CN108629304A - A kind of freezing of gait online test method - Google Patents
A kind of freezing of gait online test method Download PDFInfo
- Publication number
- CN108629304A CN108629304A CN201810386698.3A CN201810386698A CN108629304A CN 108629304 A CN108629304 A CN 108629304A CN 201810386698 A CN201810386698 A CN 201810386698A CN 108629304 A CN108629304 A CN 108629304A
- Authority
- CN
- China
- Prior art keywords
- gait
- offline
- online
- freezing
- data
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
- G06V40/23—Recognition of whole body movements, e.g. for sport training
- G06V40/25—Recognition of walking or running movements, e.g. gait recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Pathology (AREA)
- Probability & Statistics with Applications (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The present invention relates to machine learning techniques fields, more particularly to are a kind of freezing of gait online test methods.This method includes：Offline gait data and offline gait video, the offline gait data in the acquisition offline gait processes of patient include thigh acceleration, thigh angular speed, shank acceleration, shank angular speed and plantar pressure；According to the offline gait data and the offline gait video, offline sample set is established；Build the Naive Bayes Classifier of offline normal gait and offline freezing of gait；By the Naive Bayes Classifier and the online gait data of the offline normal gait and offline freezing of gait, respectively obtain in online gait processes, the probability of online normal gait and online freezing of gait obtains testing result.This method can real-time, accurately detect freezing of gait, the patient that helps to fall ill in time restores walking and normal activity.
Description
Technical field
The present invention relates to machine learning techniques fields, more particularly to are a kind of freezing of gait online test methods.
Background technology
Parkinson's disease (Parkinson ' s Diease, hereinafter referred to as PD) is a kind of common nervous system degeneration disease,
Its main clinical characteristics includes that bradykinesia and reduction, limbs are tetanic and shake, may also have the different of some postures and gait
Often.The disease is mainly in elderly population, and average age of onset is about 60 years old, and the age of onset of the disease has decline to become in recent years
Gesture.
It is the most common symptom of advanced Parkinson patient that gait, which freezes (Freezing Of Gait, hereinafter referred to as FOG),
It is mainly shown as that the transience retardance of movement, breaking out for FOG can cause patient's walking disorder even to fall, reduce its life
Bioplasm amount.Therefore, its gait is measured in real time in patient's PD walking process, it is certain that patient is given when detecting FOG
Intervention guiding measure be particularly important.
In the related technology, about there are mainly two types of the detection methods of freezing of gait：1, the gait data sequence is detected just
The often energy ratio of movement frequency band and FOG frequency bands, according to the freeze threshold of manual setting come whether judging gait normally；2, detection disease
People walk during gait and normal gait Pearson correlation coefficients, by its correlation come whether judging gait normally.
Inventor has found in the implementation of the present invention:Above two detection method is required to preset threshold manually
Value, accuracy be not high；And in above two method, when feature space have it is high-dimensional when, manual setting freeze threshold can become
It obtains very cumbersome.
Invention content
Present invention seek to address that the technical problem that the existing detection method about freezing of gait is not accurate enough, to solve the skill
Art problem, the technical solution that embodiment of the present invention uses are：A kind of freezing of gait online test method is provided, including：
Offline gait data and offline gait video, the offline gait data in the acquisition offline gait processes of patient include that thigh adds
Speed, thigh angular speed, shank acceleration, shank angular speed and plantar pressure；According to the offline gait data and described
Offline gait video, establishes offline sample set；Offline normal gait is built respectively based on the offline sample set and is freezed offline
The Naive Bayes Classifier of gait；Obtain the online gait data in online gait processes；Pass through the offline normal gait
Naive Bayes Classifier with offline freezing of gait and the online gait data, respectively obtain in online gait processes,
The probability of online normal gait and online freezing of gait；Compare the general of the online normal gait and the online freezing of gait
Rate obtains testing result.
Optionally, described according to the offline gait data and the offline gait video, it is specific to establish offline sample set
Including：Divide the offline gait data and offline gait video with preset time span and fixed time interval, obtains
Several window datas and window video；According to the corresponding window video of the window data, the window data is marked
Gait types, the gait types be freezing of gait or normal gait；Corresponding feature is extracted from the window data
Vector；According to the gait types and described eigenvector of the window data, offline sample set is established.
Optionally, the thigh acceleration includes：Thigh X-axis acceleration, thigh Y-axis acceleration and thigh Z axis accelerate
Degree；The thigh angular speed includes：Thigh X-axis angular speed, thigh Y-axis angular speed and thigh Z axis angular speed；The shank adds
Speed：Shank X-axis acceleration, shank Y-axis acceleration and shank Z axis acceleration；The shank angular speed includes：Shank X-axis
Angular speed, shank Y-axis angular speed and shank Z axis angular speed；The plantar pressure includes the first segment for being uniformly distributed in sole
Point pressure, second node pressure and third node pressure.
Optionally, described that corresponding feature vector is extracted from the window data, it specifically includes：
By following formula, calculates and obtain first eigenvector；
Wherein, FI is first eigenvector；W (t, f) is to the thigh acceleration, the shank acceleration or the foot
Bottom pressure w (t) carries out the frequency-region signal after Short Time Fourier Transform (STFT)；
By following formula, frequency-region signal is calculated in the ENERGY E of [3Hz, 8Hz] frequency band as second feature vector：
It is vectorial using the thigh angular speed or the shank angular speed as third feature；
By following formula, calculates thigh or shank is respectively offset from the angle theta of vertical direction as fourth feature vector
Wherein, a is thigh acceleration or shank acceleration, and g is acceleration of gravity.
Optionally, the simplicity for building offline normal gait and offline freezing of gait respectively based on the offline sample set
Bayes classifier specifically includes：The average value and standard deviation of all feature vectors of each window data of statistics；It calculates
In the offline sample set, the prior probability of the normal gait and the freezing of gait；According to the elder generation of the normal gait
Test probability and labeled as normal gait window data all feature vectors average value and standard value, build normal gait
Naive Bayes Classifier；The institute of window data according to the prior probability of the freezing of gait and labeled as freezing of gait
The average value and standard value for having feature vector, build the Naive Bayes Classifier of freezing of gait.
Optionally, the Naive Bayes Classifier is indicated by following formula：
Wherein, P (xi| c) belong to the conditional probability of freezing of gait or normal gait for ith feature vector, P (c) is institute
State the prior probability of normal gait or the freezing of gait.
Optionally, P (xi| c) calculated by following formula：
Wherein, xiFor ith feature vector, μc,iFor the average value, σc,iFor the standard value.
Optionally, the online gait data obtained in online gait processes, specifically includes：
With the predetermined time period and fixed time interval, the online gait data of acquisition testing window.
Optionally, by the Naive Bayes Classifier and the online gait data, online walking is respectively obtained
Cheng Zhong, the probability of online normal gait and online freezing of gait, specifically includes：Calculate the corresponding feature of the detection window to
Amount；According to described eigenvector, by the Naive Bayes Classifier of the normal gait, it is online normal to calculate detection window
Walk probability of state；According to described eigenvector, by the Naive Bayes Classifier of the freezing of gait, calculating detection window is
The probability of online freezing of gait.
Optionally, the probability of the online normal gait and the online freezing of gait, obtains testing result,
It specifically includes：Judge whether the probability of online normal gait is more than the probability of online normal gait；If so, determining testing result
For normal gait；If not, it is determined that testing result is freezing of gait.
Freezing of gait online test method provided in an embodiment of the present invention passes through the simple pattra leaves built in off-line procedure
This grader, can real-time, accurately detect freezing of gait, the testing result, on the one hand can be used as subsequently to suffer from
Person implements the basis that guiding measure is implemented, and helps fall ill patient's recovery walking and normal activity in time；On the other hand, the inspection
It surveys result and patient's freezing of gait related symptoms information can also be provided, there is important guidance to make research and the treatment of freezing of gait
With.
Description of the drawings
Fig. 1 is a kind of flow diagram of freezing of gait online test method provided in an embodiment of the present invention；
Fig. 2 is the fixed bit of two six axis inertial sensors and three diaphragm pressure sensors provided in an embodiment of the present invention
It sets；
Fig. 3 is the flow diagram of step 12 in Fig. 1；
Fig. 4 is the process schematic for the freezing of gait on-line checking that another embodiment of the present invention provides.
Specific implementation mode
In order to make the purpose of the present invention, scheme and advantage be more clearly understood, with reference to embodiments, the present invention is carried out
It is further described.It should be appreciated that described herein, specific examples are only used to explain the present invention, is not used to limit this
Invention.In addition, as long as technical characteristic involved in invention described below different embodiments is not constituted each other
Conflict can be combined with each other.
In order to which those skilled in the art are better understood from the freezing of gait online test method that following inventive embodiments provide,
The structure principle of Naive Bayes Classifier is simply introduced first below.
Naive Bayes Classifier is a kind of algorithm based on bayesian theory in sorting algorithm set.It is not single deposits
, but an algorithm family, they have common rule in this algorithm family.For example, in the algorithm family
One of them common rule is that the feature vector being each classified and other feature vectors are independent from each other；For example, should
Another common rule in algorithm family is that each feature has identical weight；Feature any in this way and result all have
Relationship, and influence degree is identical.According to the algorithmic rule, suitable feature vector is chosen, establishes offline sample set
Bayes classifier can be built.Since the embodiment of the present invention is mainly for the detection of freezing of gait, in patient's walking
It the feature vector of selection suitable type in the process and offline sample is established according to this feature vector is particularly important,
It is the important evidence that can accurately detect freezing of gait.
The embodiment of the present invention provides a kind of freezing of gait online test method as a result, first under off-line state, acquisition
The gait data of different classifications of the patient in gait processes, and the data based on the different classifications extract the spy of corresponding type
Sign vector (feature vector of the extraction is independent between each other) then establishes offline sample according to the corresponding feature of this feature vector
Collection, can finally build the Naive Bayes Classifier that can accurately detect freezing of gait.When patient is in real-time walking, adopt
Collect the gait data in the real-time walking process of patient, you can the freezing of gait in accurate judgement patient's walking process is general online
On the one hand rate, the judging result can provide corresponding symptom information for patient's freezing of gait, on the other hand can be used as follow-up
Guiding measure it is basic in real time, have important directive function to the research and treatment of freezing of gait.
It describes in detail below to the embodiment of the present invention, referring to Fig. 1, Fig. 1 is one kind provided in an embodiment of the present invention
The flow diagram of freezing of gait online test method, as shown in Figure 1, this method includes：
Step 11, the offline gait data in the acquisition offline gait processes of patient and offline gait video, the offline step
State data include thigh acceleration, thigh angular speed, shank acceleration, shank angular speed and plantar pressure.
When " the offline gait data " is that freezing of gait occurs, the data of patient's walking characteristics can be embodied, in the present invention
In embodiment, which is divided into 15 classes, respectively in patient's gait processes：
1, acceleration and angular speed of the knee in X-axis, Y-axis and Z axis；
2, ankle is in the acceleration and angular speed of X-axis, Y-axis and Z axis；
3, the pressure of 3 nodes in vola.
Above-mentioned offline gait data may be used corresponding sensor and collect, and sensor is by the above-mentioned step collected
State data are further transmitted to processor, and processor can execute the method and step of the present invention according to the offline gait data.
Wherein, in order to acquire patient's knee and ankle be in the acceleration and angular speed of X-axis, Y-axis and Z axis can be according to figure
Two six axis inertial sensors are fixed at the knee and ankle of the right leg of patient by mode shown in 2；In order to acquire patient's gait
In walking process, the pressure of 3 nodes in vola can fix three diaphragm pressure sensors according to mode shown in Fig. 2
In the right vola of patient.
By in the offline walking process of patient, six axis inertial sensors and the collected experimental data of diaphragm pressure sensor
As the offline gait data in the embodiment of the present invention.Wherein, the quantity of sensor and placement location difference, then collect
Offline gait data is also different, in other embodiments, can also be according to the thigh and calf of patient or the severity of left and right leg
Deng the reasonable adjustment for carrying out number of sensors and position, more accurately to be acquired to offline gait data.
Since the offline gait data is mainly used for building Naive Bayes Classifier, it is walked offline in acquisition patient
While state data, it is also necessary to offline gait video of the patient in offline walking process is recorded by relevant video equipment,
From the offline gait video, whether the gait that can intuitively observe patient is freezing of gait, as subsequent builds simplicity pattra leaves
One important parameter of this grader.
Step 12, according to the offline gait data and the offline gait video, establish offline sample set.
After obtaining above-mentioned offline gait data and offline gait video, can further to the offline gait data and
Offline gait video is split the extraction with feature vector respectively, and to establish offline sample set, offline sample set herein is
Build the required underlying parameter of Naive Bayes Classifier.
Specifically, as shown in figure 3, step 12 includes：
Step 121 divides the offline gait data and offline step with preset time span and fixed time interval
State video obtains several window datas and window video.
Wherein, several window datas and window video correspond, and the feature extracted in each window data
Vector is a characteristic value, as the part for establishing offline sample set；And it can be with intuitive judgment in each window video
Gait types corresponding to it are normal gait or freezing of gait.
Step 122, according to the corresponding window video of the window data, mark the gait class of the window data
Type, the gait types are freezing of gait or normal gait.
This step is the process that the gait types corresponding to each window video are judged and recorded.
Step 123 extracts corresponding feature vector from the window data.
This step is to be extracted four to normal gait and freezing of gait most discrimination property from above-mentioned window data
Mutually independent feature vector again, calculation difference are as follows：
By following formula, calculates and obtain first eigenvector；
Wherein, FI is first eigenvector；W (t, f) is to the thigh acceleration, the shank acceleration or the foot
Bottom pressure w (t) carries out the frequency-region signal after Short Time Fourier Transform (STFT)；
By following formula, frequency-region signal is calculated in the ENERGY E of [3Hz, 8Hz] frequency band as second feature vector：
It is vectorial using the thigh angular speed or the shank angular speed as third feature；
By following formula, calculates thigh or shank is respectively offset from the angle theta of vertical direction as fourth feature vector
Wherein, a is thigh acceleration or shank acceleration, and g is acceleration of gravity.
Step 124, gait types and described eigenvector according to the window data, establish offline sample set.
The offline sample set includes the offline sample set of normal gait and the offline sample set of freezing of gait.In the sample set
Ith feature vector is denoted as x in the present embodiment comprising all feature vectorsi, the i feature vector be mutually only
It is vertical, collectively constitute the offline sample set of normal gait or freezing of gait, in the present embodiment, can be calculated its normal gait or
In the offline sample set of person's freezing of gait, there are 22 feature vectors.
Step 13, the simple pattra leaves for building offline normal gait and offline freezing of gait respectively based on the offline sample set
This grader.
The Naive Bayes Classifier is indicated by following formula：
Wherein, P (xi| c) belong to the conditional probability of freezing of gait or normal gait for ith feature vector, P (c) is institute
State the prior probability of normal gait or the freezing of gait.
Specifically, P (xi| c) calculated by following formula：
Therefore, Naive Bayes Classifier is built, needs to calculate normal gait window data or freezing of gait window data
In ith feature value xiThe average value mu of upper valuec,iAnd standard deviation sigmac,i。
Wherein, average value muc,iShown in following formula：
In formula, ∑ Vc,iIndicate normal gait window data or abnormal gait the window data value in ith feature value
Summation, nc,iIndicate the number of normal gait window data or abnormal gait window data.
Wherein, standard deviation sigmac,iShown in following formula：
In the present embodiment, ith feature vector belongs to freezing of gait or the prior probability of normal gait is regarded by window
Frequency can be calculated, and calculation formula is as follows：
In formula, D indicates the sum of window video in offline sample set, DcIndicate normal gait or exception in offline sample set
The quantity of gait window video.
In the average value mu for obtaining normal gait and freezing of gait through the above stepsc,i, standard deviation sigmac,iWith prior probability it
Afterwards, you can to build the Naive Bayes Classifier of normal gait and freezing of gait respectively；Building Naive Bayes Classification
Under the Naive Bayes Classifier of the Naive Bayes Classifier of the normal gait and freezing of gait can be applied to after device
It states in the on-line checking stage.
Online gait data in step 14, the online gait processes of acquisition.
I.e. with predetermined time period and fixed time interval, the online gait data of acquisition testing window.
Step 15, by the Naive Bayes Classifier of the offline normal gait and offline freezing of gait and it is described
Line gait data respectively obtains in online gait processes, the probability of online normal gait and online freezing of gait.
The probability of step 16, the online normal gait and the online freezing of gait, obtains testing result.
Based on the Naive Bayes Classifier that step 13 is built, the probability that the moment window data is not freezing of gait is calculated
ForThe probability for being freezing of gait isWork as P1>P2When, testing result
For freezing of gait, work as P1<P2When, testing result is normal gait.
Freezing of gait online test method provided in an embodiment of the present invention, passes through the naive Bayesian built in off-line procedure
Grader can real-time, quickly and accurately detect on the one hand the freezing of gait of patient Parkinson, the testing result can be made
Subsequently to implement the basis that guiding measure is implemented to patient, fall ill patient's recovery walking and normal activity are helped in time；Separately
On the one hand, which can also be provided patient's freezing of gait related symptoms information, have to the research and treatment of freezing of gait
Important directive function.
The process of freezing of gait on-line checking is discussed in detail with a specific embodiment below, the process as shown in Figure 4
It specifically includes：Off-line learning, establishes offline sample set and on-line checking obtains testing result.
1, off-line learning is established offline sample set and is comprised the following processes：
1.1,2 six axle sensors and 3 acceleration transducers are fixed on the right leg of patient according to mode shown in Fig. 2
Knee, at ankle and vola.Acquire acceleration, angular velocity data and foot at knee and ankle in the daily walking process of patient
Base pressure force data records the process video as offline gait video as offline gait data, wherein sensor is adopted
Sample frequency is 100Hz, and the offline gait data for 15 types that sensor measures is as shown in table 1.
1 sensing data type of table
| Right thigh X-axis acceleration n1 | Right thigh Y-axis acceleration n2 | Right thigh Z axis acceleration n3 |
| Right leg X-axis acceleration n4 | Right leg Y-axis acceleration n5 | Right leg Z axis acceleration n6 |
| Right thigh X-axis angular speed n7 | Right thigh Y-axis angular speed n8 | Right thigh Z axis angular speed n9 |
| Right leg X-axis angular speed n10 | Right leg Y-axis angular speed n11 | Right leg Z axis angular speed n12 |
| 1 pressure n of right crus of diaphragm propodite point13 | 2 pressure n of right crus of diaphragm propodite point14 | 3 pressure n of right crus of diaphragm propodite point15 |
1.2, with the time interval of 100ms, the offline gait data in the time slip-window segmentation step 1.1 of 256ms wide,
Obtain 39606 window datas and 39606 window videos.Then, the corresponding offline gait video of control, marks above-mentioned each window
(normal gait is labeled as c to the corresponding gait types of mouth data1, freezing of gait is labeled as c2), it is corresponding to finally obtain normal gait
Window video is 25536, the corresponding window video of freezing of gait 14070.
1.3, extraction obtains the relevant feature vector of gait and is：FI indexes, the energy for freezing frequency band, thigh and calf angular speed,
Thigh and calf deviates the angle of vertical direction, and for calculation with reference to above-described embodiment, details are not described herein, calculates the offline sample
The quantity for all feature vectors that collection includes is 22, and ith feature vector is denoted as xi, the i feature vector is mutual indepedent,
Offline sample set is collectively constituted, as shown in table 2.
Table 2
1.4, relevant parameter is calculated, Naive Bayes Classifier is built.The process will calculate Naive Bayes Classifier
Two parameters.
Parameter one：Normal gait window data and freezing of gait window data are in ith feature value xiUpper value is averaged
Value μc,iAnd standard deviation sigmac,i, μc,iThe following formula of calculation shown in：
σc,iThe following formula of calculation shown in：
In formula, ∑ Vc,iIndicate normal gait window data or freezing of gait the window data value in ith feature value
Summation, nc,iIndicate a quantity of normal gait window video or abnormal gait window video.
Parameter two：The prior probability P (c) of normal gait and freezing of gait.Shown in following formula：
In formula, D indicates the window sum of offline sample set, DcIndicate normal gait or freezing of gait window in offline sample set
The quantity of mouth video.It is 25536 to have shown that normal gait corresponds to window video in above-described embodiment, and freezing of gait is corresponding
Window video is 14070, then the prior probability of normal gait is：Freezing of gait
Prior probability is：
1.5, Naive Bayes Classifier is builtWherein P (xi| c) it is i-th
The class conditional probability of characteristic quantity, shown in following formula：
The calculation formula of Naive Bayes Classifier can be built according to parameter one and parameter two.
2, on-line checking obtains testing result and comprises the following processes：
2.1, real-time acquisition window data, with the time interval of 100ms, the time slip-window acquisition window number of 256ms wide
According to.
2.2, to each feature vector in the window data computational chart 2, the value of ith feature vector is denoted as xi。
2.3, the window data is calculated in c classes sample (normal gait c1, freezing of gait c2) in ith feature value
Class conditional probability P (xi|c)。
2.4, the Naive Bayes Classifier built based on step 1.5, it is not freezing of gait to calculate the moment window data
Probability beThe probability for being freezing of gait isWork as P1>P2When, inspection
Survey result is freezing of gait, works as P1<P2When, testing result is normal gait.
Mode the above is only the implementation of the present invention is not intended to limit the scope of the invention, every to utilize this
Equivalent structure or equivalent flow shift made by description of the invention and accompanying drawing content, it is relevant to be applied directly or indirectly in other
Technical field is included within the scope of the present invention.
Claims (10)
1. a kind of freezing of gait online test method, which is characterized in that including：
The offline gait data in the offline gait processes of patient and offline gait video are obtained, the offline gait data includes big
Leg acceleration, thigh angular speed, shank acceleration, shank angular speed and plantar pressure；
According to the offline gait data and the offline gait video, offline sample set is established；
Build the Naive Bayes Classifier of offline normal gait and offline freezing of gait respectively based on the offline sample set；
Obtain the online gait data in online gait processes；
By the Naive Bayes Classifier and the online gait data of the offline normal gait and offline freezing of gait,
It respectively obtains in online gait processes, the probability of online normal gait and online freezing of gait；
The probability for comparing the online normal gait and the online freezing of gait, obtains testing result.
2. online test method according to claim 1, which is characterized in that described according to the offline gait data and institute
Offline gait video is stated, offline sample set is established and specifically includes：
Divide the offline gait data and offline gait video with preset time span and fixed time interval, if obtaining
Dry window data and window video；
According to the corresponding window video of the window data, the gait types of the window data, the gait class are marked
Type is freezing of gait or normal gait；
Corresponding feature vector is extracted from the window data；
According to the gait types and described eigenvector of the window data, offline sample set is established.
3. online test method according to claim 2, which is characterized in that the thigh acceleration includes：Thigh X-axis adds
Speed, thigh Y-axis acceleration and thigh Z axis acceleration；
The thigh angular speed includes：Thigh X-axis angular speed, thigh Y-axis angular speed and thigh Z axis angular speed；
The shank acceleration：Shank X-axis acceleration, shank Y-axis acceleration and shank Z axis acceleration；
The shank angular speed includes：Shank X-axis angular speed, shank Y-axis angular speed and shank Z axis angular speed；
The plantar pressure includes the first node pressure, second node pressure and third node pressure for being uniformly distributed in sole.
4. online test method according to claim 3, which is characterized in that described extracted from the window data corresponds to
Feature vector, specifically include：
By following formula, calculates and obtain first eigenvector；
Wherein, FI is first eigenvector；W (t, f) is pressed the thigh acceleration, the shank acceleration or the vola
Power w (t) carries out the frequency-region signal after Short Time Fourier Transform (STFT)；
By following formula, frequency-region signal is calculated in the ENERGY E of [3Hz, 8Hz] frequency band as second feature vector：
It is vectorial using the thigh angular speed or the shank angular speed as third feature；
By following formula, calculates thigh or shank is respectively offset from the angle theta of vertical direction as fourth feature vector
Wherein, a is thigh acceleration or shank acceleration, and g is acceleration of gravity.
5. online test method according to claim 4, which is characterized in that described to distinguish structure based on the offline sample set
The Naive Bayes Classifier for building offline normal gait and offline freezing of gait, specifically includes：
The average value and standard deviation of all feature vectors of each window data of statistics；
It calculates in the offline sample set, the prior probability of the normal gait and the freezing of gait；
According to the prior probability of the normal gait and labeled as normal gait window data all feature vectors it is flat
Mean value and standard value build the Naive Bayes Classifier of normal gait；
According to the prior probability of the freezing of gait and labeled as freezing of gait window data all feature vectors it is flat
Mean value and standard value build the Naive Bayes Classifier of freezing of gait.
6. online test method according to claim 5, which is characterized in that the Naive Bayes Classifier passes through as follows
Formula indicates：
Wherein, P (xi| c) belong to the conditional probability of freezing of gait or normal gait for ith feature vector, P (c) is described normal
The prior probability of gait or the freezing of gait.
7. according to the method described in claim 6, it is characterized in that, P (xi| c) calculated by following formula：
Wherein, xiFor ith feature vector, μc,iFor the average value, σc,iFor the standard value.
8. according to the method described in claim 6, it is characterized in that, the online gait number obtained in online gait processes
According to specifically including：
With the predetermined time period and fixed time interval, the online gait data of acquisition testing window.
9. according to the method described in claim 8, it is characterized in that, passing through the Naive Bayes Classifier and the online step
State data respectively obtain in online gait processes, and the probability of online normal gait and online freezing of gait specifically includes：
Calculate the corresponding feature vector of the detection window；
According to described eigenvector, by the Naive Bayes Classifier of the normal gait, calculate detection window be it is online just
Often step probability of state；
According to described eigenvector, by the Naive Bayes Classifier of the freezing of gait, it is online freeze to calculate detection window
Knot step probability of state.
10. according to the method described in claim 9, it is characterized in that, the online normal gait and described online
The probability of freezing of gait, obtains testing result, specifically includes：
Judge whether the probability of online normal gait is more than the probability of online normal gait；
If so, determining that testing result is normal gait；
If not, it is determined that testing result is freezing of gait.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810386698.3A CN108629304B (en) | 2018-04-26 | 2018-04-26 | Freezing gait online detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810386698.3A CN108629304B (en) | 2018-04-26 | 2018-04-26 | Freezing gait online detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108629304A true CN108629304A (en) | 2018-10-09 |
| CN108629304B CN108629304B (en) | 2020-12-08 |
Family
ID=63694729
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810386698.3A Active CN108629304B (en) | 2018-04-26 | 2018-04-26 | Freezing gait online detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108629304B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109480857A (en) * | 2018-12-29 | 2019-03-19 | 中国科学院合肥物质科学研究院 | A kind of device and method for the detection of Parkinsonian's freezing of gait |
| CN110151190A (en) * | 2019-05-23 | 2019-08-23 | 西南科技大学 | A kind of orthopaedics postoperative rehabilitation monitoring method and system |
| CN115171886A (en) * | 2022-07-25 | 2022-10-11 | 北京戴来科技有限公司 | Frozen gait detection method and device based on random forest algorithm and storage medium |
| CN115188468A (en) * | 2022-07-25 | 2022-10-14 | 北京戴来科技有限公司 | Frozen gait detection method and device based on support vector machine, and storage medium |
| CN117298449A (en) * | 2023-10-31 | 2023-12-29 | 首都医科大学宣武医院 | Closed-loop DBS regulation and control method and system based on wearable equipment |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012167328A1 (en) * | 2011-06-10 | 2012-12-13 | Bright Devices Group Pty Ltd | Freezing of gait cue apparatus |
| CN103886341A (en) * | 2014-03-19 | 2014-06-25 | 国家电网公司 | Gait behavior recognition method based on feature combination |
| CN104834888A (en) * | 2014-12-04 | 2015-08-12 | 龙岩学院 | Abnormal gait identification method capable of facilitating screening Parkinsonism |
| CN105142714A (en) * | 2013-01-21 | 2015-12-09 | 卡拉健康公司 | Devices and methods for controlling tremor |
| CN107361773A (en) * | 2016-11-18 | 2017-11-21 | 深圳市臻络科技有限公司 | For detecting, alleviating the device of Parkinson's abnormal gait |
| US20180064218A1 (en) * | 2016-09-05 | 2018-03-08 | Samsung Electronics Co., Ltd. | Walking assistance method and apparatuses |
-
2018
- 2018-04-26 CN CN201810386698.3A patent/CN108629304B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012167328A1 (en) * | 2011-06-10 | 2012-12-13 | Bright Devices Group Pty Ltd | Freezing of gait cue apparatus |
| CN105142714A (en) * | 2013-01-21 | 2015-12-09 | 卡拉健康公司 | Devices and methods for controlling tremor |
| CN103886341A (en) * | 2014-03-19 | 2014-06-25 | 国家电网公司 | Gait behavior recognition method based on feature combination |
| CN104834888A (en) * | 2014-12-04 | 2015-08-12 | 龙岩学院 | Abnormal gait identification method capable of facilitating screening Parkinsonism |
| US20180064218A1 (en) * | 2016-09-05 | 2018-03-08 | Samsung Electronics Co., Ltd. | Walking assistance method and apparatuses |
| CN107361773A (en) * | 2016-11-18 | 2017-11-21 | 深圳市臻络科技有限公司 | For detecting, alleviating the device of Parkinson's abnormal gait |
Non-Patent Citations (1)
| Title |
|---|
| SINZIANA MAZILU ETC: "Online Detection of Freezing of Gait with Smartphones and Machine Learning Techniques", 《2012 6TH INTERATIONAL CONFERENCE ON PERVASIVE COMPUTING TECHNOLOGIES FOR HEALTHCARE (PERVASIVEHEALTH) AND WORKSHOPS》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109480857A (en) * | 2018-12-29 | 2019-03-19 | 中国科学院合肥物质科学研究院 | A kind of device and method for the detection of Parkinsonian's freezing of gait |
| CN109480857B (en) * | 2018-12-29 | 2021-09-14 | 中国科学院合肥物质科学研究院 | Device and method for detecting frozen gait of Parkinson disease patient |
| CN110151190A (en) * | 2019-05-23 | 2019-08-23 | 西南科技大学 | A kind of orthopaedics postoperative rehabilitation monitoring method and system |
| CN115171886A (en) * | 2022-07-25 | 2022-10-11 | 北京戴来科技有限公司 | Frozen gait detection method and device based on random forest algorithm and storage medium |
| CN115188468A (en) * | 2022-07-25 | 2022-10-14 | 北京戴来科技有限公司 | Frozen gait detection method and device based on support vector machine, and storage medium |
| CN115188468B (en) * | 2022-07-25 | 2023-04-25 | 北京戴来科技有限公司 | Freezing gait detection method, device, equipment and storage medium based on support vector machine |
| CN117298449A (en) * | 2023-10-31 | 2023-12-29 | 首都医科大学宣武医院 | Closed-loop DBS regulation and control method and system based on wearable equipment |
| CN117298449B (en) * | 2023-10-31 | 2024-04-09 | 首都医科大学宣武医院 | Closed-loop DBS regulation and control method and system based on wearable equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108629304B (en) | 2020-12-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108629304A (en) | A kind of freezing of gait online test method | |
| EP3468450B1 (en) | Method and system for analyzing human gait | |
| Procházka et al. | Bayesian classification and analysis of gait disorders using image and depth sensors of Microsoft Kinect | |
| Polat | Freezing of gait (fog) detection using logistic regression in parkinson's disease from acceleration signals | |
| CN110659595A (en) | Tumble type and injury part detection method based on feature classification | |
| CN104091177A (en) | Abnormal gait detection method based on determined learning theory | |
| Jalloul et al. | Activity recognition using complex network analysis | |
| Hossain et al. | A direction-sensitive fall detection system using single 3D accelerometer and learning classifier | |
| KR20190105867A (en) | System and Method for Analyzing Foot Pressure Change and Gait Pattern | |
| Sampath Dakshina Murthy et al. | Gait-based person fall prediction using deep learning approach | |
| CN110598536A (en) | Falling detection method and system based on human skeleton motion model | |
| Aubol et al. | Foot contact identification using a single triaxial accelerometer during running | |
| Hasan et al. | Automated classification of autism spectrum disorders gait patterns using discriminant analysis based on kinematic and kinetic gait features | |
| CN110313918A (en) | A kind of gait phase recognition methods and system based on plantar pressure | |
| CN107019501B (en) | Remote tumble detection method and system based on genetic algorithm and probabilistic neural network | |
| Kelly et al. | An investigation into non-invasive physical activity recognition using smartphones | |
| CN114881079A (en) | Human body movement intention abnormity detection method and system for wearable sensor | |
| Ashwini et al. | Skeletal data based activity recognition system | |
| WO2008152402A1 (en) | Automatic discrimination of dynamic behaviour | |
| Saad et al. | Sensoring and features extraction for the detection of Freeze of Gait in Parkinson disease | |
| Vilas-Boas et al. | Supporting the assessment of hereditary transthyretin amyloidosis patients based on 3-D gait analysis and machine learning | |
| Martinelli et al. | Daily movement recognition for dead reckoning applications | |
| Naghavi et al. | Improving machine learning based detection of freezing of gait using data synthesis methods | |
| Nouredanesh et al. | Automated Detection of Older Adults’ Naturally-Occurring Compensatory Balance Reactions: Translation From Laboratory to Free-Living Conditions | |
| CN114913585A (en) | Household old man falling detection method integrating facial expressions |
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 |