according to one embodiment of the present application, the analytic hierarchy process is used to break down the cognitive ability assessment problem into a target layer, an evaluation index layer and a solution layer, and the second weight coefficient of each assessment test is calculated through the judgment matrix, including: comparing the evaluation indicators of each assessment test in the evaluation index layer in pairs, and constructing the comparison judgment matrix according to the obtained importance comparison results; calculating the maximum eigenvalue root of the comparison judgment matrix and the eigenvector matrix corresponding to the maximum eigenvalue root, and determining the second weight coefficient of each assessment indicator according to the eigenvector matrix.

[0022] To achieve the above-mentioned purpose, the second aspect of the present application further proposes a method for restoring human cognitive ability based on evaluation data, the method comprising the following steps:

[0023] Acquire cognitive ability assessment data of a target person to be trained, and determine the classification of the cognitive ability to be restored and the corresponding degree of decline of the target person according to the cognitive ability assessment data, wherein the cognitive ability assessment data is obtained by assessing the target person using the method described in any one of the first aspects above;

[0024] Setting a variety of cognitive recovery training, solving the causal relationship between cognitive ability assessment and cognitive recovery training based on Bayesian network, substituting the classification of cognitive ability to be recovered and the corresponding degree of decline into the causal relationship, and formulating corresponding recovery training tasks and training cycles for the target personnel;

[0025] According to the recovery training task, multimodal intervention training is performed on the target person, and during the training process, the changes in the cognitive ability of the target person are continuously monitored, and the recovery plan is adjusted in real time according to the changes in the cognitive ability.

[0026] To achieve the above purpose, the third aspect of the present application also proposes a human cognitive ability assessment device based on big data, comprising the following modules:

[0027] A testing module, used to collect basic data of a target person to be evaluated, and based on a plurality of preset cognitive ability evaluation categories, sequentially conduct a plurality of evaluation tests under each of the cognitive ability evaluation categories on the target person;

[0028] A big data analysis module, used to obtain the test data, single capability evaluation value and completion time of each of the evaluation tests, and input the basic data and each test data into a prediction model, and calculate the big data analysis index of each evaluation test based on the big data analysis capability evaluation value output by the prediction model, wherein the prediction model is pre-trained based on the big data analysis technology, and the big data analysis index represents the error between the big data analysis capability evaluation value and the single capability evaluation value;

[0029] The calculation module is used to calculate the final cognitive ability assessment result of the target person according to the single ability assessment value of each assessment test, the completion time and the big data indicator analysis index.

[0030] To achieve the above-mentioned purpose, the fourth aspect of the present application further proposes a device for restoring human cognitive ability based on evaluation data, comprising the following modules:

[0031] an acquisition module, used to acquire cognitive ability assessment data of a target person to be trained, and determine the classification of the cognitive ability to be restored and the corresponding degree of decline of the target person according to the cognitive ability assessment data, wherein the cognitive ability assessment data is obtained by assessing the target person using the method described in any one of the first aspects above;

[0032] A training formulation module is used to set a variety of cognitive recovery training, solve the causal relationship between cognitive ability assessment and cognitive recovery training based on the Bayesian network, and formulate corresponding recovery training tasks and training cycles for the target personnel based on the cognitive relationship to be restored;

[0033] The training execution module is used to perform multimodal intervention training on the target person according to the recovery training task, and continuously monitor the changes in the cognitive ability of the target person during the training process, and adjust the recovery plan in real time according to the changes in the cognitive ability.

[0034] To achieve the above-mentioned purpose, the fifth aspect of the present application further proposes an electronic device, comprising:

[0035] processor;

[0036] a memory for storing instructions executable by the processor;

[0037] Wherein, the processor is configured to execute the instructions to implement a method for assessing personnel cognitive abilities based on big data as described in any one of the first aspects above, or to implement a method for restoring personnel cognitive abilities based on assessment data as described in any one of the second aspects above.

[0038] To achieve the above-mentioned purpose, the sixth aspect of the present application also proposes a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements the personnel cognitive ability assessment method based on big data as described in any one of the first aspects above, or implements the personnel cognitive ability recovery method based on assessment data as described in any one of the second aspects above.

[0039] The technical solution provided by the embodiment of the present application brings at least the following beneficial effects: the present application adopts a computerized cognitive assessment method, which can provide accurate measurement results at the millisecond level, automatically and accurately score the test personnel, reduce the professional requirements for the tester, and facilitate data management and analysis. The present application uses computer technology and big data analysis technology to conduct a multi-dimensional comprehensive assessment of the cognitive ability of the operator. For each major aspect of the cognitive ability of the operator in a relatively closed environment, a set of cognitive test schemes and related algorithms are used to achieve a comprehensive assessment of cognitive ability, thereby improving the intelligence and comprehensiveness of the assessment. In addition, the present application adopts standardized assessment steps, model tools and algorithms, and with the help of big data analysis technology, the objectivity, scientificity and accuracy of the assessment results can be ensured. The present application integrates the assessment methods of computerized cognitive assessment and psychological principles. Through systematic measurement and assessment, it can accurately measure the level of individuals in various cognitive fields, fully consider the complexity and diversity of individuals, improve the pertinence of the assessment, and provide technical support for the accurate, comprehensive and rapid assessment of the cognitive ability of operators. It can be applied to the cognitive ability assessment in various special operating environments, enriching the application scenarios of cognitive assessment.

[0040] Moreover, for recovery training, the present application can automatically match a multi-dimensional recovery training program for a specific individual based on the comprehensive evaluation results of the personnel's cognitive ability, associate cognitive assessment with recovery training, and automatically conduct recovery training for the operating personnel in a variety of training and recovery methods through computer technology. The present application provides personalized training plans and recovery programs based on the type and degree of cognitive decline of the individual. Through systematic evaluation and training, the individual's cognitive ability can be restored in a targeted manner. The present application combines various test training tasks with visual, olfactory, auditory, and vibrotactile stimulation and recovery methods to improve the training effect from multi-dimensional intervention. The present application can also provide real-time feedback on the individual's training performance and progress, provide individuals with timely evaluation and adjustment suggestions, help individuals better understand their own cognitive status, and guide individuals to conduct targeted training, so as to achieve real-time feedback on the degree of recovery and iterative recovery training.

[0041] Therefore, this application improves the accuracy, intelligence and comprehensiveness of cognitive assessment, and on this basis also improves the pertinence, diversity and systematicness of cognitive ability recovery training, which can significantly enhance the effect of recovery training and ensure the work efficiency of personnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0043] Figure 1 A flowchart of a method for evaluating the cognitive ability of operators based on big data proposed in an embodiment of the present application;

[0044] Figure 2 A flowchart of a prediction model training method proposed in an embodiment of the present application;

[0045] Figure 3 A flowchart of a method for restoring cognitive ability of workers based on evaluation data proposed in an embodiment of the present application;

[0046] Figure 4 A flow chart of a method for solving a causal relationship proposed in an embodiment of the present application;

[0047] Figure 5 A schematic diagram of the principle of a specific method for restoring the cognitive ability of workers based on evaluation data proposed in an embodiment of the present application;

[0048] Figure 6 A schematic diagram of the structure of a device for evaluating the cognitive ability of operators based on big data proposed in an embodiment of the present application;

[0049] Figure 7 A schematic diagram of the structure of a device for restoring cognitive ability of workers based on evaluation data proposed in an embodiment of the present application;

[0050] Figure 8 A schematic diagram of the structure of an electronic device proposed in an embodiment of the present application. DETAILED DESCRIPTION

[0051] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0052] It should be noted that cognitive ability assessment is the process of using relevant tools to quantitatively describe individual cognitive abilities. Currently, the more authoritative cognitive ability assessment includes the WJIV cognitive ability assessment, which is a set of cognitive measurement methods based on the CHC cognitive theory. According to the tools used in cognitive ability assessment, the development stage of cognitive ability assessment can be divided into scale assessment and computerized cognitive assessment, which have differences and their own characteristics based on the cognitive theories and technical levels of different periods.

[0053] Among them, scale assessment, as an early assessment tool for cognitive ability, is mainly divided into intelligence scales and neuropsychological scales. The assessment form is mainly assessed through various paper-based standardized scales. Among them, the scales used for intelligence tests include the Binet-Simon Intelligence Scale, the Stanford-Binet Intelligence Scale and the Wechsler Intelligence Scale. Neuropsychological scales are mostly used to assess the cognitive level of clinical patients, including the Mini-Mental State Assessment Scale, the Montreal Cognitive Assessment (MoCA), etc. MoCA is mainly used to assess the cognitive ability of patients with mild cognitive impairment. The advantages of the scale are standardization, high reliability and validity, and strong practicality. However, the disadvantages are that it is not convenient for data processing and large-scale testing, and it requires high professionalism from the tester.

[0054] The cognitive ability assessment scheme using a scale assessment in the relevant embodiment is mainly divided into the following steps: the first step is to select a suitable cognitive measurement tool, such as the Wechsler Intelligence Scale (WAIS), the Liebert Memory Test and the Sternberg Test, according to the purpose of the assessment and the age of the subject. According to the specific situation, select a measurement tool to assess the cognitive ability of the individual. The second step is to design experimental tasks and implement them according to the selected measurement tool. The experimental tasks may include solving problems, memorizing words and completing reasoning tasks. During the implementation process, the experimental operation specifications are followed. The third step is to analyze and interpret the data based on the experimental results. Statistical methods, such as descriptive statistics, correlation analysis, etc., can be used to understand the performance of individuals in different cognitive fields. At the same time, compare with the corresponding standardized data to assess the level of cognitive function of the individual. The fourth step is to write an assessment report based on the results of the data analysis. The report should include basic information, measurement results and assessment conclusions of the individual.

[0055] However, the scale evaluation method in the above-mentioned related embodiments has the following disadvantages: First, it is not comprehensive. Its standardized score can only reflect the performance of the test-taker in a certain cognitive ability relative to the peer group, and cannot comprehensively evaluate the individual's cognitive ability level. Second, stereotyped evaluation: the standardized score simplifies the individual's ability level into a specific score, which easily ignores the complexity and diversity of the individual.

[0056] The development of computer technology has made it possible to evaluate cognitive processes and test psychomotor and other operational abilities through computer equipment. Computerized cognitive assessment is applicable to people of all ages and types. For example, it can be used to evaluate and test various cognitive abilities of various relevant personnel in military applications, such as memory, attention, executive function, perception, thinking, motor performance, judgment and decision-making ability, as well as evaluate neuromotor functions, such as reaction time, finger / hand flexibility and eye tracking. Typical evaluation systems include the Performance Evaluation Test for Environmental Research (PETER) and the Mini Cognitive Rapid Assessment Test System. However, the evaluation dimensions of computerized cognitive assessment schemes in related technologies are relatively single, and it is impossible to obtain comprehensive evaluation results.

[0057] To this end, this application proposes a method and device for evaluating the cognitive ability of operators based on big data, and realizes a comprehensive evaluation of cognitive intelligence through computer technology. This application is a computerized cognitive assessment solution. As a modern and intelligent testing method, compared with traditional scale evaluation, it can provide millisecond-level precise measurement, automatically and accurately score, reduce the professional requirements of the tester, and facilitate data management and analysis.

[0058] The following describes a method, apparatus and computer device for evaluating human cognitive ability based on big data according to an embodiment of the present invention with reference to the accompanying drawings.

[0059] Figure 1 A flowchart of a method for evaluating human cognitive ability based on big data proposed in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method comprises the following steps:

[0060] Step S101: Collect basic data of the target person to be evaluated, and based on multiple preset cognitive ability evaluation categories, perform multiple evaluation tests under each cognitive ability evaluation category on the target person in turn.

[0061] The executor of the big data-based operator cognitive ability assessment method of the present application is a computer device. First, the basic data of the operator who is currently undergoing cognitive ability assessment, such as name, age, job category and work environment, is collected to provide data support for subsequent targeted assessments. Then, the cognitive abilities of the personnel are tested and evaluated from multiple aspects such as long-term memory search, short-term memory, processing speed, auditory processing, visual-spatial thinking, knowledge comprehension, fluid reasoning and quantitative ability. There are a variety of specific test methods under each cognitive ability assessment category. For example, the various assessment tests that can be performed under each assessment category are shown in the following table:

[0062] Table 1 Evaluation test table

[0063]

[0064]

[0065] According to the evaluation test contents shown in Table 1, the operators can be tested in turn through computer equipment according to the evaluation indicators of each evaluation test. During the test, the computer equipment can display the test contents to the operators, remind and guide the operators to perform the test actions, interact with the operators, and collect the test data of the operators.

[0066] Step S102: Obtain the test data, single capability assessment value and completion time of each assessment test, input the basic data and each test data into the prediction model, and calculate the big data analysis index of each assessment test based on the big data analysis capability assessment value output by the prediction model.

[0067] Among them, the prediction model is pre-trained based on big data analysis technology, and the big data analysis index is used to represent the error between the big data analysis capability assessment value and the single capability assessment value.

[0068] Specifically, the test data of the operating personnel under each test is collected through computer equipment, the time used to complete each test is recorded, and the collected test data is input into the evaluation test system program pre-installed in the computer equipment. Based on the preset evaluation ability indicators of each evaluation test and the actual test data of the operating personnel, the single ability evaluation value corresponding to each evaluation test is generated through the relevant scoring algorithm in the system program.

[0069] For example, when conducting a cognitive ability assessment test, when evaluating a certain ability, the system receives the person's own cognitive information input, collects the person's own related collaborative actions, etc., and scores based on the accuracy and consistency of the collected cognitive information and collaborative actions. The system program then outputs the assessment results of the ability.

[0070] Furthermore, in order to improve the accuracy of the assessment, the present application also uses big data analysis capabilities to calculate the person's big data capability assessment value based on actual test data.

[0071] Specifically, for the above-mentioned individual ability assessment results, this application also outputs the corresponding big data analysis ability assessment value through the prediction model pre-trained based on big data analysis technology, and calculates the assessment error (i.e., big data analysis index) between the obtained individual ability assessment results (i.e., single ability assessment value) and the big data analysis ability assessment value. The size of the assessment error is used to assess the high and low cognitive ability of the person. When the relative error is smaller, the big data ability assessment value is higher, and then the final assessment result is higher.

[0072] In one embodiment of the present application, a big data indicator analysis index is calculated for each assessment test, including: for each assessment test, calculating the absolute value of the difference between the single ability assessment value and the big data analysis ability assessment value; dividing the absolute value by the single ability assessment value to obtain the big data indicator analysis index of the corresponding single assessment test, wherein the big data analysis ability assessment value is used as the true result of the target person's cognitive ability, and the big data indicator analysis index is negatively correlated with the final cognitive ability assessment result.

[0073] Specifically, the big data indicator analysis index can be calculated by the following formula:

[0074]

[0075] Among them, the big data indicator analysis index can be regarded as the data difference fixed value AP, which represents the relative error between the computer system evaluation result and the actual result. For each evaluation test, it can be calculated by the above evaluation function. It is only necessary to substitute the evaluation values ​​of different evaluation tests into the single-phase ability evaluation value in the formula, and compare the big data analysis ability with the evaluation test ability. The smaller the data difference fixed value, the smaller the differentiation, and the larger the value of the final cognitive ability evaluation result.

[0076] In order to more clearly explain how to obtain the above-mentioned big data analysis capability evaluation value, the specific process of constructing and training the prediction model of the present application is described in detail below using a model construction method proposed in an embodiment of the present application.

[0077] like Figure 2 As shown, the model training method in the embodiment of the present application includes the following steps:

[0078] Step S201, constructing a functional connectivity network based on the blood oxygen level-dependent BOLD signal in the functional magnetic resonance imaging data.

[0079] It should be noted that the prediction model used in the embodiments of the present application is a cognitive machine learning prediction model based on functional connectivity, which is a model that uses neuroimaging data, preferably functional connectivity information in functional magnetic resonance imaging (fMRI) data in this embodiment, to predict individual cognitive performance or specific cognitive states. This type of model reveals the neural basis of cognitive function by analyzing the synchronous changes of data signals, i.e., functional connectivity, and uses machine learning algorithms to predict individual cognitive abilities.

[0080] Specifically, a prediction model is first constructed. When constructing the model, a functional connectivity matrix or network is constructed by calculating the correlation or other statistical dependency measures between the time series of blood oxygen level-dependent (BOLD) signals in different brain regions. This step can reveal the interaction pattern between brain regions.

[0081] Step S202, extracting features from the functional connection network, and associating the extracted features with preset cognitive performance indicators through a machine learning algorithm to construct a prediction model based on functional connection.

[0082] Specifically, features are extracted from the initially constructed functional connection network. These features may include global or local properties of the network, such as node degree, clustering coefficient, path length, etc., or direct measurements of connection strength.

[0083] Furthermore, according to actual needs, machine learning algorithms such as support vector machines, random forests and neural networks are used to associate the extracted features with preset cognitive performance indicators, which may include: memory test scores, reaction time and working memory capacity, etc. Thus, a prediction model to be trained is constructed.

[0084] Step S203, obtaining cognitive assessment data of a large number of participants through a big data platform, dividing the cognitive assessment data into a training data set and a test set, and optimizing the training data set through a big data optimization algorithm.

[0085] Specifically, when the model is trained, large-scale cognitive assessment data is obtained as training data based on big data technology. As an example, on a big data platform, a large number of participants (for example, more than 1,000 normal people who participate in cognitive assessment) are tested according to the assessment test method in the above embodiment, and all the assessment test data are obtained. All participants are then divided into two groups, with the cognitive assessment test data of one group of people being used as training data and the cognitive assessment test data of the other group of people being used as a test set.

[0086] Furthermore, since the present application obtains a large amount of training data based on big data technology, in order to reduce computing resources and improve the training efficiency of the model, the embodiment of the present application also optimizes the training data through a big data optimization algorithm before performing model training.

[0087] Among them, big data optimization algorithm refers to a series of strategies and techniques used to improve data processing speed, reduce resource consumption, and enhance model accuracy and scalability when processing large-scale data sets. These algorithms usually need to deal with problems such as complexity, high dimensionality, diversity, and rapid growth of data. The combined use of these algorithms and techniques can effectively overcome computing resource limitations, improve processing efficiency, and ensure the quality and reliability of analysis results when processing big data. With the development of big data technology, new algorithms and optimization strategies are constantly emerging to cope with the growing scale and complexity of data. In this application, the corresponding optimization algorithm can be selected according to actual conditions.

[0088] As one possible implementation method, a branch and bound algorithm may be used as a big data optimization algorithm, and the training data set may be optimized through the big data optimization algorithm, including: seeking a global optimal solution for each set of cognitive assessment data through a branch and bound algorithm.

[0089] Specifically, the Branch and Bound algorithm (B&B) is a technique for searching for the optimal solution to a problem on a solution space tree, and is widely used in integer programming, combinatorial optimization, and other discrete optimization problems. It combines the exploration method of depth-first search with an effective pruning strategy to reduce the search space and find the global optimal solution.

[0090] This embodiment can use the branch and bound algorithm to find the global optimal solution, set labels and classification combinations for the collected large-scale training data, so as to achieve streamlined processing of the training data set, reduce complexity, and facilitate subsequent guidance model training.

[0091] The key strategies of the branch and bound algorithm adopted in this embodiment are described in detail below:

[0092] The first step is branching. During the search process, when a node containing a continuous variable (usually an integer) is encountered, the node is "branched", that is, it is split into two or more child nodes according to the possible values ​​of the variable. Each child node corresponds to a certain value of the variable. Choosing which variable to branch and how to branch (for example, the split point of the value) is crucial to the efficiency of the algorithm. Common branching strategies include the maximum difference between the maximum upper bound and the current optimal solution, based on variable frequency, based on influence, etc.

[0093] The second step is bounding. Bounding is to calculate a lower bound (for minimization problems) or upper bound (for maximization problems) for each child node. This bound is used to evaluate whether the subtree may contain a better solution than the currently known optimal solution. Use a heuristic function to estimate the optimal solution in the subtree, such as the linear programming relaxation (LP relaxation) solution or the solution obtained by local search. If the lower bound of a child node (for minimization problems) is higher than the currently known upper bound, it can be pruned because the subtree cannot produce a better solution.

[0094] The third step is pruning. If the lower bound of any node on a branch path is found to be inferior to the upper bound of the currently known best solution, it can be concluded that no better solution will be found by continuing to search along this path, so this path is pruned and its child nodes are no longer searched. Pruning greatly reduces the search space and improves the efficiency of the algorithm.

[0095] The fourth step is the node selection strategy, which selects the next node to expand in the order of joining. This is the most basic selection strategy, including minimum cost / maximum benefit, which selects the node with the minimum cost (for minimization problem) or maximum benefit (for maximization problem) in the active node table as the next expansion node.

[0096] In this embodiment, for the priority queue problem, a priority queue is used to maintain active nodes, and the expansion order is determined according to respective lower bounds or upper bounds and possible other heuristic information to ensure that the most promising nodes are explored first.

[0097] Eventually, when a branch path is fully explored, the algorithm backtracks to the nearest unexplored node and chooses another branch to continue searching until all possible paths have been examined or the optimal solution has been found.

[0098] The execution process of the branch algorithm can be represented by the following key parts:

[0099] First, the algorithm is initialized, and the initial solution space \(S\), the initial upper bound \(UB=+\infty\) (for maximization problems) or the lower bound \(LB=-\infty\) (for minimization problems), and the current best solution \(C^*=-\infty\) (maximization) or \(+\infty\) (minimization) are set. Then, a branch variable is selected from the current solution space to branch, denoted as \(x_i\). For the selected branch variable \(x_i\), two (or more) subproblems are created, usually by splitting the value range of \(x_i\). Each subproblem corresponds to a more restricted solution space. For each subproblem, first solve its linear programming relaxation (LP Relaxation) problem to obtain a boundary value. For the minimization problem, if the solution obtained is \(z\), the upper bound \(UB=\min(UB,z)\) is updated, and \(z\) is also a lower bound of the subproblem. If the bound of a subproblem is not better than the currently known best solution (i.e., for a minimization problem, if \(z\geq C^*\), or for a maximization problem, if \(z\leq C^*\)), prune and abandon further search for this subproblem. From the subproblems that have not been pruned and not fully explored, select the most promising node (based on principles such as minimum cost or maximum benefit) to continue branching. When all branches of a subproblem have been explored and cannot be further split, backtrack to its parent node. If all subproblems of the node have been pruned or explored, update the global best solution \(C^*\) (if a better solution is found) and select the next node to be explored. Finally, the entire process is repeated until the solution space is fully explored or the preset stopping condition is reached (such as the bound converges to be close enough, the time limit is reached, etc.).

[0100] In the above example, the symbol "\" represents an escape character, so as to express the above related content in a regular expression form.

[0101] Step S204, using the optimized training data set, a prediction model is trained based on a deep learning algorithm to predict the fluid intelligence quotient FIQ, crystallized intelligence quotient CIQ and cognitive comprehensive measurement value IC-Cognition for each assessment test.

[0102] Specifically, after the evaluation data collected through big data technology is segmented and the segmented training data set is optimized, the prediction model is trained using the optimized training data set through relevant deep learning technology, and the training prediction model predicts the evaluation result data of the test subjects in various evaluation tests.

[0103] The assessment results include Fluid Intelligence Quotient (FIQ), Crystallized Intelligence Quotient (CIQ) and Integrated Cognitive Assessment or Comprehensive Cognitive Measure (IC-Cognition). The above assessment results are concepts used in psychology and cognitive science to describe and measure different aspects of human intelligence. For example, Crystallized Intelligence Quotient can be used as an indicator in the measurement and evaluation of vocabulary information.

[0104] It should be noted that the specific implementation process of training the prediction model through deep learning technology can refer to the deep learning algorithm of the neural network model in the relevant technology, and this application does not impose any restrictions on this.

[0105] Step S205, calculating multiple evaluation indicators of the model, evaluating the prediction accuracy and generalization ability of the trained prediction model through the multiple evaluation indicators, and correcting the trained prediction model according to the evaluation results.

[0106] Specifically, after the initial training of the prediction model, multiple evaluation indicators such as accuracy, area under the AUC-ROC curve, and R are calculated through various evaluation methods, such as cross-validation and independent test sets. 2 Values, etc., are used to evaluate the prediction accuracy and generalization ability of the trained prediction model to ensure that the model does not overfit the training data. When calculating multiple evaluation indicators of the model, the data in the above test set can be used for related operations.

[0107] Furthermore, the trained prediction model can be modified according to the evaluation results. Model parameters can be adjusted or different model structures can be selected based on the performance of the validation set to improve the model's prediction accuracy and generalization ability.

[0108] For example, in one embodiment of the present application, multiple evaluation indicators include: out-of-sample test-retest reliability, calculating the out-of-sample test-retest reliability, including: inputting the test set into the trained prediction model for testing; obtaining two sets of feature weights of the prediction model during the training process and the testing process; calculating the intra-group correlation coefficient ICC between the two sets of feature weights to quantify the consistency of the feature weights between the two segmentations.

[0109] Among them, out-of-sample test-retest reliability focuses on the consistency of results when the same measurement tool is used to measure different samples at different time points. It is one of the methods to evaluate the stability of measurement tools. Unlike internal test-retest reliability (two or more tests on the same sample), out-of-sample test-retest reliability involves repeated measurements that are far apart in time or on different groups. The assessment of out-of-sample test-retest reliability is crucial to ensure the validity and reliability of research results, especially when it is necessary to verify whether the model can remain stable under different conditions, different populations or over a long time span.

[0110] In the specific implementation, as described above, the evaluation data collected based on the big data technology is segmented, and a set of training data prediction models after segmentation are used to predict FIQ, CIQ and IC-Cognition. Then, in this step, another set of test sets after segmentation is also used to evaluate the test accuracy. Specifically, the data in the test set can be input into the prediction model, and the model is used to predict FIQ, CIQ and IC-Cognition in the same way. The above two processes produce two sets of β coefficients (β1 and β2), and the β coefficient is the feature weight. The intraclass correlation coefficient (ICC) is a test-retest reliability estimation method in fMRI, which is used to quantify the consistency of the β coefficients between two segmentations. The consistency of the two β coefficients is called the out-of-sample test-retest reliability. If a higher ICC value can be obtained, it indicates that the importance of the feature can be reliably located to a specific functional connection. The ICC between the two sets of feature weights can be calculated by the relevant ICC calculation formula.

[0111] Based on the above embodiment, multiple other evaluation indicators may be calculated to evaluate the prediction model. For example, the in-sample retest reliability of the model may also be calculated.

[0112] In-sample test-retest reliability, also known as test-retest reliability or retest reliability, is a method for measuring the consistency of results when the same measurement tool is repeatedly administered to the same group of subjects or samples at different time points. In this embodiment, it can be used as one of the important indicators for evaluating the stability of psychological tests. In-sample test-retest reliability can be used to directly evaluate the stability of the same group of individuals over time, and when interpreting test-retest reliability, the influence of potential sources of bias such as memory effect and practice effect can be focused on.

[0113] For example, the following evaluation indicators can be calculated to evaluate the prediction accuracy of the prediction model, including: Mean Squared Error (MSE): The average of the sum of the squares of the differences between the predicted value and the actual value. Root Mean Squared Error (RMSE): The square root of MSE, which is easier to explain. Absolute Error (MAE): The average of the absolute values ​​of the differences between the predicted value and the actual value. Coefficient of Determination (R 2 or Adjusted R 2 ): It indicates the proportion of the variation explained by the model to the total variation.

[0114] For the evaluation indicators of classification models, if the prediction model involves classification tasks (such as assessing whether the target person has a certain psychological disorder), the following indicators can also be used: Precision: The proportion of positive examples predicted as positive. Recall: The proportion of positive examples correctly predicted as positive. F1 Score: The harmonic mean of Precision and Recall, balancing the two. Confusion Matrix: Detailed display of the number of true positives (TP), true negatives (TN), false positives (FP), and false negatives (FN), which is the basis for calculating the above indicators.

[0115] In one embodiment of the present application, k-fold cross-validation can be used when performing model validation. K-fold cross-validation is a statistical method for evaluating the performance of a machine learning model, which is used to evaluate the performance of the model on unseen data, and is particularly suitable for scenarios with small data sets or model selection and hyperparameter adjustment to avoid overfitting. This method can reduce bias, and by evaluating the model on multiple different subsets, it can reduce the contingency caused by data partitioning and improve the stability of the evaluation results.

[0116] Therefore, through the evaluation method of this embodiment, the system cognitive ability evaluation based on big data can be realized, achieving the effect of accurate evaluation, objectivity and credibility, and solving the problems of single evaluation method, low credibility of evaluation results and strong dependence on experience.

[0117] Based on the above embodiments, it can be understood that in actual applications, due to the influence of some unexpected factors, a large deviation may occur in the ability assessment test results of the target personnel. In order to avoid wasting computing resources, it is necessary to eliminate abnormal single-phase ability assessment values. Therefore, in one embodiment of the present application, after obtaining each single ability assessment value, before performing big data evaluation, it also includes: For the data of the key assessment test, the following ability assessment result function can be used for screening:

[0118]

[0119] Among them, Crg is the ability assessment result function, a is the screening parameter, which can be adjusted according to different assessment tests. The actual situation can be the actual ability assessment value determined by combining manual evaluation and other methods. Through this function, the true results of certain key assessment tests can be obtained, and the assessment data with a result of 1 will be subsequently processed.

[0120] Step S103: Calculate the final cognitive ability assessment result of the target person according to the individual ability assessment value, completion time and big data indicator analysis index of each assessment test.

[0121] In an embodiment of the present application, the final cognitive ability assessment result of the target person is calculated, including: multiplying the single ability assessment value, completion time and big data indicator analysis index of each assessment test by the corresponding first weight coefficient, respectively, to obtain a single-phase score for each assessment test, wherein the big data indicator analysis index is negatively correlated with the single-phase score.

[0122] Specifically, the single ability assessment value, completion time and big data indicator analysis index of each assessment test are multiplied by the weight coefficient of different assessment data to calculate the single-phase score of each test. Among them, the larger the single ability assessment value and the smaller the completion time, the higher the single-phase score, and the lower the big data indicator analysis index, indicating that the assessment error is lower, the higher the single-phase score.

[0123] Furthermore, the single-phase scores of all assessment tests are summarized to calculate the overall comprehensive score of the target person, that is, the final cognitive ability assessment result. When calculating the final cognitive ability assessment result based on the single-phase scores, the single-phase scores of each assessment test are multiplied by the preset weight coefficients of each test.

[0124] In one embodiment of the present application, when determining the weight coefficients of each test, a comparative judgment matrix can be generated based on the hierarchical analysis method, and the second weight coefficients of each evaluation test can be calculated by the comparative judgment matrix. Then, the single-phase scores of each evaluation test can be multiplied by the corresponding second weight coefficient to obtain the final cognitive ability evaluation result.

[0125] Specifically, in this embodiment, the hierarchical analysis method is used to refine the cognitive ability assessment problem into a target layer, an evaluation index layer, and a solution layer. That is, a hierarchical analysis structure of the cognitive ability assessment problem is established, and the cognitive ability assessment problem is refined into a three-layer structure of a target layer, an evaluation index layer, and a solution layer. The target layer is the ultimate goal to be achieved, that is, the comprehensive evaluation of cognitive ability; the indicator layer is the various factors that need to be considered to achieve the overall goal, that is, the various evaluation indicators of cognitive ability; the solution layer is the object of decision-making, that is, the object of comprehensive evaluation of cognitive ability.

[0126] Furthermore, based on the established hierarchical structure diagram, the second weight coefficient of each evaluation test is calculated through a judgment matrix, including: comparing the evaluation indicators of each evaluation test in the evaluation indicator layer in pairs, and constructing the comparison judgment matrix according to the obtained importance comparison results; calculating the maximum eigenvalue root of the comparison judgment matrix and the eigenvector matrix corresponding to the maximum eigenvalue root, and determining the second weight coefficient of each evaluation indicator according to the eigenvector matrix.

[0127] Specifically, the weight of each element is obtained by judging the eigenvector of the matrix, and the weighted sum method is used to merge them, thereby obtaining each second weight coefficient. First, by comparing the comprehensive evaluation indicators of each ability layer in the indicator layer in the hierarchical structure diagram, a comparison judgment matrix C = (cij) n × n is constructed, where n represents the number of cognitive ability evaluation indicators, and cij represents the comparison result of the importance of the i-th indicator and the j-th indicator. For example, the comparison judgment matrix can be expressed by the following formula:

[0128]

[0129] Each element in the matrix satisfies the following conditions: cij>0, cii=1, cij=1 / cji. The method for determining the value of cij can be shown in the following Table 2:

[0130] Table 2 Importance comparison result table

[0131] Factor i is better than factor j <![CDATA[c ij Value]]> Equally important 1 Slightly important 3 Strong and important 5 Strongly important 7 Extremely important 9 The middle value of two adjacent judgments 2、4、6、8

[0132] Furthermore, after constructing the judgment matrix according to the above embodiment, the maximum eigenvalue λmax of the judgment matrix and the eigenvalue matrix W corresponding to the eigenvalue are obtained. Any item Wj in the eigenvalue matrix is ​​the weight of the j-th indicator, that is, the second weight coefficient of the j-th evaluation test.

[0133] Therefore, this application collects and analyzes the data of operators through big data technology, and can evaluate based on the data to ensure the objectivity and scientificity of the evaluation results. At the same time, in the evaluation process, the relevant theories of system science can be integrated to effectively network complex problems. In this way, the difficulty of cognitive ability evaluation can be effectively reduced, and the accuracy of cognitive ability evaluation can be improved.

[0134] In summary, the personnel cognitive ability assessment method based on big data of the embodiment of the present application uses computer technology and big data analysis technology to conduct a multi-dimensional comprehensive assessment of the cognitive ability of the operating personnel, and adopts a set of cognitive test schemes and related algorithms to achieve a comprehensive assessment of cognitive ability for each major aspect of the cognitive ability of the operating personnel in a relatively closed environment, thereby improving the comprehensiveness of the assessment. In addition, the method adopts standardized assessment steps, model tools and algorithms, and with the help of big data analysis technology, it is possible to ensure the objectivity, scientificity and accuracy of the assessment results. The method integrates the assessment methods of computerized cognitive assessment and psychological principles, and through systematic measurement and assessment, it is possible to accurately measure the level of individuals in various cognitive fields, give full consideration to the complexity and diversity of individuals, and improve the pertinence of the assessment. Thus, the method improves the accuracy, intelligence and comprehensiveness of cognitive assessment.

[0135] Based on the above embodiments, using the evaluation results obtained by the cognitive ability evaluation method, the present application can also perform cognitive ability recovery training on the evaluated target personnel. It should be noted that the cognitive ability recovery training program in the relevant embodiments is generally based on the overall level of individual cognitive ability in the conclusion of the evaluation report after the cognitive evaluation of the operating personnel, and the corresponding intervention suggestions and training plans are proposed. There is a lack of in-depth correlation analysis between cognitive recovery training and cognitive evaluation. Moreover, the recovery method is single, and training is carried out through a single training method. There is a lack of diversified training recovery methods, and the final recovery effect does not meet the expected goals.

[0136] In addition, traditional cognitive training usually adopts certain experimental paradigms, and the training process is cumbersome and boring. Among the current methods of cognitive enhancement, non-invasive brain stimulation technology such as transcranial electrical stimulation, although it has shown the effect of cognitive enhancement, its application effect in actual environment still needs further verification, and studies have shown that it can cause rare but very serious side effects. Although cognitive enhancement drugs can be used to improve cognitive behavior and work ability, there are certain risks and side effects.

[0137] To this end, this application also proposes a method for restoring the cognitive ability of workers based on evaluation data. Compared with the cognitive training methods in the above-mentioned related technologies, this application uses the final cognitive ability evaluation results and other data obtained by the above-mentioned cognitive ability evaluation method as the specific data basis and support for recovery training, and adopts a computerized cognitive training method, which has the advantages of being non-invasive, non-intrusive and easy to operate, and has good application prospects. In addition, based on the comprehensive evaluation results of the cognitive ability of personnel, this method automatically matches a multi-dimensional recovery training program for specific individuals, improves the pertinence and diversity of cognitive recovery, and enhances the training effect.

[0138] Figure 3 A flowchart of a method for restoring a person's cognitive ability based on evaluation data proposed in an embodiment of the present application is shown in FIG. Figure 3 As shown, the method comprises the following steps:

[0139] Step S301: Acquire cognitive ability assessment data of the target person to be trained, and determine the classification of the target person's cognitive ability to be restored and the corresponding degree of decline according to the cognitive ability assessment data.

[0140] Among them, the cognitive ability assessment data is obtained by evaluating the target personnel using the big data-based operator cognitive ability assessment method of any one of the above-mentioned first aspect embodiments, and may include various types of cognitive ability assessment data such as the final cognitive ability assessment result of the target personnel obtained in the above-mentioned embodiments and single ability assessment values.

[0141] It should be noted that the execution subject of the method for restoring the cognitive ability of operators based on evaluation data in this application can be the computer device proposed in this application, that is, all steps in the recovery training method are executed by the computer device, and the computer device can be used to collect data, analyze and calculate, formulate training recovery plans, and interact with operators through relevant hardware devices to achieve training of operators. Therefore, this application realizes the recovery of cognitive ability of target operators through computerized cognitive training.

[0142] Specifically, this application organically integrates the comprehensive assessment of cognitive abilities and cognitive recovery training. According to the comprehensive cognitive ability assessment results of the target personnel to be trained, it automatically matches and gives a recovery training program, and then guides the personnel to complete the training tasks according to the training program. Therefore, this application first obtains the cognitive ability assessment data of the existing target personnel, and the cognitive ability assessment data is determined by pre-assessing the cognitive abilities of the target personnel based on big data. Then analyze the assessment data, and determine the classification of the cognitive abilities to be restored and the corresponding degree of decline of the target personnel based on the unqualified data. That is, with the help of a comprehensive assessment of the individual's cognitive abilities, the specific type and degree of cognitive decline are determined.

[0143] As a possible implementation method, when the cognitive ability of the target person is evaluated in advance based on big data technology, the following steps may be included: first, the basic data of the target person to be evaluated is collected, and based on multiple preset cognitive ability evaluation categories, multiple evaluation tests under each cognitive ability evaluation category are performed on the target person in turn; then, the test data, single ability evaluation value and completion time of each evaluation test are obtained, and the basic data and each test data are input into the prediction model, and the big data analysis index of each evaluation test is calculated based on the big data analysis ability evaluation value output by the prediction model, wherein the prediction model is pre-trained based on big data analysis technology, and the big data analysis index represents the error between the big data analysis ability evaluation value and the single ability evaluation value; finally, according to the single ability evaluation value, completion time and big data indicator analysis index of each evaluation test, the final cognitive ability evaluation result (i.e., the comprehensive evaluation score) of the target person is calculated.

[0144] Specifically, in this embodiment, the big data indicator analysis index of each assessment test is calculated, including: for each assessment test, calculating the absolute value of the difference between the single ability assessment value and the big data analysis ability assessment value; dividing the absolute value by the single ability assessment value to obtain the corresponding big data indicator analysis index. The big data indicator analysis index is negatively correlated with the final cognitive ability assessment result.

[0145] Furthermore, when calculating the final cognitive ability assessment result of the target person, it includes: multiplying the single ability assessment value, completion time and big data indicator analysis index of each assessment test by the corresponding weight coefficient, and obtaining the single score of each assessment test, wherein the big data indicator analysis index is negatively correlated with the single score; generating a comparative judgment matrix based on the hierarchical analysis method, calculating the weight coefficient of each assessment test through the comparative judgment matrix, and multiplying the single score of each assessment test by the corresponding weight coefficient to obtain the final cognitive ability assessment result.

[0146] For example, when evaluating the cognitive ability of personnel, the cognitive evaluation of the operating personnel is carried out from multiple key cognitive ability categories according to the content of the overall cognitive ability system of the personnel. For example, the embodiment of the present application can detect and evaluate the cognitive ability of personnel from eight cognitive ability categories, namely, long-term memory search, short-term memory, processing speed, auditory processing, visual spatial thinking, knowledge comprehension, fluid reasoning and quantitative ability. A variety of specific test methods can be carried out under each cognitive ability evaluation category. For example, the various evaluation tests that can be carried out under each evaluation category are shown in Table 3 below:

[0147] Table 3 Evaluation test table

[0148]

[0149]

[0150] According to the evaluation test contents shown in Table 3, the operators can be subjected to various tests in sequence through computer equipment according to the evaluation indicators of each evaluation test.

[0151] Thus, the cognitive ability assessment data of the target person to be trained obtained in this step includes the total score of the final cognitive ability assessment result of the target person and the individual scores of each assessment test in the above Table 1. Furthermore, the cognitive ability classification to be restored determined from the cognitive ability assessment can be the cognitive ability classification to which a certain assessment test with an unqualified score in the above Table 1 belongs, and the specific degree of decline of the cognitive ability classification can be represented by the scores of each assessment test in the classification.

[0152] Step S302: Set a variety of cognitive recovery training, solve the causal relationship between cognitive ability assessment and cognitive recovery training based on the Bayesian network, substitute the cognitive ability classification to be restored and the corresponding degree of decline into the causal relationship, and formulate corresponding recovery training tasks and training cycles for the target personnel.

[0153] Specifically, the present application can pre-formulate a variety of recovery trainings based on actual factors such as the overall system of personnel cognitive abilities and the training interactive operations executable by computer devices, and determine the specific implementation methods of each recovery training.

[0154] In one embodiment of the present application, the multiple cognitive recovery trainings set up include multiple main training projects and multiple auxiliary training projects, wherein the multiple main training projects include: attention training, cognitive flexibility training, perception training, thinking ability training, reaction training and memory training; the multiple auxiliary training projects include: mindfulness meditation training, auditory stimulation training, tactile stimulation training and olfactory stimulation training.

[0155] Specifically, the set recovery training can be adjusted according to the actual situation. For example, in this embodiment, as shown in Table 4 below, 7 categories of cognitive ability recovery training content (i.e., the above-mentioned training items) are set, and each category of training content may include multiple related specific recovery training methods.

[0156] Table 4 Cognitive ability recovery training table

[0157]

[0158] The following is a detailed description of each cognitive recovery training included in the main training items in Table 4.

[0159] Among them, the principle of the first type of Schulte grid training is to train the visual search ability and attention allocation ability by requiring the subjects to find the numbers or letters in the grid in a certain order. The relevant algorithm in the computer system will record the reaction time and accuracy of the subjects, thereby evaluating their attention level. That is, this training method can be used as an implementation method of the above-mentioned attention training.

[0160] The second type is mental arithmetic training. The principle is to strengthen the subject's mathematical memory and calculation ability through continuous mental arithmetic exercises, such as addition, subtraction, multiplication and division. The algorithm will adjust the questions according to the difficulty of the questions and the subject's answers.

[0161] The third type is long-term memory training. The principle is to make full use of the human brain's ability to store and retrieve visual information as well as its ability to associate and relate. When performing a picture recognition task, the subject identifies a previously seen picture from a group of pictures. This process involves the extraction and comparison of picture information stored in long-term memory. Accurate recognition indicates successful storage and retrieval. Associative memory training allows subjects to associate specific stimuli (such as pictures) with other stimuli. For example, when a subject sees a picture of an apple, he or she associates it with concepts such as "red", "fruit", and "sweet", thereby establishing a richer memory network and improving the storage and retrieval capabilities of long-term memory. Combining the two for training can comprehensively improve the subject's memory of visual information and the ability to strengthen memory through association.

[0162] The fourth type is short-term memory training. Its principle is to comprehensively examine and improve the short-term memory ability of the subjects. Short-term memory is a memory system that maintains and processes limited information in a short period of time. In the digital span test, the subjects remember and recall a series of numbers in order, which involves the temporary storage and repetition of digital information. The ability to accurately recall more digits indicates that the short-term memory capacity is large and can maintain more information in a short period of time. The reverse digital training increases the difficulty of the task. The subjects not only have to remember the order of numbers, but also need to perform reverse operations when recalling, which involves a deeper processing and conversion of digital information, and requires stronger attention and control. Combining the two for training can improve the depth and flexibility of the subjects' information processing, enhance the short-term memory function, and comprehensively challenge and improve the subjects' short-term memory performance in different situations.

[0163] The fifth type is reaction time training. The principle is to train the subjects’ reaction speed and decision-making ability by presenting different stimuli and requiring them to respond quickly. The algorithm will record the subjects’ reaction time, accuracy, and reaction pattern to evaluate their attention allocation and decision-making ability.

[0164] The sixth type is mental rotation training. The principle is to present a two-dimensional image through a computer device and ask the subject to imagine its three-dimensional form, training their spatial cognition and image processing abilities. The algorithm will analyze the subject's imagination process and answers to evaluate their spatial perception and cognitive abilities.

[0165] The seventh type is Stroop task training. The principle is to present stimuli with inconsistent colors and words, requiring the subjects to ignore the meaning of the words and focus on the colors, so as to train their interference suppression and attention control abilities. The algorithm will record the subjects' reaction time, error rate, and attention allocation to evaluate their ability to resist interference.

[0166] The eighth type is continuous operation training. The principle is to train the subject's task switching ability and persistence by requiring the subject to continuously perform a series of task operations. The algorithm will dynamically adjust the training rhythm and difficulty according to the difficulty of the task and the subject's performance to promote the recovery of their cognitive flexibility.

[0167] The following is a detailed description of the auxiliary training items, namely the various cognitive recovery training included in the multi-sensory stimulation training in Table 2.

[0168] Ninth, mindfulness meditation training. The principle is that mindfulness meditation promotes the reconnection of the brain's neural network by stimulating the brain's neuroplasticity mechanism. Long-term mindfulness meditation can enhance the activity of the prefrontal cortex and anterior cingulate gyrus, which are closely related to cognitive functions such as attention and self-awareness. Mindfulness meditation helps to improve self-awareness and helps individuals better identify and deal with emotions. Through mindfulness meditation, negative emotions such as anxiety and depression can be reduced, and self-control and adaptability can be enhanced. Mindfulness meditation also emphasizes focus on current perception, which helps to improve concentration. By practicing mindfulness meditation, individuals can better allocate mental resources, focus on the tasks that need to be handled at the moment, and avoid being distracted by irrelevant things.

[0169] The tenth type is auditory training, which mainly relies on music VR. The principle is to use music VR technology to provide personalized music experience, and adjust music parameters such as rhythm, melody and volume through audio processing algorithms to stimulate the subjects' brains and improve their emotional state.

[0170] The eleventh type is tactile stimulation recovery. The main plan is to change the frequency of the vibration electrode, collect the vibration signal of the subject, analyze his / her attention, relaxation and other states, and provide personalized tactile stimulation feedback based on the analysis results.

[0171] The twelfth type is olfactory stimulation recovery, which mainly relies on smell therapy. Its principle is to use the simulated natural forest and flower smells produced by aromatherapy to stimulate the subject's olfactory system, analyze their response to the smell through the olfactory perception algorithm, and improve the subject's emotional state and cognitive function.

[0172] It should be noted that in the above-mentioned cognitive recovery training, for the training using the computer human-computer interaction interface, the interaction interface can be made interesting during the training process, for example, various types of interesting pictures can be displayed and the content displayed on the interface can be enriched. For example, small human-computer interaction games can be added during the attention training and reaction training to increase the fun of the recovery training and avoid the boring training process causing mental fatigue of the trainees.

[0173] Furthermore, based on the cognitive assessment results obtained in the previous step, the training that the current user needs to perform is selected from the multiple cognitive recovery trainings that have been set above, so as to generate a personalized cognitive recovery plan for the individual and perform corresponding recovery training according to the items with insufficient assessment indicators.

[0174] As a possible implementation method, the computer system first analyzes the obtained assessment results. By comprehensively analyzing the individual assessment results, it identifies the specific deficiencies of the target person in eight cognitive ability categories, including visual-spatial thinking, fluid reasoning, auditory processing, knowledge comprehension, quantitative ability, long-term memory search, short-term memory, and processing speed. Then, the training goals are set. Based on the assessment results, the system will set specific recovery goals, such as improving several of the above eight cognitive abilities. Finally, a cognitive recovery training plan is formulated. According to the items and scores of the assessment indicators that are insufficient, the system will guide the user to operate the corresponding training tasks and training cycles according to the corresponding relationship between cognition and recovery.

[0175] Among them, in order to ensure the rationality of the formulated recovery training tasks, it is necessary to ensure the accuracy of the causal relationship between cognitive ability assessment and cognitive recovery training. In this application, the causal relationship between cognitive ability assessment and cognitive recovery training is solved based on the Bayesian network, and then the classification of cognitive abilities to be restored by the current target person and the corresponding degree of decline obtained by analysis are substituted into the causal relationship to determine the recovery training tasks that match the target person and the number of training cycles required.

[0176] In order to more clearly illustrate the specific implementation process of solving the causal relationship between cognitive ability assessment and cognitive recovery training based on the Bayesian network in this application, a solution method proposed in an embodiment of this application is exemplified below. Figure 4 This is a flow chart of a method for solving a causal relationship proposed in an embodiment of the present application. Figure 4 As shown, the method comprises the following steps:

[0177] Step S401, constructing a directed acyclic graph based on the relationship characteristics between cognitive ability assessment and cognitive recovery training, wherein the structural characteristics of the directed acyclic graph represent the direct causal relationship between cognitive variables.

[0178] Specifically, the relationship between cognitive ability assessment and cognitive recovery training is analyzed, including the characteristics that the subsequent variable will not cause the change of the preceding variable and the causal relationship between the two variables is linear. Based on these relationship characteristics, the causal relationship between cognitive ability assessment and cognitive recovery training can be represented by a directed acyclic graph (DAG).

[0179] Step S402, converting the directed acyclic graph into an equation form.

[0180] Specifically, the equation form is expressed by the following formula:

[0181] X=BX+e

[0182] Among them, X represents the sample matrix, B represents the coefficient matrix, e represents the disturbance matrix, and the sample matrix contains cognitive variables.

[0183] It should be noted that, after the sample matrix X is sorted according to the causal order between the variables in the embodiment of the present application, the coefficient matrix B can be arranged into a lower triangular matrix (the diagonal of the matrix is ​​all zero values).

[0184] For example, assuming that a directed acyclic graph has four variables from x1 to x4, and the causal relationship between the variables is linear, the causal relationship of the DAG can be expressed by the following formula:

[0185] x2=b2l*x1+e2

[0186] x3=b32*x2+e3

[0187] x4=b42*x2+b43*x3+e4

[0188] Among them, b represents the coefficient between different variables, and e represents the disturbance value.

[0189] Furthermore, the above equations are converted into the matrix form represented by the following formula:

[0190]

[0191] The formula in matrix form is the specific expanded form of the above equation X=BX+e.

[0192] Step S403, solving the sample matrix in the equation, decomposing the sample matrix into a target matrix and a disturbance matrix, and converting the target matrix into a lower triangular matrix to calculate a coefficient matrix, and solving the causal relationship according to the calculated coefficient matrix.

[0193] Specifically, the sample matrix X in the above equation X=BX+e is solved to obtain the sample matrix in the form of the target matrix multiplied by the perturbation matrix, and then the target matrix=(IB)-1 can be obtained, where I represents the identity matrix. It can be understood that after knowing the causal order of the variable X, the coefficient matrix B can be arranged into a lower triangular matrix, and the target matrix can also be arranged into a lower triangular matrix.

[0194] Therefore, the embodiment of the present application decomposes the sample matrix X into two parts, the target matrix and the perturbation matrix e, through a related algorithm, and then sorts the target matrix into a lower triangular matrix to solve the coefficient matrix B. Since the perturbation matrix e obeys a non-Gaussian distribution and is independent of each other, it does not need to be solved. The causal relationship can be solved using the equation in the above embodiment based on the calculated coefficient matrix.

[0195] In order to ensure the accuracy of the solved causal relationship, in one embodiment of the present application, when solving the causal relationship between cognitive ability assessment and cognitive recovery training based on the Bayesian network, it also includes: based on multiple probability theory principles, reasoning about the causal relationship through the Bayesian network, wherein the multiple probability theory principles include forward reasoning principle, backward reasoning principle, causal discovery principle and structural learning principle.

[0196] Specifically, Bayesian theorem and other probability theory principles can be used to assist Bayesian network reasoning. Among them, the probability theory principles used include the following:

[0197] Forward reasoning principle: Given some evidence (i.e., the observed values ​​of some nodes in the DAG), calculate the posterior probability of other nodes. This can be achieved through variable elimination, belief propagation, or sum-product algorithm.

[0198] The principle of reverse reasoning: that is, causal reasoning, if the goal is to explore the impact of a certain intervention or change in a certain variable (such as training effect) on other variables, reverse reasoning can be performed. It usually involves calculating the do-operation (do-calculus), that is, calculating the probability distribution under certain artificially set variable states to simulate causal effects.

[0199] Causal discovery principle: In practical applications, since the DAG network structure is often unknown, it needs to be learned from data. Using causal discovery algorithms, such as the PC algorithm, GES (Greedy Equivalence Search) or FCI (FastCausal Inference), can help infer possible causal structures from observed data.

[0200] Structural learning principles: Structural learning is the process of automatically learning the structure of a Bayesian network from data. It usually involves using statistical tests to identify conditional independence between variables and thus infer the directed edges of the network.

[0201] Further, as shown in Table 4 in the above example, each training item includes multiple cognitive recovery trainings. When performing recovery training for a cognitive ability classification to be recovered, multiple cognitive recovery trainings may be required. In order to ensure the rationality of the arrangement of each cognitive recovery training, in one embodiment of the present application, after solving the causal relationship according to the calculated coefficient matrix, the following steps are also included:

[0202] First, specific settings are made for each node in the directed acyclic graph. The setting process here can refer to the implementation method in the relevant technology, and this application does not limit this. Its purpose is to provide a key basis for the subsequent determination of the arrangement of cognitive recovery training.

[0203] Then, based on this setting, the specific arrangement of each cognitive recovery training corresponding to each cognitive ability classification is clarified. Further, when each cognitive ability classification in implementation corresponds to a training project, the cognitive ability classification can be restored through the multiple recovery trainings included in the training project shown in Table 4, and based on the previous settings, the specific arrangement of each cognitive recovery training corresponding to each cognitive ability classification can be reasonably determined, so that different cognitive recovery trainings can be planned more scientifically, so that they are more in line with the individual's cognitive ability recovery needs, and the effect and pertinence of the recovery training can be improved.

[0204] Therefore, based on the above embodiment, the computer system substitutes the unqualified cognitive ability classification and the individual scores of its various assessment tests into the solved causal relationship, and the system will automatically match the various recovery trainings that the target person needs to undergo and the proportion of each recovery training, and then generate a recovery training task for the target person.

[0205] The following is an exemplary description in conjunction with a specific embodiment in a practical application. Figure 5 As shown, in this embodiment, the acquired cognitive ability assessment data includes the target person's comprehensive cognitive ability assessment score and the individual scores of various assessment tests, and a corresponding recovery training task and training cycle are formulated for the target person, including: determining whether the target person should undergo recovery training and the number of training cycles based on the score range of the comprehensive cognitive ability assessment score; screening out various target assessment tests with scores below a preset threshold, determining the cognitive ability category to which each target assessment test belongs, and determining each recovery training to be performed based on the causal relationship between cognitive ability assessment and cognitive recovery training.

[0206] Step S303: According to the recovery training task, multimodal intervention training is performed on the target person, and the changes in the cognitive ability of the target person are continuously monitored during the training process, and the recovery plan is adjusted in real time according to the changes in cognitive ability.

[0207] Specifically, according to the formulated targeted recovery training tasks, the computer system guides the target personnel to carry out a variety of corresponding recovery training, and conducts multimodal intervention training on the target personnel, including cognitive training, physical therapy, smell therapy, etc., to improve the individual's cognitive ability. Furthermore, during the recovery training process, the target personnel are continuously monitored and the training tasks are adjusted. That is, during the recovery process, the changes in the individual's cognitive ability are continuously monitored, and the recovery plan is adjusted in time according to the situation. For example, for the cognitive ability that has been restored, the relevant training is stopped, etc., which can be implemented through regular cognitive function assessment.

[0208] Furthermore, a cognitive ability assessment can be conducted again after training to verify the training effect. In one embodiment of the present application, after the target person is subjected to multimodal intervention training, it also includes: re-evaluating the classification of cognitive abilities to be restored in the initial cognitive ability assessment, and updating the scores of the assessment tests that do not meet the standards in the classification of cognitive abilities to be restored; by comparing the scores of the initial assessment and the updated scores, the recovery effect of this round of recovery training is determined; when there are still assessment tests that do not meet the standards, the recovery training tasks are adjusted according to the recovery effect, and the target person is subjected to the next round of recovery training according to the adjusted recovery training tasks.

[0209] Specifically, if Figure 5 As shown, after the current round of training is completed, another evaluation is conducted. When conducting the second evaluation, the evaluation items that do not meet the evaluation score are selected for evaluation based on the most recent evaluation results. After the evaluation, a comprehensive calculation is performed again, and the training effectiveness is judged based on the increase or decrease in the score.

[0210] The implementation process of re-evaluating cognitive ability can refer to the description of pre-evaluating cognitive ability of target personnel based on big data in step S301. The evaluation principle is the same and will not be repeated here.

[0211] Therefore, after the re-evaluation is completed, the system of the embodiment of the present application forms an evaluation result again. If the evaluation is qualified, it means that the recovery training is completed; if there are still deficiencies, the recovery training needs to be continued until the expected recovery effect is achieved.

[0212] In summary, the method for recovering the cognitive ability of operators based on evaluation data in the embodiment of the present application automatically matches a multidimensional recovery training program for a specific individual based on the comprehensive evaluation results of the cognitive ability of the personnel, associates cognitive evaluation with recovery training, and automatically performs recovery training on operators in a variety of training and recovery methods through computer technology. The method provides personalized training plans and recovery programs for the type and degree of cognitive decline of individuals, and can restore the cognitive ability of individuals in a targeted manner through systematic evaluation and training. The method combines various test training tasks and stimulation recovery methods in terms of vision, smell, hearing, vibration and touch, and improves the training effect from multi-dimensional intervention. The method can also provide real-time feedback on the training performance and progress of individuals, provide timely evaluation and adjustment suggestions for individuals, help individuals better understand their cognitive status, and guide individuals to conduct targeted training, and achieve real-time feedback and iterative recovery training of the degree of recovery. As a result, the method improves the pertinence, diversity and systematicity of cognitive ability recovery training, can significantly enhance the recovery training effect, and ensure the normal operation of operators.

[0213] In order to implement the above embodiment, the present application also proposes a human cognitive ability assessment device based on big data, Figure 6 This is a schematic diagram of the structure of a device for evaluating human cognitive ability based on big data proposed in an embodiment of the present application, such as Figure 6 As shown, the device includes a testing module 100 , a big data analysis module 200 and a computing module 300 .

[0214] Among them, the test module 100 is used to collect basic data of the target person to be evaluated, and based on multiple preset cognitive ability evaluation categories, perform multiple evaluation tests under each cognitive ability evaluation category on the target person in turn.

[0215] The big data analysis module 200 is used to obtain the test data, single capability evaluation value and completion time of each evaluated test, and input the basic data and each test data into the prediction model, and calculate the big data analysis index of each evaluation test based on the big data analysis capability evaluation value output by the prediction model, wherein the prediction model is pre-trained based on big data analysis technology, and the big data analysis index represents the error between the big data analysis capability evaluation value and the single capability evaluation value.

[0216] The calculation module 300 is used to calculate the final cognitive ability assessment result of the target person according to the single ability assessment value, completion time and big data indicator analysis index of each assessment test.

[0217] In one embodiment of the present application, the big data analysis module 200 is specifically used to: for each assessment test, calculate the absolute value of the difference between the single ability assessment value and the big data analysis ability assessment value; divide the absolute value by the single ability assessment value to obtain the big data indicator analysis index of the corresponding single assessment test, wherein the big data analysis ability assessment value is used as the true result of the target person's cognitive ability, and the big data indicator analysis index is negatively correlated with the final cognitive ability assessment result.

[0218] In one embodiment of the present application, the big data analysis module 200 is specifically used to: construct a functional connection network based on the blood oxygen level-dependent BOLD signal in the functional magnetic resonance imaging data; extract features from the functional connection network, and associate the extracted features with preset cognitive performance indicators through a machine learning algorithm to construct a prediction model based on functional connectivity; obtain cognitive assessment data of a large number of participants through a big data platform, divide the cognitive assessment data into a training data set and a test set, and optimize the training data set through a big data optimization algorithm; use the optimized training data set to train a prediction model based on a deep learning algorithm to predict the fluid intelligence quotient FIQ, crystallized intelligence quotient CIQ and cognitive comprehensive measurement value IC-Cognition for various assessment tests; calculate multiple evaluation indicators of the model, evaluate the prediction accuracy and generalization ability of the trained prediction model through multiple evaluation indicators, and correct the trained prediction model according to the evaluation results.

[0219] In one embodiment of the present application, the big data analysis module 200 is also used to: input the test set into the trained prediction model for testing; obtain two sets of feature weights of the prediction model during the training process and the testing process; calculate the intra-group correlation coefficient ICC between the two sets of feature weights to quantify the consistency of the feature weights between the two segmentations.

[0220] In one embodiment of the present application, the big data analysis module 200 is further used to: seek a global optimal solution for each set of cognitive assessment data through a branch and bound algorithm.

[0221] In one embodiment of the present application, the calculation module 300 is specifically used to: multiply the single ability assessment value, completion time and big data indicator analysis index of each assessment test by the corresponding first weight coefficient, respectively, to obtain the single-phase score of each assessment test, and the big data indicator analysis index is negatively correlated with the single-phase score; generate a comparative judgment matrix based on the hierarchical analysis method, calculate the second weight coefficient of each assessment test through the comparative judgment matrix, and multiply the single-phase score of each assessment test by the corresponding second weight coefficient to obtain the final cognitive ability assessment result.

[0222] In one embodiment of the present application, the calculation module 300 is specifically used to: compare the evaluation indicators of each evaluation test in the evaluation indicator layer in pairs, and construct a comparison judgment matrix according to the obtained importance comparison results; calculate the maximum eigenvalue root of the comparison judgment matrix and the eigenvector matrix corresponding to the maximum eigenvalue root, and determine the second weight coefficient of each evaluation indicator according to the eigenvector matrix.

[0223] In summary, the big data-based operator cognitive ability assessment device of the embodiment of the present application uses computer technology and big data analysis technology to conduct a multi-dimensional comprehensive assessment of the cognitive ability of operators. For each major aspect of the cognitive ability of operators in a relatively closed environment, a set of cognitive test schemes and related algorithms are used to achieve a comprehensive assessment of cognitive ability, thereby improving the comprehensiveness of the assessment. In addition, the device integrates computerized cognitive assessment and psychological principle assessment methods. Through systematic measurement and assessment, it can accurately measure the level of individuals in various cognitive fields, fully consider the complexity and diversity of individuals, and improve the pertinence of the assessment.

[0224] In order to implement the above embodiment, the present application also proposes a device for restoring human cognitive ability based on evaluation data. Figure 7 This is a schematic diagram of the structure of a device for restoring human cognitive ability based on evaluation data proposed in an embodiment of the present application. Figure 7 As shown, the device includes an acquisition module 10, a training formulation module 20 and a training execution module 30.

[0225] Among them, the acquisition module 10 is used to obtain the cognitive ability assessment data of the target person to be trained, and determine the cognitive ability classification to be restored and the corresponding degree of decline of the target person according to the cognitive ability assessment data, wherein the cognitive ability assessment data is obtained by evaluating the target person using the operator cognitive ability assessment method based on big data as described in any one of the first aspects above.

[0226] The training formulation module 20 is used to set up a variety of cognitive recovery training, solve the causal relationship between cognitive ability assessment and cognitive recovery training based on the Bayesian network, substitute the cognitive ability classification to be restored and the corresponding degree of decline into the causal relationship, and formulate corresponding recovery training tasks and training cycles for the target personnel.

[0227] The training execution module 30 is used to perform multimodal intervention training on the target person according to the recovery training task, and continuously monitor the changes in the cognitive ability of the target person during the training process, and adjust the recovery plan in real time according to the changes in cognitive ability.

[0228] In one embodiment of the present application, the training formulation module 20 is specifically used to: construct a directed acyclic graph DAG based on the relationship characteristics between cognitive ability assessment and cognitive recovery training, and the structural characteristics of the directed acyclic graph represent the direct causal relationship between cognitive variables; convert the directed acyclic graph into an equation form shown in the following formula:

[0229] X=BX+e

[0230] Among them, X represents the sample matrix, B represents the coefficient matrix, e represents the disturbance matrix, and the sample matrix contains cognitive variables;

[0231] The sample matrix in the equation is solved, the sample matrix is ​​decomposed into a target matrix and a perturbation matrix, and the target matrix is ​​transformed into a lower triangular matrix to calculate the coefficient matrix, and the causal relationship is solved according to the calculated coefficient matrix.

[0232] In one embodiment of the present application, the training formulation module 20 is also used to: infer causal relationships through a Bayesian network based on multiple probability theory principles, wherein the multiple probability theory principles include forward reasoning principle, backward reasoning principle, causal discovery principle and structural learning principle.

[0233] In one embodiment of the present application, the training execution module 30 is also used to: re-evaluate the classification of cognitive abilities to be restored in the initial cognitive ability assessment, and update the scores of each assessment test that does not meet the standards in the classification of cognitive abilities to be restored; determine the recovery effect of this round of recovery training by comparing the scores of the initial assessment with the updated scores; when there are still assessment tests that do not meet the standards, adjust the recovery training tasks according to the recovery effect, and conduct the next round of recovery training for the target personnel according to the adjusted recovery training tasks.

[0234] In one embodiment of the present application, the training formulation module 20 is also used to: determine whether the target person should undergo recovery training and the number of training cycles based on the score range of the comprehensive score of the cognitive ability assessment; screen out each target assessment test with a score lower than a preset threshold, determine the cognitive ability category to which each target assessment test belongs, and determine the recovery training to be performed based on the causal relationship between the cognitive ability assessment and cognitive recovery training.

[0235] In summary, the device for recovering the cognitive ability of workers based on evaluation data in the embodiment of the present application improves the pertinence, diversity and systematicness of cognitive ability recovery training, can significantly enhance the recovery training effect, and ensure the normal operation of workers.

[0236] In order to implement the above embodiment, the present application also proposes a computer device, such as Figure 8As shown, the computer device 800 includes: a processor 810; a memory 820 for storing executable instructions of the processor 810; wherein the processor 410 is configured to execute instructions to implement a personnel cognitive ability assessment method based on big data as described in any of the first aspect embodiments above, or to implement a personnel cognitive ability recovery method based on assessment data as described in any of the second aspect embodiments above.

[0237] In order to implement the above embodiments, the present application also proposes a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements a personnel cognitive ability assessment method based on big data as described in any of the above-mentioned first aspect embodiments, or implements a personnel cognitive ability recovery method based on assessment data as described in any of the above-mentioned second aspect embodiments.

[0238] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0239] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0240] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0241] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute the instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0242] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0243] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0244] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0245] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for evaluating personnel cognitive ability based on big data, characterized in that: Executed by a computer device, the method comprises the following steps: Collecting basic data of the target person to be evaluated, and based on a plurality of preset cognitive ability evaluation categories, sequentially performing a plurality of evaluation tests under each of the cognitive ability evaluation categories on the target person; Obtain the test data, single capability evaluation value and completion time of each of the evaluation tests, and input the basic data and each test data into a prediction model, and calculate the big data analysis index of each evaluation test based on the big data analysis capability evaluation value output by the prediction model, wherein the prediction model is pre-trained based on big data analysis technology, and the big data analysis index represents the error between the big data analysis capability evaluation value and the single capability evaluation value; The final cognitive ability assessment result of the target person is calculated based on the single ability assessment value of each assessment test, the completion time and the big data indicator analysis index.

2. The method according to claim 1, characterized in that: The calculation of the big data indicator analysis index for each evaluation test includes: For each assessment test, calculating the absolute value of the difference between the single capability assessment value and the big data analysis capability assessment value; The absolute value is divided by the single ability assessment value to obtain a corresponding big data indicator analysis index, wherein the big data analysis ability assessment value is used as the true result of the target person's cognitive ability, and the big data indicator analysis index is negatively correlated with the final cognitive ability assessment result.

3. The method according to claim 1, characterized in that The prediction model is constructed and trained in advance based on big data analysis technology, including: Construct a functional connectivity network based on the blood oxygen level-dependent BOLD signal in functional magnetic resonance imaging data; Extracting features from the functional connectivity network, and associating the extracted features with preset cognitive performance indicators through a machine learning algorithm to construct a prediction model based on functional connectivity; Acquire cognitive assessment data of a large number of participants through a big data platform, divide the cognitive assessment data into a training data set and a test set, and optimize the training data set through a big data optimization algorithm; Using the optimized training data set, the prediction model is trained based on the deep learning algorithm to predict the fluid intelligence quotient FIQ, crystallized intelligence quotient CIQ and cognitive comprehensive measurement IC-Cognition for each assessment test; A plurality of evaluation indicators of the calculation model are used to evaluate the prediction accuracy and generalization ability of the trained prediction model through the plurality of evaluation indicators, and the trained prediction model is corrected according to the evaluation results.

4. The method according to claim 3, characterized in that The multiple evaluation indicators include: out-of-sample test-retest reliability, and calculating the out-of-sample test-retest reliability includes: Inputting the test set into the trained prediction model for testing; Obtain two sets of feature weights of the prediction model during training and testing; The intraclass correlation coefficient ICC between the two sets of feature weights is calculated to quantify the consistency of the feature weights between the two segmentations.

5. The method according to claim 4, characterized in that The big data optimization algorithm includes: a branch and bound algorithm, and the optimization of the training data set by the big data optimization algorithm includes: The global optimal solution for each set of cognitive evaluation data is sought through the branch and bound algorithm.

6. The method according to claim 1, characterized in that The calculating of the final cognitive ability assessment result of the target person comprises: The single ability assessment value, the completion time and the big data indicator analysis index of each assessment test are multiplied by the corresponding first weight coefficient respectively to obtain a single-phase score of each assessment test, and the big data indicator analysis index is negatively correlated with the single-phase score; A comparison judgment matrix is ​​generated based on the hierarchical analysis method, the second weight coefficient of each evaluation test is calculated through the comparison judgment matrix, and the single-phase score of each evaluation test is multiplied by the corresponding second weight coefficient to obtain the final cognitive ability evaluation result.

7. The method according to claim 6, characterized in that The analytic hierarchy process is used to refine the cognitive ability assessment problem into a target layer, an evaluation index layer, and a solution layer. The second weight coefficient of each assessment test is calculated by the judgment matrix, including: Compare the evaluation indicators of each evaluation test in the evaluation indicator layer in pairs, and construct the comparison judgment matrix according to the obtained importance comparison results; The maximum eigenvalue of the comparison judgment matrix and the eigenvector matrix corresponding to the maximum eigenvalue are calculated, and the second weight coefficient of each evaluation index is determined according to the eigenvector matrix.

8. A method for restoring human cognitive ability based on assessment data, characterized in that: Executed by a computer device, the recovery method comprises the following steps: Acquire cognitive ability assessment data of a target person to be trained, and determine the classification of the cognitive ability to be restored and the corresponding degree of decline of the target person according to the cognitive ability assessment data, wherein the cognitive ability assessment data is obtained by assessing the target person using the method according to any one of claims 1 to 7; Setting a variety of cognitive recovery training, solving the causal relationship between cognitive ability assessment and cognitive recovery training based on Bayesian network, substituting the classification of cognitive ability to be recovered and the corresponding degree of decline into the causal relationship, and formulating corresponding recovery training tasks and training cycles for the target personnel; According to the recovery training task, multimodal intervention training is performed on the target person, and during the training process, the changes in the cognitive ability of the target person are continuously monitored, and the recovery plan is adjusted in real time according to the changes in the cognitive ability.

9. The method according to claim 8, characterized in that The method of solving the causal relationship between cognitive ability assessment and cognitive recovery training based on Bayesian network includes: Based on the relationship characteristics between cognitive ability assessment and cognitive recovery training, a directed acyclic graph (DAG) is constructed, wherein the structural characteristics of the directed acyclic graph represent the direct causal relationship between cognitive variables; The directed acyclic graph is converted into an equation form as shown in the following formula: X=BX+e Among them, X represents the sample matrix, B represents the coefficient matrix, e represents the disturbance matrix, and the sample matrix contains cognitive variables; The sample matrix in the equation is solved, the sample matrix is ​​decomposed into a target matrix and a perturbation matrix, and the target matrix is ​​converted into a lower triangular matrix to calculate the coefficient matrix, and the causal relationship is solved according to the calculated coefficient matrix.

10. The method according to claim 8, characterized in that The method of solving the causal relationship between cognitive ability assessment and cognitive recovery training based on the Bayesian network also includes: Based on multiple probability theory principles, the causal relationship is inferred through the Bayesian network, wherein the multiple probability theory principles include forward reasoning principle, backward reasoning principle, causal discovery principle and structural learning principle.

11. The method according to claim 8, characterized in that After the multimodal intervention training is performed on the target person, the method further includes: Re-evaluate the cognitive ability classification to be restored in the initial cognitive ability assessment, and update the scores of each assessment test that does not meet the standard in the cognitive ability classification to be restored; By comparing the initial assessment score with the updated score, the recovery effect of this round of recovery training is determined; When there are still assessment tests that do not meet the standards, the recovery training tasks are adjusted according to the recovery effect, and the next round of recovery training is carried out for the target person according to the adjusted recovery training tasks.

12. The method according to claim 8, characterized in that The cognitive ability assessment data includes the comprehensive cognitive ability assessment score of the target person and the individual scores of each assessment test. The corresponding recovery training tasks and training cycles are formulated for the target person, including: Determining whether the target person should undergo recovery training and the number of training cycles according to the score range of the comprehensive score of the cognitive ability assessment; Screen out various target assessment tests with scores below a preset threshold, determine the cognitive ability category to which each target assessment test belongs, and determine the various recovery trainings to be performed based on the causal relationship between the cognitive ability assessment and the cognitive recovery training.

13. A device for evaluating personnel cognitive ability based on big data, characterized in that: Includes the following modules: A testing module, used to collect basic data of a target person to be evaluated, and based on a plurality of preset cognitive ability evaluation categories, sequentially conduct a plurality of evaluation tests under each of the cognitive ability evaluation categories on the target person; A big data analysis module, used to obtain the test data, single capability evaluation value and completion time of each of the evaluation tests, and input the basic data and each test data into a prediction model, and calculate the big data analysis index of each evaluation test based on the big data analysis capability evaluation value output by the prediction model, wherein the prediction model is pre-trained based on the big data analysis technology, and the big data analysis index represents the error between the big data analysis capability evaluation value and the single capability evaluation value; The calculation module is used to calculate the final cognitive ability assessment result of the target person according to the single ability assessment value of each assessment test, the completion time and the big data indicator analysis index.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the method for evaluating human cognitive ability based on big data as described in any one of claims 1 to 7, or it implements the method for restoring human cognitive ability based on evaluation data as described in any one of claims 8 to 12.

Images