CN114202032A - Gait detection method and device based on reservoir model and computer storage medium - Google Patents
Gait detection method and device based on reservoir model and computer storage medium Download PDFInfo
- Publication number
- CN114202032A CN114202032A CN202111539098.4A CN202111539098A CN114202032A CN 114202032 A CN114202032 A CN 114202032A CN 202111539098 A CN202111539098 A CN 202111539098A CN 114202032 A CN114202032 A CN 114202032A
- Authority
- CN
- China
- Prior art keywords
- data
- gait
- layer
- gait detection
- reserve pool
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/1116—Determining posture transitions
- A61B5/1117—Fall detection
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/112—Gait analysis
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/40—Detecting, measuring or recording for evaluating the nervous system
- A61B5/4076—Diagnosing or monitoring particular conditions of the nervous system
- A61B5/4082—Diagnosing or monitoring movement diseases, e.g. Parkinson, Huntington or Tourette
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/245—Classification techniques relating to the decision surface
- G06F18/2453—Classification techniques relating to the decision surface non-linear, e.g. polynomial classifier
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- Heart & Thoracic Surgery (AREA)
- Veterinary Medicine (AREA)
- Public Health (AREA)
- Animal Behavior & Ethology (AREA)
- Surgery (AREA)
- Neurology (AREA)
- Medical Informatics (AREA)
- Physiology (AREA)
- Mathematical Physics (AREA)
- Neurosurgery (AREA)
- Dentistry (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Computational Linguistics (AREA)
- Developmental Disabilities (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Nonlinear Science (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The application relates to a gait detection method, equipment and a computer storage medium based on a reserve pool model, wherein the reserve pool model comprises an input layer, a reserve pool, a dimensionality reduction layer and an output layer; the gait detection method comprises the following steps: acquiring gait detection data; inputting gait detection data into an input layer to convert the gait detection data into a multi-dimensional array in a preset format; the multidimensional arrays enter the reserve pool, and feature data are obtained after calculation of the reserve pool; the feature data enters a dimensionality reduction layer, and dimensionality reduction processing is carried out on the feature data in the dimensionality reduction layer; and the feature data after the dimension reduction processing enters an output layer to obtain the gait classification probability of the gait detection data. This application realizes the gait and detects based on reserve tank model, improves the degree of accuracy that the gait detected.
Description
Technical Field
The present application generally relates to the field of detection classification. More particularly, the present application relates to a reservoir model based gait detection method, apparatus and computer storage medium.
Background
Frozen gait (FOG) is a gait disorder characterized by recurrent, transient, and delayed cessation of gait, often occurring in the middle and late stages of parkinson's disease, with unknown causes of the onset. Falls in parkinson's disease are essentially caused by FOG, which has a serious impact on the physical and mental health and quality of life of the patient. Studies have shown that parkinson's disease is not completely cured and drug therapy can often only alleviate the symptoms of frozen gait, while FOG attacks are random and sensitive to drugs and the environment making it clinically difficult to detect FOG and not possible to assess its severity accurately. Therefore, the early detection of the FOG event is the basis of the implementation of intervention measures, can help the sick patient to reduce the falling frequency and recover the normal activity as soon as possible, and meanwhile, the FOG detection can also be used as the basis of disease condition evaluation, thereby having important guidance and application on the research and treatment of the Parkinson frozen gait.
At present, the research on frozen gait is mainly divided into a deep learning-based method and a machine learning-based method, wherein the deep learning-based method is an end-to-end technology and achieves a classification effect mainly by training a hidden multilayer neural network; the other method based on machine learning is divided into supervised learning and unsupervised learning. In supervised learning, most of the methods are classified based on the characteristics such as statistical characteristics, freezing indexes, freezing frequency band energy and the like and by combining an algorithm of machine learning; in unsupervised learning, features are extracted mainly by a Principal Component Analysis (PCA) method and combined with a classifier of machine learning.
The randomness and sensitivity of FOG make the detection result not particularly ideal, the detection result lacks stability, and the characteristic extraction part is based on the characteristic which is usually manually screened and extracted, so that the whole experiment process becomes more complicated,
in deep learning, one obvious but computationally expensive solution (CNN) to solve this problem is to develop a generic classifier using a large amount of training data recorded from many subjects (with different health or pathological conditions). However, such methods for FOG diagnostic purposes face several challenges, including data collection, beat annotation, and technical challenges related to hardware implementation.
Disclosure of Invention
The application provides a gait detection method and device based on a reserve pool model and a computer storage medium, which aim to solve the problem that the detection result of the existing model is inaccurate during gait detection.
In order to solve the technical problem, the application provides a gait detection method based on a reserve pool model, wherein the reserve pool model comprises an input layer, a reserve pool, a dimensionality reduction layer and an output layer; the gait detection method comprises the following steps: acquiring gait detection data; inputting gait detection data into the input layer to be converted into a multi-dimensional array in a preset format; the multidimensional array enters the reserve pool, and characteristic data are obtained after calculation of the reserve pool; the feature data enters the dimensionality reduction layer, and dimensionality reduction processing is carried out on the feature data in the dimensionality reduction layer; and the feature data after the dimension reduction processing enters the output layer to obtain the gait classification probability of the gait detection data.
In one embodiment, the performing, at the dimensionality reduction layer, dimensionality reduction processing on the feature data includes: and performing linear transformation on the characteristic data to make the characteristic data subjected to the dimensionality reduction linearly independent.
In one embodiment, the linearly transforming the feature data includes: and performing linear transformation on the characteristic data by adopting a principal component analysis method.
In one embodiment, the output layer is provided with a plurality of non-linear classifiers; the step of enabling the feature data after the dimension reduction processing to enter the output layer to obtain the gait classification probability of the gait detection data comprises the following steps: and the feature data after the dimension reduction processing enters a selected nonlinear classifier to obtain the gait classification probability of the gait detection data.
In one embodiment, inputting the gait detection data into the input layer for conversion into a multi-dimensional array in a preset format comprises: gait detection data is input into the input layer to be converted into a multi-dimensional array, and the dimensions of the multi-dimensional array comprise the number of data segments, the number of data points of each data segment and the variable number of each data point.
In one embodiment, the acquiring gait detection data comprises: acquiring gait detection data in a detection time period; sliding the time window by a fixed step length by using the time window, and intercepting gait detection data to obtain a gait detection data section; the time window has more data points than the fixed step size; the inputting gait detection data to the input layer comprises: inputting the gait detection data segment into the input layer.
In one embodiment, the output layer in the reservoir model is trained from gait training data, which includes normal gait training data and frozen gait training data.
In one embodiment, the gait detection data and the gait training data each comprise: thigh acceleration data, calf acceleration data, and lower back acceleration data.
In order to solve the above technical problem, the present application provides an electronic device, which includes a processor and a memory, where the memory stores a computer program, and the processor is configured to execute the computer program to implement the steps of the above method.
In order to solve the above technical problem, the present application proposes a computer storage medium storing a computer program executed to implement the steps of the above method.
Different from the prior art, the reserve pool model in this application includes input layer, reserve pool, dimension reduction layer and output layer, and the gait detection method includes: acquiring gait detection data; inputting gait detection data into an input layer to convert the gait detection data into a multi-dimensional array in a preset format; the multidimensional arrays enter the reserve pool, and feature data are obtained after calculation of the reserve pool; the feature data enters a dimensionality reduction layer, and dimensionality reduction processing is carried out on the feature data in the dimensionality reduction layer; and the feature data after the dimension reduction processing enters an output layer to obtain the gait classification probability of the gait detection data. The reserve pool model that this application was used only needs the training output layer, and the training is fast to introduce the dimensionality reduction layer in the reserve pool model, reduce the collinearity between the data, improve the generalization performance of reserve pool model, improve the degree of accuracy that the reserve pool model carries out the gait and detects.
Drawings
The above and other objects, features and advantages of exemplary embodiments of the present disclosure will become readily apparent from the following detailed description read in conjunction with the accompanying drawings. Several embodiments of the present disclosure are illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings and in which like reference numerals refer to similar or corresponding parts and in which:
FIG. 1 is a schematic diagram of the structure of a reservoir model of the present application;
FIG. 2 is a schematic flow chart of a gait detection method based on a reservoir model according to the present application;
FIG. 3 is a schematic diagram of an electronic device of the present application;
FIG. 4 is a schematic diagram of a computer storage medium according to the present application.
Detailed Description
The technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the drawings in the embodiments of the present disclosure, and it is obvious that the described embodiments are some, but not all embodiments of the present disclosure. All other embodiments, which can be derived by a person skilled in the art from the embodiments disclosed herein without making any creative effort, shall fall within the protection scope of the present disclosure.
Specific embodiments of the present disclosure are described in detail below with reference to the accompanying drawings. Referring to fig. 1 and fig. 2, fig. 1 is a schematic structural diagram of a reservoir model of the present application, and fig. 2 is a schematic flow chart of a gait detection method based on the reservoir model of the present application.
The gait detection method of the embodiment is used for detecting normal gait and frozen gait, and is specifically based on a reserve pool model, wherein the reserve pool model comprises an input layer, a reserve pool, a dimensionality reduction layer and an output layer.
The reserve pool calculation simulates a circuit structure of recursively connected neurons in the brain, and the reserve pool consisting of randomly sparsely connected neurons is used as a hidden layer for performing high-dimensional and nonlinear representation on input. The reserve pool is pre-generated, with the generation process being independent of the training process. The reserve pool generated in advance has good attributes, so that good performance can be obtained only by training the weight from the reserve pool to the output layer. Compared with the traditional neural network, the training process of the reservoir model is simplified, the global optimality and good generalization capability after weight determination can be realized, and the problems that the training algorithm is complex and is easy to fall into local minimum in the traditional neural network are solved.
The reserve pool is used for expanding the input signal u (t) of the input layer into x (t) in a high-dimensional manner, so that x (t) can be expanded by the weight matrix Wout. The reserve pool also has short term memory capability, and can provide time contact. The reserve tank has adjustable parameters for adjusting the performance of the reserve tank. The parameters of the reserve pool include: reservoir size, reservoir sparsity, input scaling and internal connection weight spectrum radius.
Pool size refers to the number of neurons within the pool. Different choices of reservoir size are not only related to the data of the sample, but also have a large impact on the performance of the reservoir model. Generally speaking, the larger the pool size is, the better the performance of the network is, but the larger the pool size is, the larger the calculation amount is, and the overfitting risk is increased, so that the generalization capability of the model is reduced.
Reservoir sparsity refers to the connection relationships between neurons in the reservoir. The sparsity of the reservoir can represent the richness of vectors in the reservoir and also affect the nonlinear approximation capability of the reservoir model. When the sparseness of the reserve pool is larger, the richness of vectors in the reserve pool is higher, and the nonlinear approximation capability of the reserve pool is higher.
The input scaling refers to a scaling that the input signal multiplies before entering the reservoir neurons. Because of the diversity of sample data and the selection of the neuron activation function, in order to avoid the overlarge distribution range of the input data and the generation of abnormal values when the neuron activation function is input, the input data must be limited within a certain range through a scaling scale. One setting rule for the input scaling in this embodiment is: the scale determines the degree of nonlinearity of the reservoir response, with a greater degree of nonlinearity giving a greater scale.
The radius of the internal connection weight spectrum refers to the maximum absolute eigenvalue of the connection matrix between neurons in the reservoir. The precondition for the reservoir model to be stable is that the echo state property is satisfied, i.e. the decay history of the input u (t) defines the state x (t) of the reservoir. In most cases, the spectrum radius is less than 1, so that the echo state attribute of the reserve pool can be ensured, and the reserve pool model tends to be in a stable state.
Most of the parameter choices of the reserve pool are obtained empirically, and in a given parameter space, the optimal parameter is searched by adopting a trial and error method. Because the training process of the reservoir model is simpler and the verification time is less, random parameters are generally adopted for experiments, and which set of parameters with the best model effect is selected as the optimal parameters.
The gait detection method based on the reservoir model in the embodiment comprises the following steps.
S11: gait detection data is acquired.
The method of the embodiment mainly detects the gait of the examiner, so that firstly, gait detection data of the examiner is acquired, specifically, real-time gait data of the examiner in the walking process is acquired, in the embodiment, the gait detection data is acquired through a sensor arranged on the examiner, specifically, thigh acceleration data, calf acceleration data and lower back acceleration data related to the gait are acquired through inertial sensors worn on thighs, calves and lower back of the examiner, and each acceleration data comprises X, Y, Z acceleration data in three directions.
Furthermore, for a detector with a frozen gait, the frozen gait does not exist all the time, so in the embodiment, during detection, the obtained real-time gait data in the walking process is intercepted and then detected.
After gait detection data in a detection time period are obtained, sliding the time window by a fixed step length by using the time window, and intercepting the gait detection data to obtain a gait detection data segment; the time window has more data points than the fixed step.
For example, the time window has 50 data points and the fixed step has 5 data points. The data points represent the detection time points, the sensor obtains one data every 0.015s, the corresponding time window is 0.8s, and 45 repeated data points exist between adjacent gait detection data segments.
In the embodiment, the time window does not exceed 1s, and if the intercepted data segment is too short, the data point is insufficient, and the characteristics cannot be fully expressed; if it is too long, there arises a problem that the mixture of the normal feature and the frozen feature is not easily recognized.
S12: and inputting the gait detection data into an input layer to be converted into a multi-dimensional array in a preset format.
If the gait detection data segment is obtained in step S11, the gait detection data segment is input to the input layer. The data to the input layer needs to be converted into a multidimensional NumPy array having the shape of [ N, T, V ], N being the number of data segments, T being the number of data points in each data segment, and V being the number of variables in each data point, such as thigh acceleration data, calf acceleration data, and lower back acceleration data in this embodiment, each acceleration data including X, Y, Z acceleration data in three directions, and thus the number of variables is 9. In specific calculation, the multi-dimensional array is converted from a csv format to a mat format.
S13: and the multidimensional arrays enter a reserve pool, and feature data are obtained after calculation of the reserve pool.
And calculating the multidimensional arrays in the reserve pool by each neuron to obtain characteristic data.
S14: and the feature data enters a dimensionality reduction layer, and dimensionality reduction processing is carried out on the feature data in the dimensionality reduction layer.
In a reserve pool, data has multiple collinearity, so that the space of a solution is unstable, the dimensionality is disastrous, a model is complex, and the generalization capability of the model is weak; moreover, the high-dimensional space samples in the storage pool have sparsity, so that the data characteristics are difficult to find by the model; too many variables will interfere with the model search rules; sometimes, the feature matrix is too large, so that the problems of large calculated amount and long training time are caused.
In order to solve the problem, in this embodiment, a dimension reduction method is adopted, the number of feature attributes is reduced, irrelevant or redundant features are removed, the number of features is reduced, the model accuracy is improved, and the running time is reduced.
In this embodiment, the feature data is linearly transformed at the dimensionality reduction layer to realize dimensionality reduction, so that the feature data after dimensionality reduction is linearly independent, and mutual independence between feature attributes is ensured. The method specifically comprises the steps of extracting a subspace of a large-scale reserve pool state matrix, removing linearly related components, and replacing an original reserve pool state matrix with the subspace to obtain an output weight, so that the occurrence of a ill-conditioned solution and the generalization performance and prediction accuracy of a reserve pool calculation model are avoided.
The dimensionality reduction method can adopt principal component analysis (PCA, tenPCA) to carry out linear transformation on the characteristic data. Principal Component Analysis (PCA) is a data dimension reduction method for continuous attributes, which constructs an orthogonal transformation of original data, and the base of a new space removes the correlation of data under the base of the original space, and only a few new variables are needed to explain most variables in the original data. In applications, several new variables, so-called principal components, which are fewer than the original variables and can account for the variables in most data, are typically selected for modeling instead of the original variables. By linear variation, the original data set is changed into a set of representations which are linearly independent of each dimension.
S15: and the feature data after the dimension reduction processing enters an output layer to obtain the gait classification probability of the gait detection data.
The output layer is used for classifying the characteristic data so as to determine the normal gait probability and the frozen gait probability of the gait detection data. In the present embodiment, the output layer is provided with various non-linear classifiers, such as a multilayer perceptron (multi-layer perceptron), a Ridge estimate (Ridge estimate), a Support Vector Machine (Support Vector Machine), and the like.
When the storage pool model is used for detection, a certain adaptive nonlinear classifier can be selected according to data needing to be detected so as to ensure the accuracy of classification. The feature data after the dimensionality reduction processing enters a selected nonlinear classifier to obtain the gait classification probability of the gait detection data.
The steps of training the reservoir model are similar to the steps of the gait detection method, specifically, gait training data is input into the reservoir model to be trained, and the gait training data comprises normal gait training data and frozen gait training data.
In this embodiment, the gait training data is obtained by collecting gait data of 8 patients, wherein 237 times of freezing gait time occurs in the test process, and the finally intercepted data segments include 118656 data segments, wherein 23731 data segments of the test set and 94925 data segments of the training set are obtained by dividing the data segments in a ratio of 2: 8.
The gait detection method can be implemented by an electronic device, so the present application also provides an electronic device, please refer to fig. 3, fig. 3 is a schematic structural diagram of an embodiment of the electronic device of the present application, the electronic device 100 of the present embodiment may be a computer, and includes a processor 11 and a memory 12 connected to each other, and the electronic device 100 of the present embodiment can implement the embodiment of the method. Wherein the memory 12 has stored therein a computer program, and the processor 11 is configured to execute the computer program to implement the above-mentioned method.
The processor 11 may be an integrated circuit chip having signal processing capabilities. The processor 11 may also be a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
For the method of the above embodiment, it may exist in the form of a computer program, so that the present application provides a computer storage medium, please refer to fig. 4, and fig. 4 is a schematic structural diagram of an embodiment of the computer storage medium of the present application. The computer storage medium 200 of the present embodiment stores therein a computer program 21 that can be executed to implement the method in the above-described embodiments.
The computer storage medium 200 of this embodiment may be a medium that can store program instructions, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk, or may also be a server that stores the program instructions, and the server may send the stored program instructions to other devices for operation, or may self-operate the stored program instructions.
In the several embodiments provided in the present application, it should be understood that the disclosed method and apparatus may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, a division of modules or units is merely a logical division, and an actual implementation may have another division, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
Units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
While various embodiments of the present application have been shown and described herein, it will be obvious to those skilled in the art that such embodiments are provided by way of example only. Numerous modifications, changes, and substitutions will occur to those skilled in the art without departing from the spirit and scope of the present application. It should be understood that various alternatives to the embodiments of the application described herein may be employed in practicing the application. The following claims are intended to define the scope of the application and, accordingly, to cover module compositions, equivalents, or alternatives falling within the scope of these claims.
Claims (10)
1. A gait detection method based on a reserve pool model is characterized in that the reserve pool model comprises an input layer, a reserve pool, a dimensionality reduction layer and an output layer; the gait detection method comprises the following steps:
acquiring gait detection data;
inputting gait detection data into the input layer to be converted into a multi-dimensional array in a preset format;
the multidimensional array enters the reserve pool, and characteristic data are obtained after calculation of the reserve pool;
the feature data enters the dimensionality reduction layer, and dimensionality reduction processing is carried out on the feature data in the dimensionality reduction layer;
and the feature data after the dimension reduction processing enters the output layer to obtain the gait classification probability of the gait detection data.
2. The detection method according to claim 1, wherein the performing dimension reduction processing on the feature data at the dimension reduction layer includes:
and performing linear transformation on the characteristic data to make the characteristic data subjected to the dimensionality reduction linearly independent.
3. The detection method according to claim 2, wherein said linearly transforming the feature data comprises:
and performing linear transformation on the characteristic data by adopting a principal component analysis method.
4. The detection method according to claim 1, wherein the output layer is provided with a plurality of nonlinear classifiers; the step of enabling the feature data after the dimension reduction processing to enter the output layer to obtain the gait classification probability of the gait detection data comprises the following steps:
and the feature data after the dimension reduction processing enters a selected nonlinear classifier to obtain the gait classification probability of the gait detection data.
5. The detection method according to claim 1, wherein inputting gait detection data into the input layer for conversion into a multidimensional array in a preset format comprises:
gait detection data is input into the input layer to be converted into a multi-dimensional array, and the dimensions of the multi-dimensional array comprise the number of data segments, the number of data points of each data segment and the variable number of each data point.
6. The detection method as claimed in claim 1, wherein said acquiring gait detection data comprises:
acquiring gait detection data in a detection time period;
sliding the time window by a fixed step length by using the time window, and intercepting gait detection data to obtain a gait detection data section; the time window has more data points than the fixed step size;
the inputting gait detection data to the input layer comprises: inputting the gait detection data segment into the input layer.
7. The detection method according to claim 1, wherein the output layer in the reservoir model is obtained by gait training data training, and the gait training data comprises normal gait training data and frozen gait training data.
8. The detection method as claimed in claim 5, wherein the gait detection data and the gait training data each comprise: thigh acceleration data, calf acceleration data, and lower back acceleration data.
9. An electronic device, characterized in that the electronic device comprises a processor and a memory, in which a computer program is stored, the processor being adapted to execute the computer program to implement the steps of the method according to any of claims 1-8.
10. A computer storage medium, characterized in that the computer storage medium stores a computer program which is executed to implement the steps of the method according to any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111539098.4A CN114202032B (en) | 2021-12-15 | 2021-12-15 | Gait detection method, device and computer storage medium based on reserve pool model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111539098.4A CN114202032B (en) | 2021-12-15 | 2021-12-15 | Gait detection method, device and computer storage medium based on reserve pool model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114202032A true CN114202032A (en) | 2022-03-18 |
| CN114202032B CN114202032B (en) | 2023-07-18 |
Family
ID=80654323
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111539098.4A Active CN114202032B (en) | 2021-12-15 | 2021-12-15 | Gait detection method, device and computer storage medium based on reserve pool model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114202032B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014203039A1 (en) * | 2013-06-19 | 2014-12-24 | Aselsan Elektronik Sanayi Ve Ticaret Anonim Sirketi | System and method for implementing reservoir computing using cellular automata |
| CN104323780A (en) * | 2014-10-30 | 2015-02-04 | 上海交通大学 | Support vector machine-based pedestrian gait classifying system and method |
| US20170173262A1 (en) * | 2017-03-01 | 2017-06-22 | François Paul VELTZ | Medical systems, devices and methods |
| US20170258390A1 (en) * | 2016-02-12 | 2017-09-14 | Newton Howard | Early Detection Of Neurodegenerative Disease |
| CN108090559A (en) * | 2018-01-03 | 2018-05-29 | 华南理工大学 | A kind of construction method of antithesis reserve pool neural network model |
| CN108229661A (en) * | 2018-01-03 | 2018-06-29 | 华南理工大学 | A kind of depth echo state network model building method based on multiple coding re-projection |
| CN108901033A (en) * | 2018-06-20 | 2018-11-27 | 南京邮电大学 | Base station method for predicting based on echo state network |
| US20190365287A1 (en) * | 2018-05-30 | 2019-12-05 | Industry-Academic Cooperation Foundation, Dankook University | Apparatus and method for gait type classification using pressure sensor of smart insole |
| CN111241749A (en) * | 2020-01-13 | 2020-06-05 | 广西师范大学 | Permanent magnet synchronous motor chaos prediction method based on reserve pool calculation |
-
2021
- 2021-12-15 CN CN202111539098.4A patent/CN114202032B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014203039A1 (en) * | 2013-06-19 | 2014-12-24 | Aselsan Elektronik Sanayi Ve Ticaret Anonim Sirketi | System and method for implementing reservoir computing using cellular automata |
| CN104323780A (en) * | 2014-10-30 | 2015-02-04 | 上海交通大学 | Support vector machine-based pedestrian gait classifying system and method |
| US20170258390A1 (en) * | 2016-02-12 | 2017-09-14 | Newton Howard | Early Detection Of Neurodegenerative Disease |
| US20170173262A1 (en) * | 2017-03-01 | 2017-06-22 | François Paul VELTZ | Medical systems, devices and methods |
| CN108090559A (en) * | 2018-01-03 | 2018-05-29 | 华南理工大学 | A kind of construction method of antithesis reserve pool neural network model |
| CN108229661A (en) * | 2018-01-03 | 2018-06-29 | 华南理工大学 | A kind of depth echo state network model building method based on multiple coding re-projection |
| US20190365287A1 (en) * | 2018-05-30 | 2019-12-05 | Industry-Academic Cooperation Foundation, Dankook University | Apparatus and method for gait type classification using pressure sensor of smart insole |
| CN108901033A (en) * | 2018-06-20 | 2018-11-27 | 南京邮电大学 | Base station method for predicting based on echo state network |
| CN111241749A (en) * | 2020-01-13 | 2020-06-05 | 广西师范大学 | Permanent magnet synchronous motor chaos prediction method based on reserve pool calculation |
Non-Patent Citations (8)
| Title |
|---|
| AZARAKHSH JALALVAND 等: "Radar Signal Processing for Human Identification by Means of Reservoir Computing Networks", 《2019 IEEE RADAR CONFERENCE》, pages 1 - 6 * |
| YAN YAN 等: "Topological Descriptors of Gait Nonlinear Dynamics Toward Freezing-of-Gait Episodes Recognition in Parkinson\'s Disease", 《IEEE SENSORS JOURNAL》, vol. 22, no. 05, pages 4294 - 4304, XP011901724, DOI: 10.1109/JSEN.2022.3142750 * |
| 崔珊珊: "基于神经网络及储备池计算的模拟电路故障诊断的研究与实现", 《中国优秀硕士学位论文全文数据库信息科技辑》, no. 2019, pages 135 - 89 * |
| 李莉 等: "基于压缩感知的回声状态神经网络在时间序列预测中的应用", 《软件导刊》, vol. 19, no. 04, pages 9 - 13 * |
| 胡靖雯 等: "基于深度学习的步态识别方法", 《计算机应用》, vol. 40, no. 1, pages 69 - 73 * |
| 韩敏 等: "基于储备池主成分分析的多元时间序列预测研究", 《控制与决策》, vol. 24, no. 10, pages 1526 - 1530 * |
| 颜延: "用于小样本人体传感数据机器学习的神经网络表征与拓扑印记分析机理研究", 《中国博士学位论文全文数据库基础科学辑》, no. 2021, pages 006 - 276 * |
| 黄剑 等: "基于无线穿戴式传感系统的智能步态检测研究", 《华中科技大学学报(自然科学版)》, no. 10, pages 105 - 110 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114202032B (en) | 2023-07-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Shoeibi et al. | A comprehensive comparison of handcrafted features and convolutional autoencoders for epileptic seizures detection in EEG signals | |
| Sharma et al. | Seizures classification based on higher order statistics and deep neural network | |
| Diykh et al. | Classify epileptic EEG signals using weighted complex networks based community structure detection | |
| Xie et al. | Wavelet-based sparse functional linear model with applications to EEGs seizure detection and epilepsy diagnosis | |
| Acharya et al. | Use of principal component analysis for automatic classification of epileptic EEG activities in wavelet framework | |
| Pimentel et al. | A review of novelty detection | |
| Tuncer et al. | Tetromino pattern based accurate EEG emotion classification model | |
| Aydemir et al. | A Tunable-Q wavelet transform and quadruple symmetric pattern based EEG signal classification method | |
| Martis et al. | Epileptic EEG classification using nonlinear parameters on different frequency bands | |
| Cherian et al. | Theoretical and methodological analysis of EEG based seizure detection and prediction: An exhaustive review | |
| KR102141185B1 (en) | A system of detecting epileptic seizure waveform based on coefficient in multi-frequency bands from electroencephalogram signals, using feature extraction method with probabilistic model and machine learning | |
| UR et al. | Automated diagnosis of epileptic electroencephalogram using independent component analysis and discrete wavelet transform for different electroencephalogram durations | |
| Zarei et al. | Exploring Douglas-Peucker algorithm in the detection of epileptic seizure from multicategory EEG signals | |
| Praveena et al. | Retracted article: automatic epileptic seizure recognition using relieff feature selection and long short term memory classifier | |
| Rout et al. | Multifuse multilayer multikernel RVFLN+ of process modes decomposition and approximate entropy data from iEEG/sEEG signals for epileptic seizure recognition | |
| Daftari et al. | Detection of epileptic seizure disorder using EEG signals | |
| Nahzat et al. | Classification of epileptic seizure dataset using different machine learning algorithms and PCA feature reduction technique | |
| Saminu et al. | Epilepsy detection and classification for smart IoT devices using hybrid technique | |
| Ramakrishnan et al. | Epileptic eeg signal classification using multi-class convolutional neural network | |
| Hemachandira et al. | A Framework on Performance Analysis of Mathematical Model‐Based Classifiers in Detection of Epileptic Seizure from EEG Signals with Efficient Feature Selection | |
| Zhou et al. | Phase space reconstruction, geometric filtering based Fisher discriminant analysis and minimum distance to the Riemannian means algorithm for epileptic seizure classification | |
| Cabrerizo et al. | Classification and medical diagnosis of scalp EEG using artificial neural networks | |
| CN114202032B (en) | Gait detection method, device and computer storage medium based on reserve pool model | |
| Goel et al. | Analysis of Techniques and Methods for Automated EEG signal for Epilepsy Diagnosis: A Review | |
| Ward et al. | Data balanced bagging ensemble of convolutional-LSTM neural networks for time series data classification with an imbalanced dataset |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |