Cognitive level prediction method based on optimization algorithm and machine learning
By using a regression prediction model optimized by multiverse optimization algorithm in cognitive level prediction, combining gait parameters and MoCA scores, the problem of insufficient accuracy and generalization ability in cognitive level prediction in the existing technology is solved, and a more accurate and reliable cognitive impairment risk assessment is achieved.
Patent Information
- Application Number
- CN202411832051.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-12
- Publication Date
- 2025-04-04
AI Technical Summary
The existing gait analysis methods lack sufficient accuracy and generalization ability in cognitive level prediction, making it difficult to meet the needs of different individuals and situations.
A regression prediction model based on multiverse optimization algorithm (MVO) was used, combining gait parameters with Montreal Cognitive Assessment Scale (MoCA) scores, a multiple regression prediction model was established to improve the accuracy of cognitive impairment risk assessment.
Optimizing the random forest model through multiverse optimization algorithm significantly improves the accuracy and reliability of cognitive impairment risk prediction, with the MSE value as low as 1.6597, providing a powerful tool for early identification of cognitive impairment.
Smart Images
Figure CN119763827A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of medical-engineering integration technology, and more specifically, to a cognitive level prediction method based on optimization algorithm and machine learning. Background Art
[0002] Cognition plays a vital role in human life, including thinking, memory, learning, decision-making, and problem solving. As the global population aging trend significantly intensifies, cognitive decline has become a serious health challenge. Early identification of cognitive decline, especially the risk associated with cognitive diseases such as Alzheimer's disease, is crucial for taking intervention measures and providing appropriate treatment.
[0003] Gait is an important behavioral indicator, and gait abnormality is a common manifestation of multiple diseases and involvement of different systems. Gait disorders in the elderly may not only coexist with cognitive disorders, but are also related to decreased quality of life, increased risk of death, falls, trauma and other accidents. With the development of biophysics and computer science, the advancement and standardization of analytical methods, and the combination of gait with cognition, neuroimaging, and artificial intelligence, gait analysis has great research prospects. Combining gait parameters with artificial intelligence algorithms, cognitive level assessment can be performed and real-time online assessment can be achieved. It can more conveniently assess the cognitive state of the subjects, timely detect the cognitive level of the elderly, and detect the risk of cognitive decline in the elderly as early as possible, which has important clinical significance for improving the safety of elderly patients.
[0004] Although gait analysis has shown great potential in predicting cognitive levels, existing methods often lack sufficient accuracy and generalization ability, and are difficult to meet the needs of different individuals and situations. Therefore, there is an urgent need for a cognitive level prediction method based on optimization algorithms and machine learning, which aims to achieve a more accurate and real-time assessment of the cognitive level of the elderly by integrating gait parameters with advanced algorithm models, thereby filling the gap in existing technologies in this field and providing strong support for the early identification and intervention of cognitive impairment. Summary of the invention
[0005] In order to solve the technical problems existing in the above-mentioned prior art, the present invention uses gait parameters to establish a regression prediction model for the Montreal Cognitive Assessment (MoCA) score, and proposes a regression prediction model based on the Multi-Verse Optimizer (MVO) algorithm, which is applied to cognitive function prediction; it aims to provide an early, objective and accurate cognitive impairment risk assessment tool to better identify patients, provide early intervention and improve cognitive health.
[0006] To achieve the above object, the present invention provides the following technical solutions:
[0007] A cognitive level prediction method based on optimization algorithm and machine learning, the steps are as follows:
[0008] S1. Collect gait data using wearable devices;
[0009] S2. Preprocess the collected raw data
[0010] S3, using several different types of regression models to predict the MoCA scores of CSVD patients and optimizing them using the multiverse optimization algorithm;
[0011] S4. Use mean square error as the evaluation indicator to evaluate the performance of the model.
[0012] Furthermore, in step S1, a wearable device is used to collect multiple gait parameters such as distance, angle, time, etc. during walking; and the wearable device is used to complete four gait tasks, including two single tasks and two dual tasks.
[0013] Furthermore, in step S1, before model training, the original data needs to be preprocessed first, and the missing values are filled with the average value of the gait data of each task to remove abnormal data.
[0014] Furthermore, in step S3, the regression model includes linear regression, Lasso regression, elastic network, support vector regression, decision tree, and random forest.
[0015] Furthermore, 80% of the dataset is divided into training sets and 20% into test sets, and five-fold cross validation is used to ensure the stability and reliability of the model. The SMOTE algorithm is used for oversampling: by interpolating between samples of the minority class, new synthetic samples are generated, thereby increasing the sample size of the minority class.
[0016] Furthermore, in step S3, the multiverse optimization algorithm describes the search space of the optimization problem by establishing a multiverse model; the optimization process of the multiverse optimization algorithm is divided into two stages: detection and exploitation; in the detection stage, the gravitational effects of white holes and black holes enable the population to explore extensively in the search space; in the exploitation stage, by increasing the gravitational strength of the black hole, the population gradually focuses near the local optimal solution of the search space; in model optimization, after multiple rounds of iterative optimization, the multiverse optimization algorithm finally outputs a set of optimal model parameter configurations; these parameters are applied to the machine learning model, and its predictive performance is evaluated on the test set.
[0017] Furthermore, the single task is to walk freely at a habitual speed; the dual task includes a walking task with plates and a walking task for a speech fluency test.
[0018] Furthermore, in the task of walking with a plate, participants walked with the plate in both hands and avoided dropping the cup on the plate while walking;
[0019] Verbal Fluency Test Walking Task: Participants started walking while saying any animal name.
[0020] Furthermore, the collected gait parameters include stride length, stride speed, stride frequency, stance phase, swing phase, stride time, stance time, swing time, toe-off angle, and heel-strike angle of the left and right feet.
[0021] Furthermore, the Montreal Cognitive Assessment Scale was used to assess cognitive level.
[0022] Beneficial effects of the present invention:
[0023] Compared with the prior art, the cognitive level prediction method based on optimization algorithm and machine learning described in the present invention has the following technical characteristics or beneficial effects:
[0024] (1) Accurate prediction and early identification: By collecting and analyzing gait parameters, the present invention successfully established a regression prediction model associated with the Montreal Cognitive Assessment (MoCA) score. The results in the examples show that the random forest model optimized by the multiverse optimization algorithm (MVO) performed excellently in predictive performance, with an MSE value as low as 1.6597, which significantly improved the accuracy and reliability of cognitive impairment risk prediction. This achievement provides a powerful tool for early identification of cognitive impairment, which helps to take timely intervention measures before symptoms appear or in the early stages.
[0025] (2) Comprehensive evaluation of gait characteristics: The present invention not only considers single-task gait data, but also introduces dual-task gait data (such as the plate-carrying walking task and the verbal fluency test walking task), which can more comprehensively evaluate the individual's cognitive function and motor control ability. By collecting multiple gait parameters including stride length, pace, cadence, stance phase, swing phase, etc., the present invention provides rich data support for cognitive level prediction.
[0026] (3) Optimization algorithm improves prediction performance: The introduction of the multiverse optimization algorithm (MVO) provides a new perspective and solution for the optimization of model parameters. The MVO algorithm simulates the physical phenomena in the multiverse, realizes extensive exploration in the search space and focuses on the local optimal solution, so as to find a better model parameter configuration. The results in the embodiment show that the random forest model optimized by MVO has been significantly improved in prediction performance, which further proves the effectiveness of the optimization algorithm in improving prediction accuracy.
[0027] (4) Broad prospects for clinical application: The cognitive level prediction method of the present invention has the potential for real-time online evaluation and can be conveniently applied in clinical, community and home environments. By regularly monitoring gait data through wearable devices, doctors can promptly detect changes in patients' cognitive functions and provide patients with personalized treatment plans and health management recommendations. In addition, this method can also be used for rehabilitation evaluation and efficacy monitoring of patients with cognitive impairment, providing new ideas and means for the prevention, control and treatment of cognitive impairment.
[0028] (5) Promote cognitive science research: The implementation of the present invention not only provides an effective tool for early identification and intervention of cognitive impairment, but also promotes the in-depth development of cognitive science research. By integrating gait parameters with artificial intelligence algorithms, the present invention provides a new perspective and method for cognitive function assessment, helps to reveal the intrinsic connection between cognitive impairment and gait abnormality, and provides new ideas and data support for cognitive science research.
[0029] In summary, the cognitive level prediction method based on optimization algorithm and machine learning in the present invention has shown significant beneficial effects in accurate prediction, comprehensive evaluation, optimization algorithm, clinical application and scientific research, and provides new solutions and ideas for the prevention, control and treatment of cognitive impairment. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the present invention will be described in detail below in combination with the accompanying drawings and detailed embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:
[0031] Figure 1 It is a scatter plot of the prediction performance of the MVO-RF model. DETAILED DESCRIPTION
[0032] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Figure 1 The cognitive level prediction method based on optimization algorithm and machine learning is further explained, and the technical scheme in the embodiment of the present application is clearly and completely described; obviously, the described embodiment is only a part of the embodiments of the present application, not all the embodiments, based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.
[0033] Example 1
[0034] A cognitive level prediction method based on optimization algorithm and machine learning mainly includes the following steps:
[0035] (1) Data collection: using wearable devices to collect gait data
[0036] (2) Data processing: preprocessing the collected raw data
[0037] (3) Model establishment: Six different types of regression models were used to predict the MoCA scores of CSVD patients, including linear regression, Lasso regression (Least Absolute Shrinkage and Selection Operator Regression), Elastic Net, Support Vector Regression (SVR), Decision Tree, and Random Forest (RF). The Multi-Verse Optimizer (MVO) algorithm was used for optimization.
[0038] (4) Model evaluation: Mean square error (MSE) is used as an evaluation indicator to evaluate the performance of the model.
[0039] Example 2
[0040] This embodiment is a new embodiment or a supplement to Embodiment 1. A cognitive level prediction method based on optimization algorithm and machine learning has the following steps:
[0041] (1) Data Collection
[0042] Wearable devices were used to collect multiple gait parameters such as distance, angle, and time during walking. Wearable devices were used to complete four gait tasks, including two single tasks and two dual tasks, with a walking length of more than 10 meters. The single task was to walk freely at a habitual speed; the dual tasks included a plate-carrying walking task and a verbal fluency test (VFT) walking task. Plate-carrying walking task: Participants walked with plates in both hands and avoided dropping the cups in the plates during walking. VFT walking task: Participants started walking and said the name of any animal at the same time (fruit names can be used instead for multiple repetitions). The collected gait parameters included the stride length, speed, frequency, support phase, swing phase, stride time, support time, swing time, toe-off angle, and heel-strike angle of the left and right feet.
[0043] Cognitive ability was assessed using the Montreal Cognitive Assessment (MoCA).
[0044] (2) Data preprocessing
[0045] Before model training, the raw data needs to be preprocessed first, and missing values are filled with the average value of gait data for each task to remove abnormal data. The final dataset consists of 46 processed gait parameter features and MoCA scores.
[0046] (3) Model establishment
[0047] Six different types of regression models were used to predict MoCA scores, including linear regression, Lasso regression (Least Absolute Shrinkage and Selection Operator Regression), Elastic Net (ElasticNet), Support Vector Regression (SVR), Decision Tree (DecisionTree), and Random Forest (RF).
[0048] (4) Model training
[0049] 80% of the dataset is divided into training set and 20% into test set. Five-fold cross validation is used to ensure the stability and reliability of the model. Due to the imbalance of data categories, SMOTE (Synthetic Minority Over-sampling Technique) technology is used for oversampling. This method generates new synthetic samples by interpolating between samples of minority classes, thereby increasing the sample size of minority classes and improving the imbalance of categories during training. In order to ensure the accuracy and reliability of the results, each experiment is repeated 50 times, and the average of the results is taken as the final result.
[0050] (5) Model optimization
[0051] The Multi-Verse Optimizer (MVO) is a swarm intelligence optimization algorithm. The algorithm is inspired by the multiverse theory in physics, which holds that there may be other universes besides the one we live in. MVO simulates the movement behavior of populations in the multiverse under the joint action of white holes, black holes, and wormholes, and uses this behavior to solve optimization problems.
[0052] The MVO algorithm describes the search space of the optimization problem by establishing a multiverse model. In this model, each universe represents a potential solution, and the interactions between universes are simulated by the gravitational effects of white holes, black holes, and wormholes. White holes are used to simulate the exploration process in the search space, which guides the search by attracting the surrounding universes. In contrast, black holes are used to simulate the exploitation process in the search space, which focuses the search by swallowing the surrounding universes. This process of exploration and exploitation is similar to the diversity and reinforcement mechanisms in other swarm intelligence algorithms. Wormholes are used to simulate information transmission and communication. It enables information transmission between different universes through wormholes, thereby promoting the sharing of global information and the coordination of the search process. The advantage of the MVO algorithm is that it combines the characteristics of global search and local exploitation by simulating the gravitational effects of the multiverse.
[0053] The optimization process of MVO is divided into two stages: detection and exploitation. In the detection stage, the gravitational effects of white holes and black holes make the population explore widely in the search space. In the exploitation stage, by increasing the gravitational strength of the black hole, the population gradually focuses on the local optimal solution of the search space.
[0054] In model optimization, after multiple rounds of iterative optimization, the MVO algorithm finally outputs a set of optimal model parameter configurations. These parameters are applied to the machine learning model and its prediction performance is evaluated on the test set.
[0055] (6) Model evaluation
[0056] Evaluate the performance of models including Linear Regression, Lasso Regression, Elastic Net, SVR, Decision Tree, and Random Forest.
[0057] Table 1 Prediction performance of the model
[0058] Regressor MSE Linear Regression 8.2744 Lasso 8.1035 ElasticNet 8.1399 SVR 7.3144 Decision Tree 6.3872 Random Forest 1.7351 MVO-RF 1.6597
[0059] As can be seen from Table 1, the lowest MSE value of random forest is 1.7351, indicating that it performs best in predicting the degree of cognitive impairment of patients. The random forest model reduces the risk of overfitting and improves the generalization ability of the model by integrating multiple decision trees, so the prediction error is small. In contrast, the MSE values of linear regression and support vector regression are higher, 8.2744 and 8.1035 respectively, and their performance is relatively poor.
[0060] Random forest was selected for optimization, and the MVO algorithm was used to determine the optimal random forest parameter configuration. The final MSE dropped to 1.6597, and the prediction performance was significantly improved.
[0061] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. A cognitive level prediction method based on optimization algorithm and machine learning, characterized in that: Here are the steps: S1. Collect gait data using wearable devices; S2. Preprocess the collected raw data S3, using several different types of regression models to predict the MoCA scores of CSVD patients and optimizing them using the multiverse optimization algorithm; S4. Use mean square error as the evaluation indicator to evaluate the performance of the model.
2. The cognitive level prediction method based on optimization algorithm and machine learning as claimed in claim 1, characterized in that: In step S1, a wearable device is used to collect multiple gait parameters such as distance, angle, time, etc. during walking; and the wearable device is used to complete four gait tasks, including two single tasks and two dual tasks.
3. The cognitive level prediction method based on optimization algorithm and machine learning as claimed in claim 1, characterized in that: In step S1, before model training, the original data needs to be preprocessed first, and the missing values are filled with the average value of the gait data of each task to remove abnormal data.
4. The cognitive level prediction method based on optimization algorithm and machine learning as claimed in claim 1, characterized in that: In step S3, the regression model includes linear regression, Lasso regression, elastic network, support vector regression, decision tree, and random forest.
5. The cognitive level prediction method based on optimization algorithm and machine learning as claimed in claim 1, characterized in that: 80% of the dataset is divided into training sets and 20% into test sets. Five-fold cross validation is used to ensure the stability and reliability of the model. The SMOTE algorithm is used for oversampling: new synthetic samples are generated by interpolating between samples of the minority class, thereby increasing the sample size of the minority class.
6. The cognitive level prediction method based on optimization algorithm and machine learning as claimed in claim 1, characterized in that: In step S3, the multiverse optimization algorithm describes the search space of the optimization problem by establishing a multiverse model; the optimization process of the multiverse optimization algorithm is divided into two stages: detection and exploitation; in the detection stage, the gravitational effects of white holes and black holes enable the population to explore extensively in the search space; in the exploitation stage, by increasing the gravitational strength of the black hole, the population gradually focuses near the local optimal solution of the search space; in model optimization, after multiple rounds of iterative optimization, the multiverse optimization algorithm finally outputs a set of optimal model parameter configurations; these parameters are applied to the machine learning model, and its prediction performance is evaluated on the test set.
7. The cognitive level prediction method based on optimization algorithm and machine learning as claimed in claim 2, characterized in that: The single task is to walk freely at a habitual speed; the dual task includes a walking task of carrying a plate and a walking task of a speech fluency test.
8. The cognitive level prediction method based on optimization algorithm and machine learning as claimed in claim 7, characterized in that: The walking task with a plate: participants walked with a plate in both hands and avoided dropping the cup on the plate during walking; Verbal Fluency Test Walking Task: Participants started walking while saying any animal name.
9. The cognitive level prediction method based on optimization algorithm and machine learning as claimed in claim 8, characterized in that: The gait parameters collected included stride length, stride speed, stride frequency, stance phase, swing phase, stride time, stance time, swing time, toe-off angle, and heel strike angle of the left and right feet.
10. The cognitive level prediction method based on optimization algorithm and machine learning according to any one of claims 1 to 9, characterized in that: Cognitive level was assessed using the Montreal Cognitive Assessment Scale.