Method and device for identifying cognitive impairment risk

Abstract

The invention relates to the technical field of medical information, in particular to a method and device for identifying cognitive impairment risk, and the method comprises the steps: obtaining historical clinical data and historical cognitive impairment evaluation results of historical type 2 diabetes patients; respectively training a logic regression model, a support vector regression model, a random forest model and a gradient boosting decision tree model based on historical clinical data and corresponding historical cognitive impairment evaluation results to obtain corresponding first, second, third and fourth cognitive impairment recognition models; based on the cognitive impairment recognition models, a comprehensive evaluation recognition model is obtained; the method comprises the steps of obtaining target clinical data of a target type 2 diabetes patient, determining a target cognitive impairment evaluation result of the target type 2 diabetes patient based on the target clinical data and a comprehensive evaluation recognition model, and rapidly and effectively evaluating the cognitive impairment of the type 2 diabetes patient in a mode independent of psychological evaluation quantity.

Description

Technical Field

[0001] The present invention relates to the field of medical information technology, and in particular to a method and device for identifying the risk of cognitive dysfunction. Background Art

[0002] At present, psychological assessment scales are the main method for diagnosing cognitive impairment. Commonly used diagnostic assessment tools include the Mini-Mental State Examination (MMSE), the Montreal Cognitive Assessment (MoCA), the Clinical Dementia Rating Scale, and the Alzheimer's Disease Assessment Scale (ADAS-cog). However, psychological assessment scale tests are time-consuming, usually taking 3 to 12 hours, and require professional clinical staff, a quiet environment, stable emotions, and patients with a certain level of education. Therefore, its application in non-neurology departments is greatly limited, resulting in many diabetic patients with cognitive impairment not being diagnosed in time. In addition, diabetes accelerates the development of cognitive impairment. The median time from mild cognitive impairment to dementia is 5.01 years in healthy individuals and 1.83 years in patients with diabetes or prediabetes. Mild cognitive impairment usually does not affect daily life, so it is difficult to detect in time. As a result, it is often discovered when it progresses to an irreversible dementia state.

[0003] The development of cognitive impairment can be delayed by early diagnosis and appropriate intervention.

[0004] However, how to quickly and effectively assess the cognitive dysfunction in such patients is a technical problem that needs to be solved urgently. Summary of the invention

[0005] In view of the above problems, the present invention provides a method and apparatus for identifying the risk of cognitive dysfunction that overcomes the above problems or at least partially solves the above problems.

[0006] In a first aspect, the present invention provides a method for identifying a risk of cognitive dysfunction, comprising: Obtain historical clinical data of patients with type 2 diabetes and corresponding historical cognitive dysfunction assessment results; Based on the historical clinical data and the corresponding historical cognitive dysfunction assessment results, respectively training a logistic regression model, a support vector regression model, a random forest model, and a gradient boosting decision tree model to obtain a corresponding first cognitive dysfunction risk identification model, a second cognitive dysfunction identification model, a third cognitive dysfunction identification model, and a fourth cognitive dysfunction identification model; Based on the first cognitive dysfunction risk identification model, the second cognitive dysfunction identification model, the third cognitive dysfunction identification model and the fourth cognitive dysfunction identification model, a comprehensive assessment identification model is obtained; Obtain the target clinical data of the target type 2 diabetes patients; Based on the target clinical data and the comprehensive evaluation recognition model, determine the target cognitive dysfunction evaluation result of the target type 2 diabetes patients.

[0007] Preferably, obtaining the historical clinical data and the corresponding historical cognitive dysfunction evaluation results of the historical type 2 diabetes patients includes: Obtain the historical clinical data of the historical type 2 diabetes patients; Evaluate the cognitive dysfunction of the historical type 2 diabetes patients by using MMSE screening, or MMSE screening and MoCA screening, to obtain the corresponding historical cognitive dysfunction evaluation results.

[0008] Preferably, based on the historical clinical data and the corresponding historical cognitive dysfunction evaluation results, train the logistic regression model, support vector regression model, random forest model and gradient boosting decision tree model respectively to obtain the corresponding first cognitive dysfunction risk recognition model, second cognitive dysfunction recognition model, third cognitive dysfunction recognition model and fourth cognitive dysfunction recognition model, including: Based on the historical clinical data, use the cross-validation recursive feature elimination method to determine the optimal feature variable subset; Based on the optimal feature variable subset, use the 3-fold cross-validation method and the grid search method to train the logistic regression model, support vector regression model, random forest model and gradient boosting decision tree model respectively to obtain the corresponding first cognitive dysfunction risk recognition model, second cognitive dysfunction recognition model, third cognitive dysfunction recognition model and fourth cognitive dysfunction recognition model.

[0009] Preferably, based on the first cognitive dysfunction risk recognition model, second cognitive dysfunction recognition model, third cognitive dysfunction recognition model and fourth cognitive dysfunction recognition model, obtain the comprehensive evaluation recognition model, including: taking the first output result of the first cognitive dysfunction risk recognition model, the second output result of the second cognitive dysfunction risk recognition model, the third output result of the third cognitive dysfunction risk recognition model, and the fourth output result of the fourth cognitive dysfunction risk recognition model as input features, and using the corresponding cognitive dysfunction evaluation classification diagnosis result as the predicted target value, and inputting it into the multi-layer neural network model for training to obtain the comprehensive evaluation recognition model.

[0010] Preferably, after obtaining the comprehensive evaluation model based on the first cognitive dysfunction risk identification model, the second cognitive dysfunction identification model, the third cognitive dysfunction identification model, and the fourth cognitive dysfunction identification model, it further includes: constructing a loss function for the multi-layer neural network model, where the loss function is a cross-entropy loss function between the comprehensive output result of the comprehensive evaluation and identification model and the predicted target value; Based on the loss function, train and optimize the comprehensive evaluation and identification model.

[0011] Preferably, obtaining the comprehensive evaluation and identification model based on the first cognitive dysfunction risk identification model, the second cognitive dysfunction identification model, the third cognitive dysfunction identification model, and the fourth cognitive dysfunction identification model includes: based on the first output result of the first cognitive dysfunction risk identification model, the second output result of the second cognitive dysfunction risk identification model, the third output result of the third cognitive dysfunction risk identification model, and the fourth output result of the fourth cognitive dysfunction risk identification model, using the mean method or the voting method to obtain the comprehensive evaluation and identification model.

[0012] Preferably, the obtaining of the target clinical data of the target type 2 diabetes patient includes: Obtaining the age, gender, body mass index, systolic blood pressure, diastolic blood pressure, education level, diabetes duration, retinopathy condition, diabetic nephropathy condition, peripheral neuropathy condition, peripheral vascular disease condition, comorbidities, smoking status, drinking status, sleep schedule, and diet pattern of the target type 2 diabetes patient; Or, obtaining the age, gender, body mass index, systolic blood pressure, diastolic blood pressure, education level, diabetes duration, retinopathy condition, diabetic nephropathy condition, peripheral neuropathy condition, peripheral vascular disease condition, comorbidities, smoking status, drinking status, sleep schedule, diet pattern, total cholesterol, and low-density lipoprotein cholesterol of the target type 2 diabetes patient; Or, obtaining the age, gender, body mass index, systolic blood pressure, diastolic blood pressure, education level, diabetes duration, retinopathy condition, diabetic nephropathy condition, peripheral neuropathy condition, peripheral vascular disease condition, comorbidities, smoking status, drinking status, sleep schedule, diet pattern, total cholesterol, low-density lipoprotein cholesterol, human phosphorylated Tau-181 protein, and glycated hemoglobin of the target type 2 diabetes patient.

[0013] In a second aspect, the present invention also provides a device for identifying the risk of cognitive dysfunction, including: A first acquisition module, configured to acquire the historical clinical data of historical type 2 diabetes patients and the corresponding historical cognitive dysfunction evaluation results; A training module, configured to train a logistic regression model, a support vector regression model, a random forest model, and a gradient boosting decision tree model respectively based on the historical clinical data and the corresponding historical cognitive impairment assessment results, so as to obtain a corresponding first cognitive impairment risk identification model, a second cognitive impairment identification model, a third cognitive impairment identification model, and a fourth cognitive impairment identification model; An obtaining module, configured to obtain a comprehensive evaluation and identification model based on the first cognitive impairment risk identification model, the second cognitive impairment identification model, the third cognitive impairment identification model, and the fourth cognitive impairment identification model; A second obtaining module, configured to obtain the target clinical data of a target type 2 diabetes patient; A determining module, configured to determine the target cognitive impairment assessment result of the target type 2 diabetes patient based on the target clinical data and the comprehensive evaluation and identification model.

[0014] In a third aspect, the present invention further provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where when the processor executes the program, the method described in the first aspect is implemented.

[0015] In a fourth aspect, the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the method described in the first aspect is implemented.

[0016] One or more technical solutions in the embodiments of the present invention have at least the following technical effects or advantages: The present invention provides a method for identifying the risk of cognitive impairment, including: obtaining the historical clinical data and the corresponding historical cognitive impairment assessment results of historical type 2 diabetes patients; respectively training a logistic regression model, a support vector regression model, a random forest model, and a gradient boosting decision tree model based on the historical clinical data and the corresponding historical cognitive impairment assessment results to obtain a corresponding first cognitive impairment risk identification model, a second cognitive impairment risk identification model, a third cognitive impairment identification model, and a fourth cognitive impairment identification model; obtaining a comprehensive evaluation and identification model based on the first cognitive impairment risk identification model, the second cognitive impairment identification model, the third cognitive impairment identification model, and the fourth cognitive impairment identification model; obtaining the target clinical data of a target type 2 diabetes patient, and determining the target cognitive impairment assessment result of the target type 2 diabetes patient based on the target clinical data and the comprehensive evaluation and identification model, and quickly and effectively evaluating the cognitive impairment of type 2 diabetes patients through a comprehensive evaluation and identification model that does not rely on the construction of a psychological assessment scale. Description of the Drawings

[0017] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered a limitation of the present invention. Also, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings: Figure 1 It shows a schematic flow chart of the steps of the method for identifying the risk of cognitive impairment in an embodiment of the present invention; Figure 2 It shows a schematic diagram of the logical process for MoCA screening and MMSE screening in an embodiment of the present invention; Figure 3 It shows a schematic diagram of the process of obtaining a comprehensive evaluation and recognition model in an embodiment of the present invention; Figure 4 It shows a schematic diagram of the performance comparison of training various models with the first type of clinical data in an embodiment of the present invention; Figure 5 It shows a schematic diagram of the performance comparison of training various models with the second type of clinical data in an embodiment of the present invention; Figure 6 It shows a schematic diagram of the performance comparison of training various models with the third type of clinical data in an embodiment of the present invention; Figure 7 It shows a schematic diagram of the device for identifying the risk of cognitive impairment in an embodiment of the present invention; Figure 8 It shows a schematic diagram of the computer device for realizing the identification of the risk of cognitive impairment in an embodiment of the present invention. Detailed Embodiments

[0018] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0019] Embodiment 1 An embodiment of the present invention provides a method for identifying the risk of cognitive impairment, as Figure 1 shown, including: S101, obtaining the historical clinical data of historical type 2 diabetes patients and the corresponding historical cognitive impairment evaluation results; S102. Train a logistic regression model, a support vector regression model, a random forest model, and a gradient boosting decision tree model respectively based on historical clinical data and corresponding historical cognitive impairment assessment results to obtain corresponding first, second, third, and fourth cognitive impairment risk identification models; S103. Obtain a comprehensive assessment and identification model based on the first, second, third, and fourth cognitive impairment risk identification models; S104. Obtain the target clinical data of the target type 2 diabetes patient; S105. Determine the target cognitive impairment assessment result of the target type 2 diabetes patient based on the target clinical data and the comprehensive assessment and identification model.

[0020] The method for identifying the risk of cognitive impairment provided in the present invention can more quickly and accurately predict the risk of cognitive impairment in type 2 diabetes patients, thus contributing to early intervention and treatment, delaying the development of cognitive impairment, and improving the quality of life of patients.

[0021] In a specific implementation manner, according to the process of S101, collect the clinical data of historical type 2 diabetes (T2DM) patients, and incorporate relevant clinical information into the analysis. The patient inclusion criteria are as follows: (1) Age ≥ 18 years old; (2) Diagnosed with type 2 diabetes.

[0022] The exclusion criteria are as follows: (1) Patients with psychological or mental disorders; (2) Patients with a history of drug or alcohol abuse; (3) Patients who are unable to complete the psychological scale screening (such as hearing and speech disorders).

[0023] Next, for the screened patients, obtain the historical clinical data of the historical type 2 diabetes patients, including: age, gender, body mass index, human phosphorylated Tau-181 protein, glycated hemoglobin, systolic blood pressure, diastolic blood pressure, education level, diabetes duration, retinopathy, diabetic nephropathy, peripheral neuropathy, peripheral vascular disease, comorbidities, total cholesterol, low-density lipoprotein, cholesterol, smoking status, drinking status, sleep schedule, and diet regularity.

[0024] Then, evaluate the cognitive impairment of historical type 2 diabetes patients by using MMSE screening, or MMSE screening and MoCA screening to obtain corresponding historical cognitive impairment assessment results.

[0025] Specifically, the results of MoCA screening and MMSE screening are used as comprehensive criteria for diagnosing cognitive impairment. Each patient with a history of type 2 diabetes is first screened by MMSE to be classified as either having dementia or not. For patients without dementia, further MoCA screening is performed to be classified as normal or having mild cognitive impairment (MCI).

[0026] Among them, the dementia diagnosis criteria for MMSE screening scores are adjusted according to the educational level of the participants: 1. Education duration exceeds 12 years: MMSE screening score is higher than 24; 2. Education duration exceeds 6 years: MMSE screening score is higher than 20; 3. Education duration is less than 6 years: MMSE screening score is higher than 17.

[0027] A MoCA screening score between 18 and 25 indicates mild cognitive impairment; a score above 25 is considered normal. For patients with an education duration of less than 12 years, their total MoCA screening score will be adjusted and increased by 1 point to correct for educational level bias. As Figure 2 shown, it is the logical process of MoCA screening and MMSE screening.

[0028] To ensure the authenticity and validity of the data, it is necessary to filter the historical clinical data and the corresponding historical cognitive impairment assessment results.

[0029] First, data with contradictory MoCA screening results and MMSE screening results are regarded as unreliable noise data and deleted.

[0030] Then, the null values in the historical clinical data are filled. Specifically, the mean filling method is used, and data standardization and feature engineering methods are used to preprocess the historical clinical data.

[0031] Next, execute S102. Based on the historical clinical data and the corresponding historical cognitive impairment assessment results, train the logistic regression model, support vector regression model, random forest model, and gradient boosting decision tree model respectively to obtain the corresponding first cognitive impairment risk identification model, second cognitive impairment identification model, third cognitive impairment identification model, and fourth cognitive impairment identification model.

[0032] Specifically, based on historical clinical data, the cross-validation recursive feature elimination method is used to determine the optimal feature variable subset. Based on the optimal feature variable subset, the 3-fold cross-validation method and the grid search method are used to train the logistic regression model, the support vector regression model, the random forest model, and the gradient boosting decision tree model respectively, to obtain the corresponding first cognitive impairment risk identification model, second cognitive impairment identification model, third cognitive impairment identification model, and fourth cognitive impairment identification model.

[0033] When determining the optimal feature variable subset, the preprocessed historical clinical data is first used as the initial feature set. Then, a sub-model is constructed to evaluate the importance of each feature and sort them according to the degree of importance; the unimportant features are deleted to obtain a new feature set; then, based on the new feature set, the 3-fold cross-validation is used to evaluate the performance of the sub-model of the new feature set, and the above method is repeated until the feature variable data reaches 1, thereby obtaining the optimal feature subset as the training feature set of the sub-model.

[0034] Among them, the 3-fold cross-validation evaluation specifically divides the training set into 3 equal parts, takes any two of them as training data, and takes the separate one as validation data, with 3 combination methods.

[0035] Each independent sub-model, namely the logistic regression model, the support vector regression model, the random forest model, and the gradient boosting decision tree model. Logistic regression is a machine learning algorithm for classification problems. It assumes a non-linear relationship between the input and output, and uses a sigmoid function to map the input variables to the output between 0 and 1. Support vector regression is a machine learning algorithm for classification and regression problems. It is based on the idea of maximizing the margin of the classifier to find a hyperplane to separate different classes. The random forest model is an ensemble learning algorithm that combines multiple decision trees to reduce the overfitting risk of a single decision tree. The random forest algorithm can be used for classification and regression problems. The gradient boosting decision tree model is a tree-structured machine learning algorithm that divides the data set into multiple small subsets until each subset can be described by a simple rule and can be used for classification and regression problems.

[0036] The above-obtained optimal feature subset is used to train the above four sub-models respectively. Among them, the 3-fold cross-validation method and the network search method are combined for parameter tuning to obtain their respective identification models, namely the first cognitive impairment risk identification model, the second cognitive impairment risk identification model, the third cognitive impairment risk identification model, and the fourth cognitive impairment risk identification model.

[0037] Specifically, the optimal feature subset and the corresponding historical cognitive impairment assessment classification diagnosis results are used as prediction target values and input into a logistic regression model, a support vector regression model, a random forest model, and a gradient boosting decision tree model for training respectively. Through continuous iterative optimization, the corresponding recognition models are obtained.

[0038] Next, execute S103 to obtain a comprehensive evaluation recognition model based on the first cognitive impairment risk recognition model, the second cognitive impairment risk recognition model, the third cognitive impairment recognition module, and the fourth cognitive impairment recognition model.

[0039] Specifically, the first output result of the first cognitive impairment risk recognition model, the second output result of the second cognitive impairment risk recognition model, the third output result of the third cognitive impairment risk recognition model, and the fourth output result of the fourth cognitive impairment risk recognition model are used as input features, and the corresponding cognitive impairment assessment classification diagnosis result is used as the prediction target value and input into a multi-layer neural network model for training to obtain a comprehensive evaluation recognition model.

[0040] The preset target value, that is, the target value Y preset based on medical diagnosis classification label is normal: 0, dementia or mild cognitive impairment: 1.0. Regarding this as the fitting target, the output results of the above four models are used as input data and input into a multi-layer neural network model (Multi-layer Neural_Network Perceptron regressor) for training, thereby obtaining a comprehensive evaluation recognition model.

[0041] Among them, the hidden layer structure of the multi-layer neural network model is [3, 5, 8, 8, 8, 5, 3], and each number in this array represents the number of nodes in each hidden layer respectively.

[0042] If the output results of the four independent sub-models are y1, y2, y3, and y4 respectively, then the prediction result of the comprehensive evaluation recognition model is Y ensemble and Y ensemble is a continuous value between 0 and 1. The function of the multi-layer neural network model can be expressed by the following function: Y ensemble = F MLN (y1, y2, y3, y4) Furthermore, construct the loss function of the multi-layer neural network model. This loss function is the cross-entropy loss function between the comprehensive output result of the comprehensive evaluation recognition model and the prediction target value; Based on this loss function, the comprehensive evaluation recognition model is trained and optimized.

[0043] Specifically, in the process of continuous iterative optimization, ensuring that the loss function is small enough, the comprehensive evaluation recognition model reaches the optimal, as Figure 3 shown.

[0044] Since in a machine learning model, a single model often has limitations, integrating multiple single models into one model can make up for these limitations by leveraging the advantages of multiple models, thereby improving the accuracy and robustness of prediction.

[0045] The above is a method for obtaining a comprehensive evaluation recognition model through an integration approach. The comprehensive evaluation recognition model can also be obtained through the mean method or the voting method.

[0046] Specifically, for the mean method, the output results of multiple single models are added and then averaged as the output result of the comprehensive evaluation recognition model.

[0047] For the voting method, votes are cast on the output results of single models, and the model with the most votes is used as the comprehensive evaluation recognition model.

[0048] Therefore, based on the first output result of the first cognitive dysfunction risk recognition model, the second output result of the second cognitive dysfunction risk recognition model, the third output result of the third cognitive dysfunction risk recognition model, and the fourth output result of the fourth cognitive dysfunction risk recognition model, the comprehensive evaluation recognition model is obtained by using the mean method or the voting method.

[0049] Finally, S104 - S105 are executed to obtain the target clinical data of the target type 2 diabetes patient. Then, based on the target clinical data and the comprehensive evaluation recognition model, the target cognitive dysfunction evaluation result of the target type 2 diabetes patient is determined.

[0050] In a specific implementation manner, there are three situations for the obtained target clinical data: The first is that it includes not only the basic information of the target type 2 diabetes patient but also some blood test data, and the blood test data includes biomarker information.

[0051] For example: The target clinical data includes: Age, gender, body mass index, systolic blood pressure, diastolic blood pressure, education level, diabetes duration, retinopathy status, diabetic nephropathy status, peripheral neuropathy status, peripheral vascular disease status, comorbidities, smoking status, drinking status, sleep schedule, diet regularity, total cholesterol, low-density lipoprotein cholesterol, human phosphorylated Tau-181 protein, and glycated hemoglobin, where total cholesterol, low-density lipoprotein cholesterol, human phosphorylated Tau-181 protein, and glycated hemoglobin are blood test data.

[0052] The second type only includes the basic information of the target type 2 diabetes patients and some blood test data.

[0053] For example, the target clinical data includes: Age, gender, body mass index, systolic blood pressure, diastolic blood pressure, education level, diabetes duration, retinopathy status, diabetic nephropathy status, peripheral neuropathy status, peripheral vascular disease status, comorbidities, smoking status, drinking status, sleep schedule status, diet regularity status, total cholesterol, and low-density lipoprotein cholesterol. Total cholesterol and low-density lipoprotein cholesterol are part of the blood test data.

[0054] The third type only includes the basic information of the target type 2 diabetes patients.

[0055] For example, the target clinical data includes: Age, gender, body mass index, systolic blood pressure, diastolic blood pressure, education level, diabetes duration, retinopathy status, diabetic nephropathy status, peripheral neuropathy status, peripheral vascular disease status, comorbidities, smoking status, drinking status, sleep schedule status, and diet regularity status.

[0056] For the latter two types of target clinical data, in the case of insufficient medical conditions, blood test data can be dispensed with, and still accurate prediction results of the risk of cognitive impairment in type 2 diabetes patients can be obtained.

[0057] The ROC curves of the recognition models obtained by training the above three different types of clinical data using an Ensemble, a logistic regression (LR) module, a Support Vector Regression (SVR) model, a Random Forest Classifier (RF), and a LightGBM (Light Gradient Boosting Machine) are used to evaluate their performance by comparing the areas under their curves (AUC). The ROC (Receiver Operating Characteristic) curve, that is, the Receiver Operating Characteristic curve, is a tool for evaluating the performance of binary classification models. It shows the classification performance of the model by plotting the curve (ROC) of the True Positive Rate (TPR) against the False Positive Rate (FPR) at different thresholds. The closer the curve is to the upper left corner, the better the performance of the model. The area under the curve (AUC, Area Under the Curve) is an important indicator for measuring the performance of the model. The value of AUC ranges from 0 to 1, and the closer the value is to 1, the better the classification performance of the model. Specifically: The True Positive Rate (TPR), also known as Sensitivity, is the ratio of the number of samples correctly identified as positive by the model to the total number of actual positive samples. TPR = TP / (TP + FN), where TP (True Positive) is the number of samples that are actually positive and predicted to be positive, and FN (False Negative) is the number of samples that are actually positive but predicted to be negative; The False Positive Rate (FPR) is the ratio of the number of samples misidentified as positive by the model to the total number of actual negative samples. FPR = FP / (FP + TN), where FP (False Positive) is the number of samples that are actually negative but predicted to be positive, and TN (True Negative) is the number of samples that are actually negative and predicted to be negative. For the first type, the obtained AUC value is 0.839, as Figure 4 shown; for the second type, the obtained AUC value is 0.791, as Figure 5 shown; for the third type, the obtained AUC value is 0.770, as Figure 6 shown.

[0058] The risk prediction model for cognitive impairment constructed in the present invention can use this method to explore new risk factors, treatment methods, and prevention strategies for cognitive impairment, thereby promoting the progress of medical research. The model construction method and the constructed model in the present invention can be used as auxiliary means for diagnosis and treatment to improve the timeliness and accuracy of cognitive impairment diagnosis. Combining early intervention measures can effectively delay the development of cognitive impairment. For primary hospitals, outpatients, and other patients who are difficult to undergo blood or imaging examinations, the prediction model in the present invention that requires less or no such examination information can play a key role in early detection of cognitive impairment. Specifically, model performances with AUC values of 0.791 and 0.770 can be obtained by using little or no blood test information. This further verifies the excellent performance of the risk prediction model for cognitive impairment and its construction method provided by the present invention, which can assist in the diagnosis and treatment of clinical cognitive impairment.

[0059] One or more technical solutions in the embodiments of the present invention have at least the following technical effects or advantages: The present invention provides a method for identifying the risk of cognitive impairment, including: obtaining the historical clinical data of historical type 2 diabetes patients and the corresponding historical cognitive impairment assessment results; based on the historical clinical data and the corresponding historical cognitive impairment assessment results, training a logistic regression model, a support vector regression model, a random forest model, and a gradient boosting decision tree model respectively to obtain the corresponding first cognitive impairment risk identification model, second cognitive impairment risk identification model, third cognitive impairment identification model, and fourth cognitive impairment identification model; based on the first cognitive impairment risk identification model, second cognitive impairment identification model, third cognitive impairment identification model, and fourth cognitive impairment identification model, obtaining a comprehensive evaluation and identification model; obtaining the target clinical data of a target type 2 diabetes patient, and based on the target clinical data and the comprehensive evaluation and identification model, determining the target cognitive impairment assessment result of the target type 2 diabetes patient, and rapidly and effectively evaluating the cognitive impairment of type 2 diabetes patients through a comprehensive evaluation and identification model constructed without relying on a psychological assessment scale.

[0060] Embodiment 2 Based on the same inventive concept, the embodiments of the present invention also provide a device for identifying the risk of cognitive impairment, as Figure 7 shown, including: A first acquisition module 701, configured to acquire the historical clinical data of historical type 2 diabetes patients and the corresponding historical cognitive impairment assessment results; A training module 702, configured to train a logistic regression model, a support vector regression model, a random forest model, and a gradient boosting decision tree model respectively based on the historical clinical data and the corresponding historical cognitive impairment assessment results, so as to obtain corresponding first, second, third, and fourth cognitive impairment risk identification models; An obtaining module 703, configured to obtain a comprehensive evaluation and identification model based on the first, second, third, and fourth cognitive impairment risk identification models; A second obtaining module 704, configured to obtain the target clinical data of a target type 2 diabetic patient; A determining module 705, configured to determine the target cognitive impairment assessment result of the target type 2 diabetic patient based on the target clinical data and the comprehensive evaluation and identification model.

[0061] In an alternative embodiment, the first obtaining module 701 is configured to: Obtain the historical clinical data of historical type 2 diabetic patients; Evaluate the cognitive impairment of the historical type 2 diabetic patients by using MMSE screening, or MMSE screening and MoCA screening, so as to obtain corresponding historical cognitive impairment assessment results.

[0062] In an alternative embodiment, the training module 702 is configured to: Based on the historical clinical data, determine an optimal feature variable subset by using cross-validation recursive feature elimination; Based on the optimal feature variable subset, use a 3-fold cross-validation method and a grid search method to train a logistic regression model, a support vector regression model, a random forest model, and a gradient boosting decision tree model respectively, so as to obtain corresponding first, second, third, and fourth cognitive impairment risk identification models.

[0063] In an alternative embodiment, the obtaining module 703 is configured to: Use the first output result of the first cognitive impairment risk identification model, the second output result of the second cognitive impairment risk identification model, the third output result of the third cognitive impairment risk identification model, and the fourth output result of the fourth cognitive impairment risk identification model as input features, and use the corresponding cognitive impairment assessment classification diagnosis result as the predicted target value, and input them into a multi-layer neural network model for training to obtain a comprehensive evaluation and identification model.

[0064] In an alternative embodiment, the obtaining module 703 is further configured to: Construct a loss function for the multi-layer neural network model, where the loss function is a cross-entropy loss function between the comprehensive output result of the comprehensive evaluation and recognition model and the predicted target value; Based on the loss function, perform training and optimization on the comprehensive evaluation and recognition model.

[0065] In an alternative embodiment, the obtaining module 703 is further configured to: Based on the first output result of the first cognitive dysfunction risk recognition model, the second output result of the second cognitive dysfunction risk recognition model, the third output result of the third cognitive dysfunction risk recognition model, and the fourth output result of the fourth cognitive dysfunction risk recognition model, use the mean method or the voting method to obtain a comprehensive evaluation and recognition model.

[0066] In an alternative embodiment, the second obtaining module 704 is configured to: Obtain the age, gender, body mass index, systolic blood pressure, diastolic blood pressure, education level, diabetes duration, retinopathy, diabetic nephropathy, peripheral neuropathy, peripheral vascular disease, comorbidities, smoking status, alcohol consumption status, sleep schedule, and diet pattern of a target type 2 diabetes patient; Or, obtain the age, gender, body mass index, systolic blood pressure, diastolic blood pressure, education level, diabetes duration, retinopathy, diabetic nephropathy, peripheral neuropathy, peripheral vascular disease, comorbidities, smoking status, alcohol consumption status, sleep schedule, diet pattern, total cholesterol, and low-density lipoprotein cholesterol of a target type 2 diabetes patient; Or, obtain the age, gender, body mass index, systolic blood pressure, diastolic blood pressure, education level, diabetes duration, retinopathy, diabetic nephropathy, peripheral neuropathy, peripheral vascular disease, comorbidities, smoking status, alcohol consumption status, sleep schedule, diet pattern, total cholesterol, low-density lipoprotein cholesterol, human phosphorylated Tau-181 protein, and glycated hemoglobin of a target type 2 diabetes patient.

[0067] Embodiment III Based on the same inventive concept, an embodiment of the present invention provides a computer device, as Figure 8 shown, including a memory 804, a processor 802, and a computer program stored on the memory 804 and executable on the processor 802. When the processor 802 executes the program, it implements the steps of the above method for identifying the risk of cognitive dysfunction.

[0068] Among them, in Figure 8Among them, there is a bus architecture (represented by bus 800). Bus 800 may include any number of interconnected buses and bridges. Bus 800 links together various circuits of one or more processors represented by processor 802 and a memory represented by memory 804. Bus 800 may also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and thus will not be further described herein. Bus interface 806 provides an interface between bus 800 and receiver 801 and transmitter 803. Receiver 801 and transmitter 803 may be the same element, i.e., a transceiver, which provides a unit for communicating with various other devices over a transmission medium. Processor 802 is responsible for managing bus 800 and general processing, while memory 804 may be used to store data used by processor 802 when performing operations.

[0069] Embodiment 4 Based on the same inventive concept, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the steps of the method for identifying the risk of cognitive impairment as described above.

[0070] The algorithms and displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems may also be used in conjunction with the teachings provided herein. The structure required to construct such systems will be apparent from the above description. In addition, the present invention is not directed to any particular programming language. It should be understood that the content of the present invention described herein can be implemented using various programming languages, and the description of the specific language above is for disclosing the best mode of the present invention.

[0071] In the specification provided herein, a large number of specific details are set forth. However, it can be understood that the embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures, and technologies have not been shown in detail so as not to obscure the understanding of this specification.

[0072] Similarly, it should be understood that, in order to streamline this disclosure and assist in understanding one or more of the various inventive aspects, in the above description of the exemplary embodiments of the present invention, the various features of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, the disclosed method should not be construed as reflecting an intention that the claimed invention requires more features than are expressly recited in each embodiment. Rather, as reflected in each embodiment, the inventive aspects lie in less than all the features of the single embodiment disclosed above. Thus, the claims following the detailed description are hereby expressly incorporated into the detailed description, where each claim itself serves as a separate embodiment of the present invention.

[0073] Those skilled in the art can understand that the modules in the devices in the embodiments can be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments can be combined into one module or unit or component, and in addition, they can be divided into multiple sub-modules or sub-units or sub-components. Except that at least some of such features and / or processes or units are mutually exclusive, any combination can be adopted to combine all the features disclosed in this specification (including the accompanying claims, abstract and drawings) and all the processes or units of any method or device so disclosed. Unless otherwise explicitly stated, each feature disclosed in this specification (including the accompanying claims, abstract and drawings) can be replaced by an alternative feature that provides the same, equivalent or similar purpose.

[0074] In addition, those skilled in the art can understand that although some of the embodiments herein include certain features included in other embodiments rather than other features, the combination of the features of different embodiments means that it is within the scope of the present invention and forms different embodiments. For example, in the specific implementation manner, any one of the claimed embodiments can be used in any combination.

[0075] Each component embodiment of the present invention can be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art should understand that a microprocessor or a digital signal processor (DSP) can be used in practice to implement some or all of the functions of some or all of the components of the device and computer equipment for identifying the risk of cognitive impairment according to the embodiments of the present invention. The present invention can also be implemented as a device or device program (such as a computer program and a computer program product) for executing part or all of the methods described herein. Such a program implementing the present invention can be stored on a computer-readable medium, or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.

[0076] It should be noted that the above embodiments are illustrative of the present invention and not restrictive thereof, and alternative embodiments can be designed by those skilled in the art without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in the claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention can be implemented by means of hardware including several different elements and by means of a suitably programmed computer. In a unit claim listing several devices, several of these devices can be embodied by the same item of hardware. The use of the words first, second, and third, etc. does not denote any order. These words can be interpreted as names.

Claims

1. A method for identifying the risk of cognitive dysfunction, characterized in that, Including: Obtaining the historical clinical data of historical type 2 diabetes patients and the corresponding historical cognitive impairment assessment results; Based on the historical clinical data and the corresponding historical cognitive impairment assessment results, training a logistic regression model, a support vector regression model, a random forest model, and a gradient boosting decision tree model respectively to obtain the corresponding first cognitive impairment risk identification model, second cognitive impairment identification model, third cognitive impairment identification model, and fourth cognitive impairment identification model; Based on the first cognitive impairment risk identification model, the second cognitive impairment identification model, the third cognitive impairment identification model, and the fourth cognitive impairment identification model, obtaining a comprehensive evaluation and identification model; Obtaining the target clinical data of a target type 2 diabetes patient; Based on the target clinical data and the comprehensive evaluation and identification model, determining the target cognitive impairment assessment result of the target type 2 diabetes patient.

2. The method according to claim 1, wherein The obtaining of the historical clinical data of historical type 2 diabetes patients and the corresponding historical cognitive impairment assessment results includes: Obtaining the historical clinical data of historical type 2 diabetes patients; By using MMSE screening, or MMSE screening and MoCA screening, evaluating the cognitive impairment of the historical type 2 diabetes patients to obtain the corresponding historical cognitive impairment assessment results.

3. The method according to claim 1, characterized in that, Based on the historical clinical data and the corresponding historical cognitive impairment assessment results, training a logistic regression model, a support vector regression model, a random forest model, and a gradient boosting decision tree model respectively to obtain the corresponding first cognitive impairment risk identification model, second cognitive impairment identification model, third cognitive impairment identification model, and fourth cognitive impairment identification model, including: Based on the historical clinical data, using cross-validation recursive feature elimination method to determine the optimal feature variable subset; Based on the optimal feature variable subset, using 3-fold cross-validation method and grid search method to train a logistic regression model, a support vector regression model, a random forest model, and a gradient boosting decision tree model respectively to obtain the corresponding first cognitive impairment risk identification model, second cognitive impairment identification model, third cognitive impairment identification model, and fourth cognitive impairment identification model.

4. The method according to claim 1, characterized in that, Based on the first cognitive impairment risk identification model, the second cognitive impairment identification model, the third cognitive impairment identification model, and the fourth cognitive impairment identification model, obtaining a comprehensive evaluation and identification model, including: Taking the first output result of the first cognitive impairment risk identification model, the second output result of the second cognitive impairment risk identification model, the third output result of the third cognitive impairment risk identification model, and the fourth output result of the fourth cognitive impairment risk identification model as input features, and using the corresponding cognitive impairment assessment classification diagnosis result as the predicted target value, inputting them into a multi-layer neural network model for training to obtain a comprehensive evaluation and identification model.

5. The method according to claim 4, characterized in that, After obtaining the comprehensive evaluation and recognition model based on the first cognitive dysfunction risk recognition model, the second cognitive dysfunction recognition model, the third cognitive dysfunction recognition model, and the fourth cognitive dysfunction recognition model, it further includes: Construct a loss function for the multi-layer neural network model, where the loss function is the cross-entropy loss function between the comprehensive output result of the comprehensive evaluation and recognition model and the predicted target value; Based on the loss function, train and optimize the comprehensive evaluation and recognition model.

6. The method according to claim 1, characterized in that, Obtaining the comprehensive evaluation and recognition model based on the first cognitive dysfunction risk recognition model, the second cognitive dysfunction recognition model, the third cognitive dysfunction recognition model, and the fourth cognitive dysfunction recognition model includes: Based on the first output result of the first cognitive dysfunction risk recognition model, the second output result of the second cognitive dysfunction risk recognition model, the third output result of the third cognitive dysfunction risk recognition model, and the fourth output result of the fourth cognitive dysfunction risk recognition model, use the mean method or the voting method to obtain the comprehensive evaluation and recognition model.

7. The method according to claim 1, characterized in that, The obtaining of the target clinical data of the target type 2 diabetic patients includes: Obtaining the age, gender, body mass index, systolic blood pressure, diastolic blood pressure, education level, diabetes duration, retinopathy status, diabetic nephropathy status, peripheral neuropathy status, peripheral vascular disease status, comorbidities, smoking status, alcohol consumption status, sleep schedule status, and diet regularity status of the target type 2 diabetic patients; Or, obtaining the age, gender, body mass index, systolic blood pressure, diastolic blood pressure, education level, diabetes duration, retinopathy status, diabetic nephropathy status, peripheral neuropathy status, peripheral vascular disease status, comorbidities, smoking status, alcohol consumption status, sleep schedule status, diet regularity status, total cholesterol, and low-density lipoprotein cholesterol of the target type 2 diabetic patients; Or, obtaining the age, gender, body mass index, systolic blood pressure, diastolic blood pressure, education level, diabetes duration, retinopathy status, diabetic nephropathy status, peripheral neuropathy status, peripheral vascular disease status, comorbidities, smoking status, alcohol consumption status, sleep schedule status, diet regularity status, total cholesterol, low-density lipoprotein cholesterol, human phosphorylated Tau-181 protein, and glycated hemoglobin of the target type 2 diabetic patients.

8. A device for identifying the risk of cognitive dysfunction, characterized in that, It includes: A first acquisition module for acquiring the historical clinical data of historical type 2 diabetic patients and the corresponding historical cognitive dysfunction evaluation results; A training module for training a logistic regression model, a support vector regression model, a random forest model, and a gradient boosting decision tree model respectively based on the historical clinical data and the corresponding historical cognitive dysfunction evaluation results to obtain the corresponding first cognitive dysfunction risk recognition model, second cognitive dysfunction recognition model, third cognitive dysfunction recognition model, and fourth cognitive dysfunction recognition model; An obtaining module, configured to obtain a comprehensive evaluation and recognition model based on a first cognitive dysfunction risk recognition model, a second cognitive dysfunction recognition model, a third cognitive dysfunction recognition model, and a fourth cognitive dysfunction recognition model; A second acquisition module, configured to acquire target clinical data of a target type 2 diabetes patient; A determination module, configured to determine a target cognitive dysfunction evaluation result of the target type 2 diabetes patient based on the target clinical data and the comprehensive evaluation and recognition model.

9. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, the method according to any one of claims 1 to 7 is implemented.

Images