Method and apparatus for constructing post-stroke cognitive impairment prediction model, device, and medium

Abstract

The present application relates to the technical field of intelligent healthcare, and discloses a method and apparatus for constructing a post-stroke cognitive impairment prediction model, a device, and a medium. The method comprises: acquiring a historical data set of a target cohort, performing a significance test on each parameter within the historical data set by using a statistical method, and selecting a parameter meeting a set condition, the parameter meeting the set condition and labeled data forming a training set, and the labeled data being a probability of occurrence of acute post-stroke cognitive impairment or a probability of occurrence of early-onset post-stroke cognitive impairment; dividing the training set into a first training subset and a second training subset according to the labeled data; and using a machine learning method to construct an acute post-stroke cognitive impairment prediction model according to the first training subset, and using the machine learning method to construct an early-onset post-stroke cognitive impairment prediction model according to the second training subset. The present application improves accuracy of post-stroke cognitive impairment prediction.

Description

Post-stroke cognitive impairment prediction model construction method, device, equipment and medium

[0001] The application claims priority to the Chinese patent application No. 202510276359.X, filed on March 10, 2025, and entitled "Post-stroke cognitive impairment prediction model construction method, device, equipment and medium", the whole content of which is incorporated herein by reference. TECHNICAL FIELD

[0002] The application relates to the technical field of intelligent medical treatment, in particular to a post-stroke cognitive impairment prediction model construction method, device, equipment and medium. BACKGROUND

[0003] Post-stroke cognitive impairment (PSCI) is often accompanied by memory loss and decreased computing power. In some patients, the problem can be more serious, with decreased executive function, which can affect the quality of life of the patient. In the related art, the diagnosis of PSCI mainly uses a screening scale. The patient is required to perform a series of demonstrations, and each function is scored. In combination with the clinical diagnosis result of stroke, the cognitive impairment development of the patient is comprehensively evaluated. There is a problem of human subjective factors affecting the accuracy of judgment. At present, there is a lack of effective objective prediction model, which cannot realize the early warning of PSCI in the acute stage and long-term, and limits the development of early intervention and individualized treatment strategies. SUMMARY

[0004] The application aims to provide a post-stroke cognitive impairment prediction model construction method, device, equipment and medium, which can improve the accuracy of post-stroke cognitive impairment prediction.

[0005] To achieve the above-mentioned purpose, the application provides the following solutions.

[0006] In a first aspect, the application provides a post-stroke cognitive impairment prediction model construction method, comprising:

[0007] Collecting a historical data set of a target population, each sample in the historical data set comprising a plurality of parameters of a stroke patient;

[0008] Performing importance test on each parameter in the historical data set by a statistical method, and screening parameters meeting a set condition; the parameters meeting the set condition and label data constitute a training set, the label data being a probability of cognitive impairment after acute stroke or a probability of early-onset cognitive impairment after stroke; the cognitive impairment after acute stroke being cognitive impairment occurring within 2 weeks after stroke, and the early-onset cognitive impairment after stroke being cognitive impairment occurring within 3 to 6 months after stroke;

[0009] dividing the training set into a first training subset and a second training subset according to the label data; the first training subset is composed of samples with label data of the probability of occurrence of cognitive impairment after acute stroke, and the second training subset is composed of samples with label data of the probability of occurrence of cognitive impairment after early-onset stroke;

[0010] constructing a cognitive impairment after acute stroke prediction model according to the first training subset by using a machine learning method, and constructing a cognitive impairment after early-onset stroke prediction model according to the second training subset by using a machine learning method.

[0011] In a second aspect, the present application provides a cognitive impairment after stroke prediction model construction device, which applies the cognitive impairment after stroke prediction model construction method, and comprises:

[0012] a historical data set collection module, configured to collect a historical data set of a target population, each sample in the historical data set comprising a plurality of parameters of a stroke patient;

[0013] a training set determination module, configured to perform significance test on each parameter in the historical data set by using a statistical method, and screen out parameters meeting a set condition; the parameters meeting the set condition and label data constitute a training set, the label data being the probability of occurrence of cognitive impairment after acute stroke or the probability of occurrence of cognitive impairment after early-onset stroke; the cognitive impairment after acute stroke is cognitive impairment occurring within 2 weeks after stroke, and the cognitive impairment after early-onset stroke is cognitive impairment occurring within 3 to 6 months after stroke;

[0014] a training subset determination module, configured to divide the training set into a first training subset and a second training subset according to the label data; the first training subset is composed of samples with label data of the probability of occurrence of cognitive impairment after acute stroke, and the second training subset is composed of samples with label data of the probability of occurrence of cognitive impairment after early-onset stroke;

[0015] a prediction model construction module, configured to construct a cognitive impairment after acute stroke prediction model according to the first training subset by using a machine learning method, and construct a cognitive impairment after early-onset stroke prediction model according to the second training subset by using a machine learning method.

[0016] In a third aspect, the present application provides a cognitive impairment after stroke prediction model construction device, which comprises a first processor, a second processor, a third processor and a fourth processor.

[0017] A first processor is configured to retrieve a historical dataset of the target population from a memory, wherein each sample in the historical dataset includes multiple parameters of a stroke patient;

[0018] The second processor is used to perform importance tests on each parameter in the historical dataset obtained from the first processor using statistical methods, and to filter out parameters that meet the set conditions. The parameters that meet the set conditions and the label data constitute a training set, wherein the label data is the probability of cognitive impairment after acute stroke or the probability of cognitive impairment after early-onset stroke. Acute stroke cognitive impairment is defined as cognitive impairment occurring within 2 weeks after stroke, and early-onset stroke cognitive impairment is defined as cognitive impairment occurring within 3 to 6 months after stroke.

[0019] The third processor is used to divide the training set in the second processor into a first training subset and a second training subset based on the label data in the second processor; the first training subset consists of samples whose label data is the probability of cognitive impairment after acute stroke, and the second training subset consists of samples whose label data is the probability of cognitive impairment after early-onset stroke.

[0020] The fourth processor is used to construct a predictive model of cognitive impairment after acute stroke using machine learning methods based on the first training subset of the third processor, and to construct a predictive model of cognitive impairment after early stroke using machine learning methods based on the second training subset of the third processor.

[0021] Fourthly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for constructing a prediction model for post-stroke cognitive impairment as described above.

[0022] Fifthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for constructing a prediction model for post-stroke cognitive impairment as described above.

[0023] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides a method, apparatus, device, and medium for constructing a prediction model for post-stroke cognitive impairment. Parameters meeting set conditions are obtained through importance screening. A training set is constructed from these parameters and labeled data. The labeled data includes the probability of acute PSCI and the probability of early-onset PSCI. This data is generated independently of patient-generated demonstrations, reducing prediction errors caused by the uncertainty of patient demonstration behavior and improving the accuracy of PSCI prediction. Furthermore, this application creates separate prediction models for acute early-onset PSCI, further improving the accuracy of PSCI prediction. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 is a flowchart illustrating a method for constructing a predictive model for post-stroke cognitive impairment according to an embodiment of this application.

[0026] Figure 2 is a schematic diagram of the importance assignment of model variables based on the random forest algorithm provided in an embodiment of this application.

[0027] Figure 3 is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0028] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0029] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0030] This application provides a method for constructing a predictive model of post-stroke cognitive impairment, as shown in Figure 1. The method for constructing a predictive model of post-stroke cognitive impairment includes steps 101-104.

[0031] Step 101: Collect historical datasets of the target population, where each sample in the historical dataset includes multiple parameters of a stroke patient.

[0032] Step 102: Use statistical methods to perform importance tests on each parameter in the historical dataset and screen out parameters that meet the set conditions; the parameters that meet the set conditions and the label data constitute the training set, wherein the label data is the probability of cognitive impairment after acute stroke or the probability of cognitive impairment after early-onset stroke; the cognitive impairment after acute stroke is defined as cognitive impairment occurring within 2 weeks after stroke, and the cognitive impairment after early-onset stroke is defined as cognitive impairment occurring within 3 to 6 months after stroke.

[0033] Step 103: Divide the training set into a first training subset and a second training subset according to the labeled data; the first training subset consists of samples labeled with the probability of cognitive impairment after acute stroke, and the second training subset consists of samples labeled with the probability of cognitive impairment after early-onset stroke.

[0034] Step 104: Construct a predictive model for cognitive impairment after acute stroke using machine learning methods based on the first training subset, and construct a predictive model for cognitive impairment after early-onset stroke using machine learning methods based on the second training subset.

[0035] In an exemplary embodiment, step 101 specifically includes: the target population is 243 stroke patients counted from December 2022 to April 2024, and cognitive assessment and structural imaging are completed for the 243 stroke patients 14 days after admission and 3-6 months after onset to obtain historical datasets.

[0036] Acute PSCI definition: Within 2 weeks after stroke, a Mini-Mental State Examination (MMSE) or Montreal Cognitive Assessment (MoCA) score below the threshold for the elderly population in a designated region.

[0037] Early-onset PSCI definition: 3-6 months after stroke, with an MMSE or MoCA score below the threshold for the elderly population in the designated region.

[0038] In one exemplary embodiment, step 102 specifically includes:

[0039] Statistical methods were used to perform t-tests, Mann-Whitney U tests, chi-square tests, or Fisher tests on each parameter in the historical dataset to obtain the p-value for each parameter. Parameters with p-values ​​less than a set value were used as parameters after p-value screening. More specifically, Python was used for statistical processing of the data. Normally distributed clinical variables were expressed as mean ± standard deviation, and t-tests were used for comparisons between groups. Non-normally distributed clinical variables were expressed as M(Q1, Q3), and Mann-Whitney U tests were used for comparisons between groups. Categorical variables were expressed as frequency and relative percentage (%), and chi-square tests or Fisher tests were used for comparisons between groups. M(Q1, Q3) represents the combination of median and interquartile range.

[0040] The random forest method is used to select the top N parameters by importance from the parameters after filtering by the P-value of the first training subset, thus obtaining the first set of parameters.

[0041] The random forest method is used to select the top N parameters by importance from the parameters after filtering by the P-value of the second training subset, thus obtaining the second set of parameters.

[0042] The union of the first set of parameters and the second set of parameters is taken to obtain the parameters that meet the set conditions. The parameter importance is assigned using the random forest method, as shown in Figure 2.

[0043] In one exemplary embodiment, the value is set to 0.05 and N is set to 10. Variables with P < 0.05 in the univariate analysis are included in the random forest variable importance ranking.

[0044] Of the 243 participants, 189 were male and 54 were female, with an average age of 61 (53, 70) years at enrollment. A total of 112 patients had acute post-stroke cognitive impairment (PSCI). The PSCI group had significantly higher rates of age, white matter hyperintensity, diabetes, hyperlipidemia, stroke history, lacunar infarction, small vessel disease burden, and infarction at key sites and in the cortical region compared to the non-PSNCI group, and lower rates of education compared to the PSNCI group (all P < 0.05). Three to six months post-stroke, the PSCI group had significantly higher rates of age, white matter hyperintensity, hyperlipidemia, small vessel disease imaging burden, and cortical infarction compared to the PSNCI group (all P < 0.05).

[0045] Historical datasets include imagery data, and methods for processing imagery data include:

[0046] (1) Calculation of high signal volume of brain white matter: Based on T2-FLAIR sequence, the high signal volume of brain white matter was segmented using MATLAB-SPM12-LST toolkit to obtain the high signal volume of brain white matter.

[0047] (2) Infarct volume calculation: Based on the diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) sequences, baseline segmentation was performed using the MATLAB-SPM12-LST toolkit to obtain the infarct volume.

[0048] (3) Infarction in key areas: Based on DWI and ADC sequence interpretation, determine whether there is infarction in the basal ganglia, thalamus, hippocampus, inferior medial temporal gyrus and angular gyrus.

[0049] In an exemplary embodiment, the parameters include age, years of education, volume of high signal in brain white matter, presence of diabetes, presence of stroke, presence of cortical infarction, presence of lacunar lesions, and presence of infarction in key areas; key areas include the basal ganglia, thalamus, hippocampus, inferior medial temporal gyrus, and angular gyrus.

[0050] In an exemplary embodiment, a predictive model for post-acute stroke cognitive impairment is constructed using machine learning methods based on a first training subset, specifically including:

[0051] Multiple first prediction models are constructed using different machine learning methods based on the first training subset.

[0052] Each of the first prediction models is tested using the first test set, and the area under the curve (AUC) value corresponding to each first prediction model is obtained. The data structure of the first test set is the same as that of the first training subset.

[0053] The first prediction model with the largest AUC value was used as the prediction model for cognitive impairment after acute stroke.

[0054] In one exemplary embodiment, a machine learning method is used to construct a predictive model for cognitive impairment after early-onset stroke based on a second training subset, specifically including:

[0055] Multiple second prediction models are constructed using different machine learning methods based on the second training subset.

[0056] The second test set is used to test each of the second prediction models, and the AUC value corresponding to each second prediction model is obtained. The data structure of the second test set is the same as that of the second training subset.

[0057] The second prediction model with the largest AUC value was used as the prediction model for cognitive impairment after early-onset stroke.

[0058] The parameters that meet the set conditions selected in the steps are used as the target parameters.

[0059] The predictive model for acute post-stroke cognitive impairment is used to input the patient's target parameters, especially the target parameters at admission, and output the probability of acute PSCI.

[0060] The early-onset post-stroke cognitive impairment prediction model is used to input the patient's target parameters, especially the target parameters at admission, and output the probability of early-onset PSCI.

[0061] The machine learning methods mentioned include Gradient Boost, K-Nearest Neighbors, Logistic Regression, Artificial Neural Network (ANN), Random Forest, Support Vector Machines, and XGBoost.

[0062] Among them, the AUCs of the prediction models for acute post-stroke cognitive impairment constructed by gradient boosting algorithm, K-nearest neighbor algorithm, random forest algorithm, logistic regression algorithm, artificial neural network support vector machine, and XGBoost were 0.87, 0.78, 0.96, 0.78, 0.80, 0.83, and 0.96, respectively, all exceeding 0.75. The AUCs of the prediction models for early-onset post-stroke cognitive impairment constructed by gradient boosting algorithm, K-nearest neighbor algorithm, random forest algorithm, logistic regression algorithm, artificial neural network support vector machine, and XGBoost were 0.84, 0.86, 0.85, 0.75, 0.77, 0.76, and 0.82, respectively, all exceeding 0.75.

[0063] In one exemplary embodiment, this application provides a method for predicting post-stroke cognitive impairment, comprising: inputting the target parameters to be predicted into an acute post-stroke cognitive impairment prediction model and outputting the probability of acute PSCI; and inputting the target parameters to be predicted into an early-onset post-stroke cognitive impairment prediction model and outputting the probability of early-onset PSCI.

[0064] Based on the same inventive concept, this application also provides a device for constructing a post-stroke cognitive impairment prediction model to implement the above-mentioned method for constructing a post-stroke cognitive impairment prediction model. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the post-stroke cognitive impairment prediction model construction device provided below can be found in the limitations of the post-stroke cognitive impairment prediction model construction method described above, and will not be repeated here.

[0065] In an exemplary embodiment, a device for constructing a predictive model for post-stroke cognitive impairment is provided. The device applies the aforementioned method for constructing a predictive model for post-stroke cognitive impairment. The device includes:

[0066] The historical dataset acquisition module is used to collect historical datasets of the target population. Each sample in the historical dataset includes multiple parameters of a stroke patient.

[0067] The training set determination module is used to perform importance tests on each parameter in the historical dataset using statistical methods, and to screen out parameters that meet the set conditions. The parameters that meet the set conditions and the label data constitute the training set. The label data is the probability of cognitive impairment after acute stroke or the probability of cognitive impairment after early-onset stroke. Cognitive impairment after acute stroke is defined as cognitive impairment occurring within 2 weeks after stroke, and cognitive impairment after early-onset stroke is defined as cognitive impairment occurring within 3 to 6 months after stroke.

[0068] The training subset determination module is used to divide the training set into a first training subset and a second training subset based on the labeled data; the first training subset consists of samples whose labeled data is the probability of cognitive impairment after acute stroke, and the second training subset consists of samples whose labeled data is the probability of cognitive impairment after early-onset stroke.

[0069] The prediction model building module is used to build a prediction model for cognitive impairment after acute stroke using machine learning methods based on a first training subset, and to build a prediction model for cognitive impairment after early stroke using machine learning methods based on a second training subset.

[0070] In an exemplary embodiment, this application provides a device for constructing a prediction model for post-stroke cognitive impairment. The device is a computer-implemented device for constructing a prediction model for post-stroke cognitive impairment. The device applies the method for constructing a prediction model for post-stroke cognitive impairment. The device includes a first processor, a second processor, a third processor, and a fourth processor.

[0071] A first processor is used to retrieve a historical dataset of the target population from memory, wherein each sample in the historical dataset includes multiple parameters of a stroke patient.

[0072] The second processor is used to perform importance tests on each parameter in the historical dataset obtained from the first processor using statistical methods, and to screen out parameters that meet the set conditions. The parameters that meet the set conditions and the label data constitute a training set. The label data is the probability of cognitive impairment after acute stroke or the probability of cognitive impairment after early-onset stroke. Cognitive impairment after acute stroke is defined as cognitive impairment occurring within 2 weeks after stroke, and cognitive impairment after early-onset stroke is defined as cognitive impairment occurring within 3 to 6 months after stroke.

[0073] The third processor is used to divide the training set in the second processor into a first training subset and a second training subset based on the labeled data in the second processor; the first training subset consists of samples whose labeled data is the probability of cognitive impairment after acute stroke, and the second training subset consists of samples whose labeled data is the probability of cognitive impairment after early-onset stroke.

[0074] The fourth processor is used to construct a predictive model of cognitive impairment after acute stroke using machine learning methods based on the first training subset of the third processor, and to construct a predictive model of cognitive impairment after early stroke using machine learning methods based on the second training subset of the third processor.

[0075] In an exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram is shown in Figure 3. The computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is connected to the system bus via the I / O interfaces. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device stores data for constructing a post-stroke cognitive impairment prediction model. The I / O interfaces of the computer device are used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for constructing a post-stroke cognitive impairment prediction model.

[0076] Those skilled in the art will understand that the structure shown in Figure 3 is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. A specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0077] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0078] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0079] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0080] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0081] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units, etc., and are not limited to these.

[0082] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0083] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for constructing a predictive model for post-stroke cognitive impairment, characterized in that, The method for constructing the post-stroke cognitive impairment prediction model includes: Collect historical datasets of the target population, where each sample in the historical dataset includes multiple parameters of a stroke patient; Statistical methods were used to perform importance tests on each parameter in the historical dataset, and parameters that met the set conditions were selected. The parameters that met the set conditions and the labeled data constituted the training set. The labeled data were the probability of cognitive impairment after acute stroke or the probability of cognitive impairment after early-onset stroke. Cognitive impairment after acute stroke was defined as cognitive impairment occurring within 2 weeks after stroke, and cognitive impairment after early-onset stroke was defined as cognitive impairment occurring within 3 to 6 months after stroke. The training set is divided into a first training subset and a second training subset based on the labeled data; the first training subset consists of samples labeled with the probability of cognitive impairment after acute stroke, and the second training subset consists of samples labeled with the probability of cognitive impairment after early-onset stroke. Based on the first training subset, a machine learning method was used to construct a predictive model for cognitive impairment after acute stroke, and based on the second training subset, a machine learning method was used to construct a predictive model for cognitive impairment after early-onset stroke.

2. The method for constructing a predictive model for post-stroke cognitive impairment according to claim 1, characterized in that, Statistical methods were used to perform importance tests on each parameter in the historical dataset, and parameters that met the set criteria were selected, including: Statistical methods were used to perform t-tests, Mann-Whitney U tests, chi-square tests, or Fisher tests on each parameter in the historical dataset to obtain the p-values ​​for each parameter. Parameters with p-values ​​less than a set value were selected as parameters after p-value screening. The random forest method is used to select the top N parameters by importance from the parameters after filtering by the p-value of the first historical data subset to obtain the first set of parameters; the first historical data subset is composed of samples in the historical dataset whose labeled data is the probability of cognitive impairment after acute stroke. The random forest method is used to select the top N parameters by importance from the parameters after filtering by the p-value of the second historical data subset, to obtain the second set of parameters; the second historical data subset is composed of samples in the historical dataset whose labeled data is the probability of cognitive impairment after early-onset stroke; Take the union of the first set of parameters and the second set of parameters to obtain the parameters that meet the set conditions.

3. The method for constructing a predictive model for post-stroke cognitive impairment according to claim 1, characterized in that, The parameters include age, years of education, volume of high signal in brain white matter, presence of diabetes, presence of stroke, presence of cortical infarction, presence of lacunar lesions, and presence of infarction in key areas; key areas include the basal ganglia, thalamus, hippocampus, inferior medial temporal gyrus, and angular gyrus.

4. The method for constructing a predictive model for post-stroke cognitive impairment according to claim 1, characterized in that, Based on the first training subset, a machine learning method was used to construct a predictive model for cognitive impairment after acute stroke, specifically including: Multiple first prediction models are constructed using different machine learning methods based on the first training subset; Each of the first prediction models is tested using the first test set, and the AUC value corresponding to each of the first prediction models is obtained. The first prediction model with the largest AUC value was used as the prediction model for cognitive impairment after acute stroke.

5. The method for constructing a predictive model for post-stroke cognitive impairment according to claim 1, characterized in that, Based on the second training subset, a machine learning method was used to construct a predictive model for cognitive impairment after early-onset stroke, specifically including: Multiple second prediction models are constructed using different machine learning methods based on the second training subset; The second test set was used to test each of the second prediction models, and the AUC value corresponding to each of the second prediction models was obtained. The second prediction model with the largest AUC value was used as the prediction model for cognitive impairment after early-onset stroke.

6. The method for constructing a predictive model for post-stroke cognitive impairment according to claim 4 or 5, characterized in that, The machine learning methods include gradient boosting, K-nearest neighbors, logistic regression, random forest, artificial neural networks, support vector machines, and XGBoost.

7. A device for constructing a predictive model for post-stroke cognitive impairment, characterized in that, The post-stroke cognitive impairment prediction model construction device applies the post-stroke cognitive impairment prediction model construction method according to any one of claims 1-6, and the post-stroke cognitive impairment prediction model construction device comprises: The historical dataset acquisition module is used to collect historical datasets of the target population. Each sample in the historical dataset includes multiple parameters of a stroke patient. The training set determination module is used to perform importance tests on each parameter in the historical dataset using statistical methods, and to screen out parameters that meet the set conditions. The parameters that meet the set conditions and the label data constitute the training set. The label data is the probability of cognitive impairment after acute stroke or the probability of cognitive impairment after early-onset stroke. Cognitive impairment after acute stroke is defined as cognitive impairment occurring within 2 weeks after stroke, and cognitive impairment after early-onset stroke is defined as cognitive impairment occurring within 3 to 6 months after stroke. The training subset determination module is used to divide the training set into a first training subset and a second training subset based on the label data; the first training subset consists of samples whose label data is the probability of cognitive impairment after acute stroke, and the second training subset consists of samples whose label data is the probability of cognitive impairment after early-onset stroke. The prediction model building module is used to build a prediction model for cognitive impairment after acute stroke using machine learning methods based on a first training subset, and to build a prediction model for cognitive impairment after early stroke using machine learning methods based on a second training subset.

8. A device for constructing a predictive model for post-stroke cognitive impairment, characterized in that, The device for constructing a predictive model for post-stroke cognitive impairment includes a first processor, a second processor, a third processor, and a fourth processor. A first processor is configured to retrieve a historical dataset of the target population from a memory, wherein each sample in the historical dataset includes multiple parameters of a stroke patient; The second processor is used to perform importance tests on each parameter in the historical dataset obtained from the first processor using statistical methods, and to filter out parameters that meet the set conditions. The training set consists of parameters and label data that meet the set conditions. The label data is the probability of cognitive impairment after acute stroke or the probability of cognitive impairment after early-onset stroke. Cognitive impairment after acute stroke is defined as cognitive impairment occurring within 2 weeks after stroke, and cognitive impairment after early-onset stroke is defined as cognitive impairment occurring within 3 to 6 months after stroke. The third processor is used to divide the training set in the second processor into a first training subset and a second training subset based on the label data in the second processor; the first training subset consists of samples whose label data is the probability of cognitive impairment after acute stroke, and the second training subset consists of samples whose label data is the probability of cognitive impairment after early-onset stroke. The fourth processor is used to construct a predictive model of cognitive impairment after acute stroke using machine learning methods based on the first training subset of the third processor, and to construct a predictive model of cognitive impairment after early stroke using machine learning methods based on the second training subset of the third processor.

9. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the method for constructing a predictive model of post-stroke cognitive impairment as described in any one of claims 1-6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the method for constructing a predictive model for post-stroke cognitive impairment as described in any one of claims 1-6.

Images