Portable intelligent evaluation and rehabilitation training integrated equipment for cognitive impairment patient

Abstract

The invention discloses portable intelligent assessment and rehabilitation training integrated equipment for cognitive disorder patients, and belongs to the technical field of cognitive disorder assessment. Comprising a device body composed of a base and a display screen, and the device body is electrically connected with an intelligent evaluation training system. The problem that an existing assessment method has limitation is solved, and the portability of existing medical equipment is improved by performing integrated design on the equipment in a notebook computer form; multi-dimensional evaluation data of the patient are collected through a MoCA Monte-Evaluation scale, scores and change trends of all cognitive functions are extracted by using a correlation analysis method, and the scores and change trends are predicted through a neural network model to obtain a prediction result of the patient; corresponding rehabilitation training suggestions are retrieved from the decision tree classification model, optimization is carried out in combination with actual conditions, and then rehabilitation training is carried out on the patient; and by introducing an intelligent technology, personalized design and comprehensive intervention, the evaluation and rehabilitation training effects are further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of cognitive impairment assessment, and in particular to a portable integrated device for intelligent assessment and rehabilitation training of patients with cognitive impairment. Background Art

[0002] Cognitive impairment refers to significant damage in an individual's cognitive areas such as memory, attention, language, executive function, and visual-spatial ability, which affects daily life and social function. Cognitive impairment ranges from mild cognitive impairment (MCI) to severe dementia, such as Alzheimer's disease. The assessment and rehabilitation training methods of cognitive impairment play an important role in clinical practice.

[0003] However, in the existing technology, traditional cognitive impairment assessment methods still have some limitations. For example, the assessment method relies on the patient's subjective answers and is easily affected by the patient's emotions, education level and cultural background. For example, patients with lower education levels may score lower in language and abstract thinking tests, but this does not necessarily reflect the true cognitive function, making it difficult to ensure the accuracy of the cognitive impairment assessment results.

[0004] Therefore, the existing needs are not met, and we propose a portable integrated device for intelligent assessment and rehabilitation training of patients with cognitive impairment. Summary of the Invention

[0005] The purpose of the present invention is to provide a portable integrated device for intelligent assessment and rehabilitation training of patients with cognitive impairment. By integrating the device in the form of a laptop computer, the portability of existing medical equipment is improved. The MoCA Montreal Cognitive Assessment scale is used to collect multidimensional assessment data of patients, and the scores and change trends of various cognitive functions are extracted using a correlation analysis method. These scores and change trends are predicted using a neural network model to obtain the patient's prediction results. Corresponding rehabilitation training suggestions are retrieved from a decision tree classification model, optimized based on actual conditions, and then rehabilitation training is performed on the patient. By introducing intelligent technology, personalized design and comprehensive intervention, the effects of assessment and rehabilitation training are further improved, thereby solving the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A portable integrated device for intelligent assessment and rehabilitation training for patients with cognitive impairment comprises: a device body, the device body comprising a base and a display screen, the base being provided with buttons and a touch screen, the side of the base being provided with a card slot, a sound hole, and an interface, a camera being provided at the top of the display screen, and a stylus for writing training on the touch screen being embedded in the card slot, the device body being electrically connected to an intelligent assessment and training system via wireless communication technology;

[0008] The intelligent assessment training system includes:

[0009] Intelligent assessment unit, used to evaluate patients' visual-spatial and executive abilities, abstract abilities, memory, attention, calculation abilities, language abilities, and orientation abilities using the Montreal Cognitive Assessment (MoCA) scale, obtaining multidimensional assessment data of patients;

[0010] A data collection unit is used to collect multidimensional assessment data of historical patients in the MoCA Montreal Cognitive Assessment Scale, pre-process the multidimensional assessment data of historical patients, and then save them in time series to form data samples;

[0011] The feature extraction unit is used to extract feature information from the preprocessed data samples using the correlation analysis method, including the scores and change trends of various cognitive functions, which serve as the feature set for subsequent model prediction training;

[0012] The model prediction unit is used to create a neural network model, train the neural network model based on the preprocessed data samples and the filtered feature set, enable the neural network model to learn the rules and patterns in the data samples, and output the prediction results;

[0013] The evaluation summary unit is used to analyze the prediction results and determine whether the patient has cognitive impairment and the degree of impairment of various cognitive functions;

[0014] The rehabilitation training unit is used to provide patients with personalized rehabilitation training recommendations based on the judgment results, and uses a decision tree classification model to cluster and divide the rehabilitation training recommendations according to the cause type, to assist doctors in the diagnosis and treatment of cognitive impairment.

[0015] Furthermore, the intelligent assessment training system further includes:

[0016] A user registration unit is used to register the patient's personal account in the intelligent assessment and training system through a display screen and buttons to form a patient's personal database;

[0017] The function selection unit is used to display the cognitive function assessment selection option on the display screen after the patient completes login, for the doctor or family member to select.

[0018] The human-computer interaction unit is used to enable the patient to interact with the rehabilitation training course through the sound hole and camera.

[0019] Furthermore, the feature extraction unit includes:

[0020] The data processing module is used to organize the multidimensional assessment data in the data sample into a structured table, where each row represents a time point and each column represents the score of a cognitive function;

[0021] The correlation analysis module is used to perform time series analysis on the scores of each cognitive function, calculate the mean, variance, and change trend of each cognitive function, and then use the Pearson correlation coefficient to calculate the correlation between various cognitive functions. The calculation formula is as follows:

[0022]

[0023] Where r represents the correlation between the two cognitive functions; x i and y i are the scores of two cognitive functions, and are the means of the two cognitive functions respectively;

[0024] Perform a p-value significance test on the correlation coefficient to determine whether the correlation coefficient is accurate; after confirming its accuracy, use a heat map to display the correlation coefficients and correlation strengths between various cognitive functions.

[0025] Furthermore, the feature extraction unit further includes:

[0026] The trend extraction module is used to calculate the linear change slope of each cognitive function score using the linear regression method, and perform time series analysis on the score of each cognitive function to extract the change trend of each cognitive function score;

[0027] The feature extraction module is used to cluster various cognitive functions according to the similarity of their changing trends and identify functions with similar changing patterns as feature information of the multidimensional evaluation data in the current data sample.

[0028] Furthermore, the feature extraction module includes:

[0029] A processing module is used to establish a dimensionality elimination processing degree between any two cognitive functions based on the dimensional differences between the various cognitive functions, and to unify the change trends of the cognitive functions based on the dimensionality elimination processing degree to obtain a target change trend;

[0030] A center determination module is used to divide the target change trend into high change, medium change and low change to obtain multiple trend segments, determine the silhouette coefficient of each trend segment in the high change trend segment, select the high trend segment with the highest silhouette coefficient as the high change cluster center, determine the medium trend segment corresponding to the average distance of the medium change trend segment content as the medium change cluster center, and randomly select any low trend segment in the low change trend segment as the low change cluster center;

[0031] The clustering module is used to perform cluster analysis on the target change trend based on the high change cluster center, the medium change cluster center and the low change cluster center to obtain the clustering results;

[0032] An analysis module is used to obtain, from the clustering results, trend segments to be analyzed whose similarities with the high-variation cluster center, the medium-variation cluster center, and the low-variation cluster center are within a preset range;

[0033] The judgment module is used to analyze the time characteristics of the trend segments to be analyzed in different cognitive functions and determine whether the time series are consistent;

[0034] If so, it is determined that the cognitive functions corresponding to the trend segments to be analyzed have similar change patterns;

[0035] Otherwise, it is determined that the cognitive functions corresponding to the trend segments to be analyzed do not have similar change patterns;

[0036] The information determination module is used to use a set of cognitive functions with similar change patterns as feature information of the multidimensional evaluation data in the current data sample.

[0037] Furthermore, the rehabilitation training unit includes:

[0038] The plan-making module is used to develop targeted rehabilitation training plans based on the type and severity of the patient's cognitive impairment. The rehabilitation training plans include: memory training, language ability training, attention training, visual-spatial and executive ability training, calculation training, orientation training, and comprehensive cognitive training;

[0039] The plan division module is used to use the cause type of cognitive impairment, cognitive function score and severity as characteristic variables and rehabilitation training recommendations as target variables;

[0040] The cluster analysis module is used to build a decision tree classification model using CART, and gradually divide the rehabilitation training recommendations according to the cause type and severity of the patient's cognitive impairment; and based on the division results of the decision tree classification model, the rehabilitation training recommendations are clustered by cause type.

[0041] Furthermore, the rehabilitation training unit further includes:

[0042] The program generation module is used to assist doctors in optimizing rehabilitation plans and developing personalized rehabilitation training plans for each patient based on the prediction results of the neural network model, combined with the rehabilitation training suggestions and family members' opinions given by the decision tree classification model;

[0043] The dynamic adjustment module is used to regularly evaluate the patient's cognitive function and adjust the training plan in a timely manner according to changes.

[0044] Furthermore, the data collection unit includes:

[0045] The data sorting module is used to clean the multi-dimensional evaluation data of historical patients in the collected data samples and remove outliers and missing values;

[0046] The data storage module is used to save the pre-processed data samples in time series according to the collection date, and mark the timestamp and the cause type of cognitive impairment to form a data sample library.

[0047] Furthermore, the process of removing outliers and missing values ​​in the data sorting module is as follows:

[0048] Obtaining an evaluation data set of a historical patient in a data sample in a current dimension, obtaining a data evaluation maximum value and a data evaluation minimum value in the evaluation data set, and obtaining a standard deviation of the evaluation data set in a previous dimension;

[0049] Based on the maximum and minimum data evaluation values ​​and the standard deviation, the current data evaluation value is judged as abnormal to obtain the degree of data abnormality;

[0050] The calculation formula for the degree of data anomaly is as follows:

[0051]

[0052] Among them, K represents the degree of data anomaly of the current data evaluation value, σ represents the standard deviation of the evaluation data set, and δ max Indicates the maximum value of data evaluation, δ min represents the minimum value of data evaluation, γ1 represents the maximum value range ratio, γ2 represents the minimum value range ratio, and H represents a constant with a value of 5;

[0053] When the data abnormality level is greater than a preset level, the current data evaluation value is determined to be an abnormal value and removed; otherwise, the current data evaluation value is determined to be normal;

[0054] Obtain the data type and data distribution characteristics of the evaluation data set of the current dimension, and determine whether there are missing values ​​in the evaluation data set based on the data type and data distribution characteristics;

[0055] The missing value judgment method is as follows:

[0056] and

[0057] Among them, n represents the number of data types in the evaluation data set, N i represents the amount of data of the i-th data type, N0 represents the standard data amount, τ i represents the data volume weight of the i-th data type, R represents the capacity threshold, Q represents the distribution representation value of the data distribution characteristics under the evaluation data set, Q0 represents the standard distribution representation value, and Z represents the data distribution threshold;

[0058] If there are missing values, determine the location of the missing values ​​and remove the evaluation data range where the missing values ​​are located.

[0059] Furthermore, the model prediction unit includes:

[0060] The sample partitioning module is used to divide the preprocessed data samples and the filtered feature set into a training set and a test set, train the neural network model with the training set, and verify the accuracy of the neural network model with the test set;

[0061] The performance verification module is used to compare the actual diagnostic results in the data sample with the predicted results of the neural network model to analyze whether the performance of the neural network model is accurate; if the verification passes, the neural network model is applied to actual clinical diagnosis;

[0062] The model optimization module is used to regularly update and iterate the neural network model with new multi-dimensional evaluation data to optimize the performance of the neural network model.

[0063] Compared with the prior art, the present invention has the following beneficial effects:

[0064] 1. In the present invention, by adopting the form of a laptop computer, medical equipment capable of intelligent assessment and rehabilitation training of cognitive impairment is integrated into an all-in-one device, so as to improve the portability of existing equipment.

[0065] 2. In the present invention, the Montreal Cognitive Assessment (MoCA) is used to collect multidimensional assessment data from patients. Correlation analysis is then used to extract the scores and change trends of various cognitive functions from the multidimensional assessment data. The characteristic information is then predicted using a neural network model to obtain a prediction result of whether the patient suffers from cognitive impairment and the severity of the condition. Based on the obtained prediction results, corresponding rehabilitation training recommendations are retrieved from a decision tree classification model, and the recommendations are optimized based on the patient's actual situation, thereby providing rehabilitation training for the patient. Based on this, by introducing intelligent technology, personalized design, and comprehensive intervention, doctors can be assisted in further improving the effectiveness of assessment and rehabilitation training, thereby improving the quality of life of patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 This is a structural diagram of the portable integrated device for intelligent assessment and rehabilitation training of patients with cognitive impairment according to the present invention;

[0067] Figure 2 This is a schematic diagram of the module composition of the intelligent assessment training system of the present invention.

[0068] In the figure: 1. Device body; 2. Base; 21. Button; 22. Touch screen; 3. Display; 31. Camera; 4. Card slot; 41. Stylus; 5. Sound hole; 6. Interface. DETAILED DESCRIPTION

[0069] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0070] To address the existing technical issues, traditional cognitive impairment assessment methods still have some limitations, such as: the assessment method relies on the patient's subjective answers and is easily affected by the patient's mood, education level and cultural background. For example, patients with low education levels may score lower in language and abstract thinking tests, but this does not necessarily reflect the true cognitive function, making it difficult to ensure the accuracy of cognitive impairment assessment results. Figure 1-2 , this embodiment provides the following technical solutions:

[0071] A portable integrated device for intelligent assessment and rehabilitation training of patients with cognitive impairment, comprising: a device body 1, which is composed of a base 2 and a display screen 3, and the base 2 and the display screen 3 are connected by a screen axis, and is composed of two upper and lower hinge brackets, a stretching belt and a rotating shaft. The hinge bracket is responsible for carrying the screen and keyboard and allowing them to rotate freely; the stretching belt is used to adjust the tension of the hinge to ensure the stability of the screen in various states; and the rotating shaft is the core component connecting the upper and lower hinge brackets, which not only enables the screen to rotate, but also fixes it at a specific angle; a button 21 and a touch screen 22 are provided on the base 2, the button 21 is used to log in to the patient's personal account and select the corresponding device function, and the touch screen 22 is related to the stylus 41 The patient writes the corresponding training content on the touch screen 22 using a stylus 41. For example, in the Montreal Cognitive Assessment (MoCA) scale, the visual-spatial and executive function parts require items such as an alternating line connection test, drawing a cube, and drawing a clock to assess the patient's visual-spatial and executive functions. Based on this, the assessment content can be completed on the touch screen 22 using the stylus 41. A card slot 4, a sound hole 5, and an interface 6 are provided on the side of the base 2. A camera 31 is provided on the upper end of the display screen 3. A stylus 41 for writing training on the touch screen 22 is embedded in the card slot 4. The sound hole 5 is used to send and receive sound. The interface 6 is used to connect a charging cable or other data cable. The device body 1 is electrically connected to the intelligent assessment and training system through wireless communication technology.

[0072] Intelligent assessment and training system, including:

[0073] A user registration unit is used to register a patient's personal account in the intelligent assessment and training system through the display screen 3 and the button 21 to form a patient personal database;

[0074] A function selection unit, for displaying a cognitive function assessment option on the display screen 3 via the key 21 after the patient completes login, for the doctor or family member to select;

[0075] The human-computer interaction unit is used to enable the patient to interact with the rehabilitation training course through the sound hole 5 and the camera 31; so that the patient can communicate with the teacher or other animated characters used to issue instructions in the rehabilitation training content during rehabilitation training;

[0076] The intelligent assessment unit is used to assess the patient's visual-spatial and executive abilities, abstract thinking, memory, attention, calculation, language, and orientation using the MoCA Montreal Cognitive Assessment scale to obtain multidimensional assessment data of the patient. Specifically, the MoCA Montreal Cognitive Assessment scale is suitable for screening for cognitive dysfunction and includes 11 examination items in eight cognitive areas, including attention, executive function, memory, language, visual-spatial ability, abstract thinking, calculation, and orientation. The total score is 30 points, and a score of 26 or more is considered normal. The MoCA Montreal Cognitive Assessment scale has a short testing time, high sensitivity, and covers important cognitive areas, making it suitable for clinical application. Therefore, the portable integrated intelligent assessment and rehabilitation training device for patients with cognitive impairment in this embodiment uses the MoCA Montreal Cognitive Assessment scale to assess patients.

[0077] The data collection unit is used to collect the multidimensional assessment data of historical patients in the MoCA Montreal Cognitive Assessment Scale, pre-process the multidimensional assessment data of historical patients, and then save them in time series to form data samples; the data collection unit includes:

[0078] The data cleaning module is used to clean the multidimensional assessment data of historical patients in the collected data sample and remove outliers and missing values. Specifically, when processing the MoCA Montreal Cognitive Assessment data, it sets reasonable thresholds or ranges to identify outliers and exclude them from the dataset. For example, if the score of certain assessment items is significantly outside the normal range, it can be considered an outlier and removed. Various strategies can be adopted to handle missing values, such as filling them with the mean or predicting them based on other relevant variables. This ensures data integrity and consistency, providing a reliable foundation for subsequent data analysis.

[0079] The data storage module is used to save the pre-processed data samples in time series according to the collection date, and mark the timestamp and the cause type of cognitive impairment to form a data sample library.

[0080] The feature extraction unit is used to extract feature information from the preprocessed data samples using the correlation analysis method, including the scores and change trends of various cognitive functions, which serve as feature sets for subsequent model prediction training. The feature extraction unit includes:

[0081] The data processing module is used to organize the multidimensional assessment data in the data sample into a structured table, where each row represents a time point and each column represents the score of a cognitive function, as shown in Table 1 below;

[0082] The correlation analysis module is used to perform time series analysis on the scores of each cognitive function, calculate the mean, variance, and change trend of each cognitive function, and then use the Pearson correlation coefficient to calculate the correlation between various cognitive functions. The calculation formula is as follows:

[0083]

[0084] Where r represents the correlation between the two cognitive functions; x i and y i are the scores of two cognitive functions, and are the means of the two cognitive functions respectively;

[0085] Perform a p-value significance test on the correlation coefficient to determine whether the correlation coefficient is accurate; once it is confirmed to be accurate, use a heat map to display the correlation coefficients and correlation strengths between various cognitive functions;

[0086] The trend extraction module is used to calculate the linear change slope of each cognitive function score using the linear regression method, and perform time series analysis on the score of each cognitive function to extract the change trend of each cognitive function score;

[0087] The feature extraction module is used to cluster various cognitive functions according to the similarity of their change trends and identify functions with similar change patterns as feature information of the multidimensional evaluation data in the current data sample;

[0088] In one embodiment, the feature extraction module includes:

[0089] A processing module is used to establish a dimensionality elimination processing degree between any two cognitive functions based on the dimensional differences between the various cognitive functions, and to unify the change trends of the cognitive functions based on the dimensionality elimination processing degree to obtain a target change trend;

[0090] A center determination module is used to divide the target change trend into high change, medium change and low change to obtain multiple trend segments, determine the silhouette coefficient of each trend segment in the high change trend segment, select the high trend segment with the highest silhouette coefficient as the high change cluster center, determine the medium trend segment corresponding to the average distance of the medium change trend segment content as the medium change cluster center, and randomly select any low trend segment in the low change trend segment as the low change cluster center;

[0091] The clustering module is used to perform cluster analysis on the target change trend based on the high change cluster center, the medium change cluster center and the low change cluster center to obtain the clustering results;

[0092] An analysis module is used to obtain, from the clustering results, trend segments to be analyzed whose similarities with the high-variation cluster center, the medium-variation cluster center, and the low-variation cluster center are within a preset range;

[0093] The judgment module is used to analyze the time characteristics of the trend segments to be analyzed in different cognitive functions and determine whether the time series are consistent;

[0094] If so, it is determined that the cognitive functions corresponding to the trend segments to be analyzed have similar change patterns;

[0095] Otherwise, it is determined that the cognitive functions corresponding to the trend segments to be analyzed do not have similar change patterns;

[0096] The information determination module is used to use a set of cognitive functions with similar change patterns as feature information of the multidimensional evaluation data in the current data sample.

[0097] In this embodiment, the silhouette coefficient is used to indicate the degree of compactness in clustering.

[0098] In this embodiment, the high-change trend segment is more effective in identifying functional similarity, so the complex but accurate method of collecting silhouette coefficients is used to determine the cluster center, while the medium-change trend segment has the second most important effect on functional similarity, and the low-change trend segment has the least effect on functional similarity, so a less complex and rough method is selected to determine the cluster center to ensure clustering efficiency and clustering effect.

[0099] The beneficial effects of the above design scheme are: by dividing the target change trend into high change, medium change and low change, multiple trend segments are obtained, the silhouette coefficient of each trend segment in the high change trend segment is determined, the high trend segment with the highest silhouette coefficient is selected as the high change cluster center, the medium trend segment corresponding to the average distance of the medium change trend segment content is determined as the medium change cluster center, and any low trend segment in the low change trend segment is randomly selected as the low change cluster center, so that the clustering process is more targeted, ensuring that the clustering results can better represent the similarity between cognitive functions, ensuring the accuracy of the characteristic information of the multidimensional evaluation data in the current data sample, and providing a basis for the intelligent evaluation of patients with cognitive impairment.

[0100] Specifically, when conducting cluster analysis, regarding correlation analysis: if two cognitive functions are significantly correlated, it may indicate that they share certain neural mechanisms or influence each other; for example, memory and language ability may be highly correlated because language tasks usually rely on memory; regarding change trend analysis: if the score of a certain cognitive function shows a downward trend, it may indicate that the function is impaired; for example, a decrease in orientation scores may indicate a deterioration in spatial cognitive ability; based on this, through correlation analysis and trend extraction, we can deeply understand the changes in patients' cognitive functions and their interrelationships, and provide a scientific basis for clinical decision-making.

[0101] In this embodiment, for example, it is assumed that the evaluation data of a patient is as follows:

[0102] Time point Visual space and execution Abstract power memory attention Computing power Language skills Orientation 1 8 8 10 8 8 10 9 2 7 7 6 7 7 6 6 3 5 6 7 5 6 6 5

[0103] Table 1. Patient Assessment Data Sheet

[0104] Calculations based on the above method revealed that: Correlation analysis revealed a high correlation between memory and language ability (r = 0.95, p < 0.05); trend analysis revealed a downward trend in the scores of various cognitive functions, with orientation decreasing the fastest. This suggests that the patient's spatial cognitive ability is gradually deteriorating.

[0105] The beneficial effects achieved by the above content: By performing correlation analysis and trend extraction on the patient's multidimensional data, we can deeply understand the changes in the patient's cognitive function and their interrelationships, and provide a scientific basis for clinical decision-making.

[0106] The model prediction unit is used to create a neural network model, train the neural network model based on the preprocessed data samples and the filtered feature set, enable the neural network model to learn the rules and patterns in the data samples, and output the prediction results; the model prediction unit includes:

[0107] The sample partitioning module is used to divide the preprocessed data samples and the filtered feature set into training and test sets. The neural network model is trained using the training set, and the accuracy of the neural network model is verified using the test set. Specifically, by dividing the training and test sets into a 70%-30% or 80%-20% ratio, the neural network model learns the mapping between input features (i.e., scores for various cognitive functions in the Montreal Cognitive Assessment scale) and output targets (i.e., diagnosis of cognitive impairment, severity of illness, etc.). The training process includes forward propagation, loss calculation, backpropagation, and parameter updates. Through multiple iterations of training, the neural network model gradually converges and achieves good performance.

[0108] The performance verification module is used to compare the actual diagnostic results in the data sample with the predicted results of the neural network model to analyze whether the performance of the neural network model is accurate; if the verification passes, the neural network model is applied to actual clinical diagnosis;

[0109] The model optimization module is used to regularly update and iterate the neural network model with new multi-dimensional evaluation data to optimize the performance of the neural network model.

[0110] In one embodiment, the process of removing outliers and missing values ​​in the data sorting module is as follows:

[0111] Obtaining an evaluation data set of a historical patient in a data sample in a current dimension, obtaining a data evaluation maximum value and a data evaluation minimum value in the evaluation data set, and obtaining a standard deviation of the evaluation data set in a previous dimension;

[0112] Based on the maximum and minimum data evaluation values ​​and the standard deviation, the current data evaluation value is judged as abnormal to obtain the degree of data abnormality;

[0113] The calculation formula for the degree of data anomaly is as follows:

[0114]

[0115] Among them, K represents the degree of data anomaly of the current data evaluation value, σ represents the standard deviation of the evaluation data set, and δ max Indicates the maximum value of data evaluation, δ min represents the minimum value of data evaluation, γ1 represents the maximum value range ratio, γ2 represents the minimum value range ratio, and H represents a constant with a value of 5;

[0116] When the data abnormality level is greater than a preset level, the current data evaluation value is determined to be an abnormal value and removed; otherwise, the current data evaluation value is determined to be normal;

[0117] Obtain the data type and data distribution characteristics of the evaluation data set of the current dimension, and determine whether there are missing values ​​in the evaluation data set based on the data type and data distribution characteristics;

[0118] The missing value judgment method is as follows:

[0119] and

[0120] Among them, n represents the number of data types in the evaluation data set, N i represents the amount of data of the i-th data type, N0 represents the standard data amount, τ i represents the data volume weight of the i-th data type, R represents the capacity threshold, Q represents the distribution representation value of the data distribution characteristics under the evaluation data set, Q0 represents the standard distribution representation value, and Z represents the data distribution threshold;

[0121] If there are missing values, determine the location of the missing values ​​and remove the evaluation data range where the missing values ​​are located.

[0122] In this embodiment, different data distribution characteristics correspond to different values, which are preset.

[0123] In this embodiment, the maximum range ratio and the minimum range ratio have a value range between (0, 1) to represent a normal range condition.

[0124] In this embodiment, the data volume weight of a data type is pre-set according to the specific situation of the data type in the evaluation.

[0125] In this embodiment, the capacity threshold and the data distribution threshold are preset according to actual conditions and requirements.

[0126] The beneficial effects of the above design scheme are: by collecting and screening the multidimensional data of patients with cognitive impairment, removing outliers and missing values ​​to support the accuracy of the multidimensional data of patients with cognitive impairment and support reliable analysis conclusions, and further providing accurate and effective multidimensional data for the intelligent assessment and rehabilitation training integrated equipment for patients with cognitive impairment, through standardized outlier and missing value processing, optimizing the multidimensional data structure and calculation process and providing a high-quality data foundation for the intelligent assessment of patients with cognitive impairment.

[0127] The evaluation summary unit is used to analyze the prediction results and determine whether the patient has cognitive impairment and the degree of impairment of various cognitive functions;

[0128] The rehabilitation training unit is used to provide patients with personalized rehabilitation training recommendations based on the judgment results. It also uses a decision tree classification model to cluster and divide rehabilitation training recommendations by etiology type, assisting doctors in diagnosing and treating cognitive impairment. The rehabilitation training unit includes:

[0129] The plan-making module is used to develop targeted rehabilitation training plans based on the type and severity of the patient's cognitive impairment. The rehabilitation training plans include: memory training, language ability training, attention training, visual-spatial and executive ability training, calculation training, orientation training, and comprehensive cognitive training;

[0130] The plan division module is used to use the cause type of cognitive impairment, cognitive function score and severity as characteristic variables and rehabilitation training recommendations as target variables;

[0131] The cluster analysis module is used to build a decision tree classification model using CART to gradually classify rehabilitation training recommendations based on the cause and severity of the patient's cognitive impairment; and based on the classification results of the decision tree classification model, the rehabilitation training recommendations are clustered by cause type;

[0132] Based on the above, in this example, memory training, language training, attention training, visuospatial and executive function training, calculation training, orientation training, and comprehensive cognitive training are clustered and divided. For example, memory training is suitable for patients with Alzheimer's disease and mild cognitive impairment. Training methods include using flashcards or memory games, practicing recalling daily events or word lists, and combining visual or auditory cues to enhance memory. Language training is suitable for patients with aphasia and frontotemporal dementia. Training methods include performing naming exercises, such as picture descriptions, practicing retelling sentences or stories, and using speech therapy apps. Attention training is suitable for patients with attention deficit and vascular cognitive impairment. Training methods include using attention training software, such as the digit span test, performing selective attention tasks, such as the Stroop test, and practicing multitasking. Visuospatial and executive function training is suitable for patients with Parkinson's disease and Lewy body dementia. Training methods include performing clock drawing tests or copying patterns, using puzzles or building blocks to train spatial abilities, and practicing planning and organizing tasks, such as creating a schedule. Numerology training is suitable for patients with vascular cognitive impairment and Alzheimer's disease. Training methods include simple arithmetic exercises, using math games or apps, and integrating them into everyday situations, such as shopping calculations. Orientation training is suitable for patients with severe dementia and Alzheimer's disease. Training methods include practicing identifying time, place, and people, using maps or navigation tools to practice spatial orientation, and integrating them into everyday situations, such as recalling home addresses. Comprehensive cognitive training is suitable for patients with all cognitive impairments. Training methods include using comprehensive cognitive training software such as Lumosity and Cogni Fit, combining multiple cognitive tasks, such as memory, attention, and language.

[0133] The program generation module is used to assist doctors in optimizing rehabilitation plans based on the prediction results of the neural network model, combined with the rehabilitation training suggestions given by the decision tree classification model and the opinions of family members, and to develop a personalized rehabilitation training plan for each patient. Continuing with the above embodiment, for example, if the patient's orientation ability declines the fastest, it indicates that the patient's spatial cognitive ability is gradually deteriorating. Retrieval can be performed based on orientation ability as a characteristic variable, and the training method obtained is: practicing identifying time, place and people, using maps or navigation tools to train spatial orientation, and combining daily life scenes, such as recalling home addresses, as rehabilitation training suggestions for the current patient.

[0134] The dynamic adjustment module is used to regularly assess the patient's cognitive function and adjust the training program in a timely manner based on changes. Specifically, the purpose of regular assessments is to monitor the patient's cognitive status, detect changes in a timely manner, and adjust the rehabilitation program. The frequency of assessments can be determined based on the patient's specific situation, such as once every 3-6 months. If the assessment finds that the patient's orientation and cognitive function have improved through training, the training difficulty can be appropriately reduced or new training tasks can be introduced. If new cognitive issues are discovered, such as decreased attention, the corresponding training content needs to be increased, and attention training tasks such as digit span tests can be introduced.

[0135] The beneficial effects achieved by the above content are: by adopting the form of a laptop computer, medical equipment capable of intelligent assessment and rehabilitation training for cognitive impairment is integrated into an all-in-one device to improve the portability of existing equipment; by introducing intelligent technology, personalized design and comprehensive intervention, it can assist doctors to further improve the effectiveness of assessment and rehabilitation training, thereby improving the quality of life of patients.

[0136] Working Principle: The Montreal Cognitive Assessment (MoCA) is used to collect multidimensional assessment data from patients. Correlation analysis is used to extract the scores and change trends of various cognitive functions from the multidimensional assessment data. The characteristic information is then predicted using a neural network model to obtain predictions of whether the patient suffers from cognitive impairment and the severity of the condition. Based on the prediction results, corresponding rehabilitation training suggestions are retrieved from the decision tree classification model, and the rehabilitation training suggestions are optimized based on the patient's actual situation, so as to provide rehabilitation training for the patient.

[0137] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "including," "having," or any other variations thereof are intended to cover non-exclusive possessors, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or includes elements that are inherent to such process, method, article, or apparatus.

[0138] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions, and alterations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A portable device for intelligent assessment and rehabilitation training of patients with cognitive impairment, comprising: The device body (1) is characterized in that: the device body (1) is connected to an intelligent assessment training system, and the intelligent assessment training system includes: Intelligent assessment unit, using the Montreal Cognitive Assessment Scale to obtain multidimensional assessment data of patients; The data collection unit collects historical multi-dimensional evaluation data, determines outliers based on maximum, minimum, and standard deviation, and eliminates missing values ​​based on data type and data distribution characteristics to form data samples; A feature extraction unit extracts feature information from the data sample using a correlation analysis method, including scores and change trends of various cognitive functions as a feature set. The feature extraction process includes: unifying the change trends of cognitive functions based on the degree of dimension elimination, clustering based on the level of trends, and extracting cognitive functions with similar change patterns as feature information; Model prediction unit, which predicts the neural network model based on data samples and feature sets and outputs the prediction results; The assessment summary unit determines whether the patient has cognitive impairment and the degree of impairment based on the prediction results.

2. The portable device for intelligent assessment and rehabilitation training for patients with cognitive impairment according to claim 1, characterized in that: The feature extraction unit includes: The data processing module is used to organize the multidimensional assessment data in the data sample into a structured table, where each row represents a time point and each column represents the score of a cognitive function; The correlation analysis module is used to perform time series analysis on the scores of each cognitive function, calculate the mean, variance, and change trend of each cognitive function, and then use the Pearson correlation coefficient to calculate the correlation between various cognitive functions. The calculation formula is as follows: Where r represents the correlation between the two cognitive functions; x i and y i are the scores of two cognitive functions, and are the means of the two cognitive functions respectively; Perform a p-value significance test on the correlation coefficient to determine whether the correlation coefficient is accurate; once it is confirmed to be accurate, use a heat map to display the correlation coefficients and correlation strengths between various cognitive functions; The trend extraction module is used to calculate the linear change slope of each cognitive function score using the linear regression method, and perform time series analysis on the score of each cognitive function to extract the change trend of each cognitive function score; The feature extraction module is used to cluster various cognitive functions according to the similarity of their changing trends and identify functions with similar changing patterns as feature information of the multidimensional evaluation data in the current data sample.

3. The portable device for intelligent assessment and rehabilitation training for patients with cognitive impairment according to claim 2, characterized in that: The feature extraction module includes: A processing module is used to establish a dimensionality elimination processing degree between any two cognitive functions based on the dimensional differences between the various cognitive functions, and to unify the change trends of the cognitive functions based on the dimensionality elimination processing degree to obtain a target change trend; A center determination module is used to divide the target change trend into high change, medium change and low change to obtain multiple trend segments, determine the silhouette coefficient of each trend segment in the high change trend segment, select the high trend segment with the highest silhouette coefficient as the high change cluster center, determine the medium trend segment corresponding to the average distance of the medium change trend segment content as the medium change cluster center, and randomly select any low trend segment in the low change trend segment as the low change cluster center; The clustering module is used to perform cluster analysis on the target change trend based on the high change cluster center, the medium change cluster center and the low change cluster center to obtain the clustering results; An analysis module is used to obtain, from the clustering results, trend segments to be analyzed whose similarities with the high-variation cluster center, the medium-variation cluster center, and the low-variation cluster center are within a preset range; The judgment module is used to analyze the time characteristics of the trend segments to be analyzed in different cognitive functions and determine whether the time series are consistent; If so, it is determined that the cognitive functions corresponding to the trend segments to be analyzed have similar change patterns; Otherwise, it is determined that the cognitive functions corresponding to the trend segments to be analyzed do not have similar change patterns; The information determination module is used to use a set of cognitive functions with similar change patterns as feature information of the multidimensional evaluation data in the current data sample.

4. The portable device for intelligent assessment and rehabilitation training for patients with cognitive impairment according to claim 1, characterized in that: The data collection unit comprises: The data sorting module is used to clean the multi-dimensional evaluation data of historical patients in the collected data samples and remove outliers and missing values; The data storage module is used to save the pre-processed data samples in time series according to the collection date, and mark the timestamp and the cause type of cognitive impairment to form a data sample library.

5. The portable intelligent assessment and rehabilitation training device for patients with cognitive impairment according to claim 4, characterized in that: The process of removing outliers in the data sorting module is as follows: Obtaining an evaluation data set of a historical patient in a data sample in a current dimension, obtaining a data evaluation maximum value and a data evaluation minimum value in the evaluation data set, and obtaining a standard deviation of the evaluation data set in a previous dimension; Based on the maximum and minimum data evaluation values ​​and the standard deviation, the current data evaluation value is judged as abnormal to obtain the degree of data abnormality; When the data abnormality level is greater than a preset level, the current data evaluation value is determined to be an abnormal value and is removed; otherwise, the current data evaluation value is determined to be normal.

6. The portable device for intelligent assessment and rehabilitation training for patients with cognitive impairment according to claim 4, characterized in that: The process of removing missing values ​​in the data sorting module is as follows: Obtain the data type and data distribution characteristics of the evaluation data set of the current dimension, and determine whether there are missing values ​​in the evaluation data set based on the data type and data distribution characteristics; If there are missing values, determine the location of the missing values ​​and remove the evaluation data range where the missing values ​​are located.

7. The portable device for intelligent assessment and rehabilitation training for patients with cognitive impairment according to claim 1, characterized in that: The intelligent assessment training system further includes: The rehabilitation training unit provides patients with rehabilitation training suggestions based on the judgment results, and uses a decision tree classification model to cluster and divide the rehabilitation training suggestions according to the cause type; A user registration unit, used to register a patient's personal account in the intelligent assessment and training system through a display screen (3) and a button (21) to form a patient's personal database; A function selection unit is used to display a cognitive function assessment selection option on a display screen (3) after the patient completes login, for the doctor or family member to select; The human-computer interaction unit is used to enable the patient to interact with the rehabilitation training course through the sound hole (5) and the camera (31).

8. The portable device for intelligent assessment and rehabilitation training for patients with cognitive impairment according to claim 7, characterized in that: The rehabilitation training unit comprises: The plan-making module is used to develop targeted rehabilitation training plans based on the type and severity of the patient's cognitive impairment. The rehabilitation training plans include: memory training, language ability training, attention training, visual-spatial and executive ability training, calculation training, orientation training, and comprehensive cognitive training; The plan division module is used to use the cause type of cognitive impairment, cognitive function score and severity as characteristic variables and rehabilitation training recommendations as target variables; The cluster analysis module is used to build a decision tree classification model using CART to gradually classify rehabilitation training recommendations based on the cause and severity of the patient's cognitive impairment; and based on the classification results of the decision tree classification model, the rehabilitation training recommendations are clustered by cause type; The program generation module is used to assist doctors in optimizing rehabilitation plans and developing personalized rehabilitation training plans for each patient based on the prediction results of the neural network model, combined with the rehabilitation training suggestions and family members' opinions given by the decision tree classification model; The dynamic adjustment module is used to regularly evaluate the patient's cognitive function and adjust the training plan in a timely manner according to changes.

9. The portable device for intelligent assessment and rehabilitation training for patients with cognitive impairment according to claim 1, characterized in that: The model prediction unit includes: The sample partitioning module is used to divide the preprocessed data samples and the filtered feature set into a training set and a test set, train the neural network model with the training set, and verify the accuracy of the neural network model with the test set; The performance verification module is used to compare the actual diagnostic results in the data sample with the predicted results of the neural network model to analyze whether the performance of the neural network model is accurate; if the verification passes, the neural network model is applied to actual clinical diagnosis; The model optimization module is used to regularly update and iterate the neural network model with new multi-dimensional evaluation data to optimize the performance of the neural network model.

10. The portable device for intelligent assessment and rehabilitation training for patients with cognitive impairment according to claim 1, characterized in that: The device body (1) is composed of a base (2) and a display screen (3); a button (21) and a touch screen (22) are provided on the base (2); a card slot (4), a sound hole (5) and an interface (6) are provided on the side of the base (2); a camera (31) is provided at the upper end of the display screen (3); a stylus (41) for performing writing training on the touch screen (22) is embedded in the card slot (4).

Images