CN109793645A - A kind of auxiliary patient Parkinson gait rehabilitation training device - Google Patents
A kind of auxiliary patient Parkinson gait rehabilitation training device Download PDFInfo
- Publication number
- CN109793645A CN109793645A CN201910052596.2A CN201910052596A CN109793645A CN 109793645 A CN109793645 A CN 109793645A CN 201910052596 A CN201910052596 A CN 201910052596A CN 109793645 A CN109793645 A CN 109793645A
- Authority
- CN
- China
- Prior art keywords
- gait
- training device
- parkinson
- patient
- training
- 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
Abstract
The present invention discloses a kind of auxiliary patient Parkinson gait rehabilitation training device, comprising: gait training device mainboard, lithium battery, infrared light transmitter elements, waist band, training program aid decision-making system.Wherein, lithium battery, infrared light transmitter elements are connect by route with gait training device mainboard, gait training device mainboard, lithium battery, infrared light transmitter elements are fixed on composition gait training device on waist band, gait training device is fixed on the rehabilitation training that patient's waist carries out gait；Training program aid decision-making system provides accurately gait rehabilitation training program using variation self-encoding encoder algorithm for patient, and scheme is forwarded to gait training device by way of Bluetooth communication.The present invention can be directed to different Kieren Perkins patients, provide the gait rehabilitation training program of specificity, and patient is helped to carry out gait rehabilitation training.
Description
Technical field
The present invention relates to medical device and system regions, in particular to a kind of auxiliary patient Parkinson gait rehabilitation training cartridge
It sets.
Background technique
Parkinson's disease (Parkinson ' s disease) is a kind of common nervous system degeneration disease, in the elderly
It is common.In China over-65s crowd, the illness rate of Parkinson's disease is about 1.7%.The most important pathological change of Parkinson's disease
It is the denaturation death of substantia nigra of midbrain dopaminergic neuron, striatal dopaminergic neuron content conspicuousness is thus caused to reduce
And it causes a disease.Cause the definite cause of disease of this pathological change still unclear at present, inherent cause, environmental factor, age ageing, oxidation
The denaturation death process that may participate in dopaminergic neuron stress be waited.
Parkinson's disease insidious onset, makes slow progress.Onset symptoms be usually side limbs tremble or activity is clumsy, in turn
Involve contralateral limbs.Clinically, static tremor, bradykinesia, myotonia and posture gait disorder are mainly shown as.Posture is anti-
It penetrates to disappear often and occur in the middle and advanced stage of disease, patient is not easy to maintain the balance of body, and the slightly road surface of out-of-flatness is possible to fall
?.Parkinsonian usually the more can walk the more fast when walking, and be not easy to halt, referred to as festinating gait.Advanced Parkinson patient can
There is freezeout, show as occurring briefly taking a step suddenly when walking, biped seems to be sticked on the ground, needs the several seconds of pausing
It can just be further continued for moving ahead or can not being again started up after clock.
Existing research shows that if there is " visual cues object " before patient, as the grid or zebra stripes on floor will be limited
The case where avoiding freezing of gait generation.Studied carefully based on this, it is clinical at present to have for Parkinsonian's festinating gait and freezing of gait
The method of a variety of rehabilitation trainings, such as using striation as " visual cues object " Lai Gaishan Parkinsonian of Parkinsonian
The freezing of gait problem occurred in daily life；Improve the flurried step of Parkinsonian using metronome and step song systematic training
State problem.
The art existing apparatus includes: induced with laser crutch, laser shoe etc..But there are certain defects.Laser lures
It is big, inconvenient to carry to lead crutch volume, and laser rays is not fixed body front position is opposite, is influenced by upper limb is movable；Laser
When having barrier in front of shoes, barrier can cause to block to laser rays, can not project correct position, and guidance is caused to lose
It loses.
In addition, the shortcomings that existing apparatus further include: the training program in device is single, can not according to patient's degree,
Patient age etc. formulates specific gait rehabilitation training program；The light effect of light beam in the dark is obvious, the secondary light source under strong light
Identification it is low etc..
Summary of the invention
Above-mentioned apparatus and method there are aiming at the problem that, it is trained that the present invention provides a kind of auxiliary patient's Parkinson gait rehabilitation
Device.
The present invention is realized with following technical solution: a kind of auxiliary patient Parkinson gait rehabilitation training device, including
Waist band, gait training device mainboard, lithium battery, infrared light transmitter elements and training program aid decision-making system；
The waist band, beam are each for placing patient's Parkinson gait rehabilitation training device in disturbances in patients with Parkinson disease waist
Building block；
The gait training device mainboard, is fixed on waist band, mentions for patient's Parkinson gait rehabilitation training device
For control and communication function；
The lithium battery is fixed on waist band, is connect with gait training device mainboard, is patient's Parkinson gait health
Multiple training device power supply；
The infrared light transmitter elements are fixed on waist band, are connect with gait training device mainboard, and gait health is generated
The experienced infrared light of refreshment；
The gait training device mainboard includes main processor modules, and the main processor modules are connected with voltage modulus of conversion
Block, bluetooth module, speech processing module, serial communication modular, function button module and on-off circuit module；
Training program aid decision-making system is realized by bluetooth and is communicated with gait training device mainboard, is mentioned for Kieren Perkins patient
For accurate gait rehabilitation training program.
Preferably, the main processor modules use STM32F407 chip, are the main control chips of gait training device, to
Each compositing chip issues dispatch command.
The voltage transformation module uses LM2596S-3.3V chip, LM2596S-3.3 chip and its peripheral drive circuit,
It realizes the voltage that the lithium battery voltage of 12V is converted to 3.3V, provides operating voltage for each chip of gait training device.
The bluetooth module uses NRF51822 chip, NRF51822 chip and its peripheral drive circuit, realizes training side
Communication between case aid decision-making system and gait training device, the specific gait that training program aid decision-making system is generated are instructed
Practice scheme and is forwarded to gait training device.
The speech processing module uses XFS5152CE chip, XFS5152CE chip and its peripheral drive circuit, realizes
Gait training device generates the function of voice, provides voice indication signal for the training of patient's Parkinson gait rehabilitation.
The serial communication modular uses MAX232CSE chip, MAX232CSE chip and its peripheral drive circuit, realizes
Communication between gait training device host processor chip STM32F407 and speech processing chip XFS5152CE is that data turn
Change chip.
The function button module is specially four road key circuits, and key K1 and K2 are the size function for adjusting speech volume
Can, wherein K1 is to tune up, and K2 is to turn down；Key K3 is switch function of Bluetooth communication；Key K4 is gait training device main process task
Device reset function.
The on-off circuit module uses transistor switching circuit, and the input of transistor circuit connects host processor chip
STM32F407, output termination infrared light transmitter elements, realizes the control to infrared light light on and off.
Preferably, the waist band is customization elastic straps, and elastic straps length is adjustable, there is fixator in elastic straps
The structure of part.
Preferably, the lithium battery capacity is 2800mAh, having a size of 56*22*67mm, weight 158g.
Preferably, the infrared light transmitter elements be red laser lamp device, assembly dia 12mm, length 40mm,
Output power is 5mw.
Preferably, the training program aid decision-making system specific workflow is as follows:
Firstly, realizing variation self-encoding encoder algorithm；
Then, it using patient's Parkinson gait rehabilitation training data training variation self-encoding encoder model of clinical statistics, obtains
To can according to input Parkinson's patient information, obtain the program bag of accurately specific treatment regimens；
Then, using the software systems of C# language exploitation training program aid decision-making system, and by housebroken variation
Self-encoding encoder model integrated obtains having the training for generating specific training program function into training program aid decision-making system
Scheme aid decision-making system；The information of patient Parkinson is inputted to the training program aid decision-making system for being mounted on computer end, this
When, training program aid decision-making system calls variation self-encoding encoder model according to the information of input, generates gait training scheme；
Finally, the training program aid decision-making system of computer end turns the gait training scheme of generation by blueteeth network
It is dealt into gait training device.
Preferably, patient's Parkinson information includes: age, gender, sick age, illness grade.
Preferably, the variation self-encoding encoder algorithm is to generate model algorithm, including three parts: encoder, priori, solution
Code device；Specific step is as follows:
(1) model by hidden variable Z generation target data X, variation self-encoding encoder are constructed using variation self-encoding encoder
Set hidden variable Z Normal Distribution；
(2) using Parkinson's patients clinical gait rehabilitation training sample of statistics, remember sample data are as follows: { X1,…,Xn,
Sample data integrally indicates that p (X) indicates the distribution of X, specially formula (1) with X, wherein setting hidden variable Z is obeying standard just
State distribution, i.e. p (Z)=N (0,1),
(3) setting Posterior distrbutionp p (Z | X) is normal distribution, for given sample Xk, there are one to be specific to X for settingk
Posterior distrbutionp p (Z | Xk), from p (Z | Xk) profile samples go out hidden variable Z, then hidden variable Z is reduced into Xk；
(4) it finds and is specific to XkNormal distribution p (Z | Xk) two groups of parameters: mean μ and variances sigma2, in the present invention two
Person is vector；
(5) two neural network μ are constructedk=f (Xk) and log σ2=f2(Xk) be fitted, calculating is specific to sample Xk's
Normal distribution p (Z | Xk) mean value and variance, acquire and be specific to sample XkMean value and variance i.e. obtain normal distribution, from normal state
A hidden variable Z is sampled in distributionk, utilize generatorIt obtainsThen generation is minimizedIt is right with its
The original sample X answeredkDifference, use formulaIt calculates；
(6) variation self-encoding encoder allows all p (Z | X) all to standard normal point during variation self-encoding encoder encodes
Cloth is dressed, and sampling can also generate data from standardized normal distribution N (0,1) in decoding process；
(7) variation self-encoding encoder model allows all p (Z | X) to be all to calculate each isolated component to the process that N (0,1) is dressed
KL divergence KL (N (μ, the σ of normal distribution and standardized normal distribution2) | | N (0,1)) between least disadvantage, least disadvantage is denoted as
Loss, as shown in formula (2)；
(8) from normal distribution N (μ, σ2) one hidden variable Z of middle sampling, it is equivalent to and is adopted from standardized normal distribution N (0,1)
One, sample value, indicates the value with ε, then utilize formula Z=μ+ε × σ, hidden variable Z is calculated, wherein mean μ and variance
Square root σ obtains for model training.
The invention has the advantages that device light source contrast is strong, ideal training effect can be reached under strong and weak luminous environment；Using waist
Portion's band is convenient for carrying and uses, and solves the problems, such as that light is influenced by upper limb activity；It can suffer from for different Parkinsons
Person provides the gait rehabilitation training program of specificity.
Detailed description of the invention
Fig. 1 is schematic structural view of the invention；
Fig. 2 is gait training device main board function block diagram of the invention；
Fig. 3 is gait training device motherboard circuit schematic diagram of the invention；
Fig. 4 is training program aid decision-making system flow chart of the invention；
Fig. 5 is variation self-encoding encoder model schematic of the invention；
Fig. 6 is variation self-encoding encoder model training procedure chart of the invention；
Fig. 7 variation self-encoding encoder sets hidden variable Z Normal Distribution figure.
Specific embodiment
The present invention is further described for explanation and specific embodiment with reference to the accompanying drawing.
As shown in Figure 1, a kind of auxiliary patient Parkinson gait rehabilitation training device, including waist band, gait training dress
Set mainboard, lithium battery, infrared light transmitter elements and training program aid decision-making system.Wherein, lithium battery, infrared light emission list
Member is connect by route with gait training device mainboard, and gait training device mainboard, lithium battery, infrared light transmitter elements are fixed
Gait training device is constituted on waist band, and gait training device is fixed on the rehabilitation training that patient's waist carries out gait；
Training program aid decision-making system is mounted in the software systems of computer end, and core algorithm is variation self-encoding encoder, according to input
Age of patient Parkinson, gender, sick age, illness grade, generate accurately specific gait rehabilitation training program, and pass through
The mode of Bluetooth communication is forwarded to gait training device, uses for patient Parkinson.
As shown in Fig. 2, the main processor modules use STM32F407 chip, the voltage transformation module is used
LM2596S-3.3V voltage conversion chip, the bluetooth module use NRF51822 chip, and the speech processing module uses
XFS5152CE chip, the serial communication modular use MAX232CSE chip, and the function button module is specially that four roads are pressed
Key circuit, the on-off circuit module use transistor switching circuit.The input terminal of voltage conversion chip LM2596S-3.3V connects
Lithium battery is connect, the voltage for 3.3V is exported, output is connect with host processor chip STM32F407；Function button K1, K2, K3 with
K4 is directly connect with host processor chip STM32F407；The input terminal and host processor chip of serial communication chip MAX232CSE
STM32F407 connection, output end are connect with speech chip XFS5152CE；The output end of speech chip XFS5152CE and loudspeaking
Device connection；The previous stage circuit of Bluetooth communication chip NRF51822 is connect with host processor chip STM32F407, rear stage circuit
It is connect with Bluetooth antenna；The input terminal of transistor switching circuit is connect with host processor chip STM32F407, output end with it is infrared
Light emitting unit laser connection.
As shown in figure 3, STM32F407 is host processor chip, STM32F407 is the Cortex-M4 kernel of ARM framework,
Support uCOS system.Pin VDD1, VDD2, VDD3, VDD4 of STM32F407 connect the voltage of 3.3V；Pin VSS1, VSS2,
VSS3, VSS4 meet GND；PD0, PD1 are R2OUT the and T2IN pin that serial communication pin meets MAX3232CSE respectively；PA6,
PA7 is SWDIO and the SWCLK pin that clock communication pin connects NRF51822 chip respectively；Pin PB12, PB13, PB14,
PB15 meets key K4, K3, K2, K1 respectively；Pin PA9 meets triode Q1 by resistance R11.
LM2596S-3.3V is that 12V turns 3.3V voltage conversion chip, and the input VCC of LM2596S-3.3V chip is to provide
The lithium battery of 12V voltage；The composition circuit of LM2596S-3.3V includes capacitor: C1, C2, C3, C4, resistance: R1, inductance: L1,
One-way conduction diode: D1；The output of LM2596S-3.3V chip is the DC voltage of 3.3V.
MAX232CSE is the bis- RS232 transmitters of CMOS and receiver, is serial communication chip, operating voltage 3.3V, structure
It include capacitor at circuit: C5, C6, C7, C8；Pin R2OUT, T2IN are connect with the PD0 of STM32F407 chip with PD1 respectively；Pipe
Foot R2IN, T2OUT are connect with the TXD of XFS5152CE chip with RXD respectively.
XFS5152CE is speech production chip, operating voltage 3.3V；Constitute circuit include capacitor: C9, C10, C11,
C12, C13, C14, C15, resistance: R2, R3, R4, R5, R6, R7, R8, R9, R17；Pin TXD, RXD respectively with MAX232CSE
R2IN connect with T2OUT.
NRF51822 is wireless blue tooth chip, operating voltage 3.3V；Constitute circuit include capacitor: C16, C17, C18,
C19, C21, C22, C23, resistance: R16, crystal oscillator X1；Pin SWDIO, SWCLK respectively with pin PA6, PA7 of STM32F407
Connection.
The composition circuit of four function buttons K1, K2, K3, K4 include resistance: R12, R13, R14, R15, respectively with
PB15, PB14, PB13, PB12 pin of STM32F407 chip connect.
The triode Q1 of transistor switching circuit is IN4148, and the composition of transistor switching circuit includes capacitor: C20, electricity
Resistance: R10, R11；The input of transistor switching circuit is connect by R11 with the pin PA9 of STM32F407 chip, triode switch
The output of circuit is connect with infrared light transmitter elements laser.
As shown in figure 4, realizing variation self-encoding encoder algorithm using python language in the case where pyCharm develops environment；It connects
, using patient's Parkinson gait rehabilitation training data training variation self-encoding encoder model of clinical statistics, obtaining being capable of basis
Parkinson's patient information is inputted, the program bag of accurately specific treatment regimens is obtained；Then, training side is developed using C# language
The software systems of case aid decision-making system, and by housebroken variation self-encoding encoder model integrated to training program aid decision
In system, obtain having the training program aid decision-making system for generating specific training program function；To being mounted on computer end
The information of training program aid decision-making system input patient Parkinson, comprising: age, gender, sick age, illness grade, at this point, instruction
Practice scheme aid decision-making system according to the information of input, calls variation self-encoding encoder model, generate gait training scheme；Finally,
By blueteeth network, the gait training scheme of generation is forwarded to gait training dress by the training program aid decision-making system of computer end
It sets, gait training device can execute specific gait training scheme, carry out gait rehabilitation training to patient Parkinson.
As shown in Figure 5 and Figure 6, (1) variation self-encoding encoder is made of three parts: encoder, priori, decoder.The present invention
A model by hidden variable Z generation target data X is constructed using variation self-encoding encoder, variation self-encoding encoder sets hidden variable Z
Normal Distribution is as shown in Figure 7.
(2) using Parkinson's patients clinical gait rehabilitation training sample of statistics, remember sample data are as follows: { X1,…,Xn,
Sample data integrally indicates that p (X) indicates the distribution of X, specially formula (1) with X, wherein setting hidden variable Z is obeying standard just
State distribution, i.e. p (Z)=N (0,1),
(3) setting Posterior distrbutionp p (Z | X) is normal distribution, for given sample Xk, there are one to be specific to X for settingk
Posterior distrbutionp p (Z | Xk), from p (Z | Xk) profile samples go out hidden variable Z, then hidden variable Z is reduced into Xk；
(4) it finds and is specific to XkNormal distribution p (Z | Xk) two groups of parameters: mean μ and variances sigma2, in the present invention two
Person is vector；
(5) two neural network μ are constructedk=f (Xk) and log σ2=f2(Xk) be fitted, calculating is specific to sample Xk's
Normal distribution p (Z | Xk) mean value and variance, acquire and be specific to sample XkMean value and variance i.e. obtain normal distribution, from normal state
A hidden variable Z is sampled in distributionk, utilize generatorIt obtainsThen generation is minimizedIt is right with its
The original sample X answeredkDifference, use formulaIt calculates；
(6) variation self-encoding encoder allows all p (Z | X) all to standard normal point during variation self-encoding encoder encodes
Cloth is dressed, and preventing the noise in generating process is 0, guarantees the generative capacity of model, this makes it possible to the priori before reaching point
Cloth p (Z)=N (0,1) sampling can also generate data from standardized normal distribution N (0,1) in decoding process, and variation encodes certainly
Device model schematic is as described in Figure 5.
(7) variation self-encoding encoder model allows all p (Z | X) to be all to calculate each isolated component to the process that N (0,1) is dressed
KL divergence KL (N (μ, the σ of normal distribution and standardized normal distribution2) | | N (0,1)) between least disadvantage, least disadvantage is denoted as
Loss, as shown in formula (2)；
(8) from normal distribution N (μ, σ2) one hidden variable Z of middle sampling, it is equivalent to and is adopted from standardized normal distribution N (0,1)
One, sample value, indicates the value with ε, then utilize formula Z=μ+ε × σ, hidden variable Z is calculated, wherein mean μ and variance
Square root σ obtains for model training.
The present invention improves light source generator, redesigns each funtion part of the device of exploitation gait rehabilitation training, especially
It is to can effectively solve the problem of light is influenced by upper limb activity, using strong using fixed structure of the waist band as device
Contrast light source, so that secondary light source is attained by ideal effect under circumstances.Furthermore the invention also includes training programs
Aid decision-making system, the system can formulate the gait health of specificity according to information such as age, the medical histories of different disturbances in patients with Parkinson disease
Multiple training program.
Claims (8)
1. a kind of auxiliary patient Parkinson gait rehabilitation training device, it is characterised in that: including waist band, gait training device
Mainboard, lithium battery, infrared light transmitter elements and training program aid decision-making system；
The waist band, beam are respectively formed in disturbances in patients with Parkinson disease waist for placing patient's Parkinson gait rehabilitation training device
Component；
The gait training device mainboard, is fixed on waist band, provides control for patient's Parkinson gait rehabilitation training device
System and communication function；
The lithium battery is fixed on waist band, is connect with gait training device mainboard, is instructed for patient's Parkinson gait rehabilitation
Practice device power supply；
The infrared light transmitter elements are fixed on waist band, are connect with gait training device mainboard, and gait rehabilitation instruction is generated
Experienced infrared light；
The gait training device mainboard includes main processor modules, the main processor modules be connected with voltage transformation module,
Bluetooth module, speech processing module, serial communication modular, function button module and on-off circuit module；
Training program aid decision-making system realizes by bluetooth and communicates with gait training device mainboard that patient provides essence for Kieren Perkins
Quasi- gait rehabilitation training program.
2. a kind of auxiliary patient Parkinson gait rehabilitation training device according to claim 1, it is characterised in that: the master
Processor module uses STM32F407 chip, and the voltage transformation module uses LM2596S-3.3V voltage conversion chip, described
Bluetooth module uses NRF51822 chip, and the speech processing module uses XFS5152CE chip, and the serial communication modular is adopted
With MAX232CSE chip, the function button module is specially four road key circuits, and the on-off circuit module uses triode
Switching circuit.
3. a kind of auxiliary patient Parkinson gait rehabilitation training device according to claim 1, it is characterised in that: the waist
Portion's band is customization elastic straps.
4. a kind of auxiliary patient Parkinson gait rehabilitation training device according to claim 1, it is characterised in that: the lithium
Battery capacity is 2800mAh, having a size of 56*22*67mm, weight 158g.
5. a kind of auxiliary patient Parkinson gait rehabilitation training device according to claim 1, it is characterised in that: described red
Outer light emitting unit is red laser lamp device, assembly dia 12mm, length 40mm, output power 5mw.
6. a kind of auxiliary patient Parkinson gait rehabilitation training device according to claim 1, it is characterised in that: the instruction
It is as follows to practice scheme aid decision-making system specific workflow:
Firstly, realizing variation self-encoding encoder algorithm；
Then, using patient's Parkinson gait rehabilitation training data training variation self-encoding encoder model of clinical statistics, energy is obtained
Enough according to input Parkinson's patient information, the program bag of accurately specific treatment regimens is obtained；
Then, the software systems of training program aid decision-making system are developed using C# language, and housebroken variation is self-editing
Code device model integrated obtains having the training program for generating specific training program function into training program aid decision-making system
Aid decision-making system；The information of patient Parkinson is inputted to the training program aid decision-making system for being mounted on computer end, at this point, instruction
Practice scheme aid decision-making system according to the information of input, calls variation self-encoding encoder model, generate gait training scheme；
Finally, by blueteeth network, the gait training scheme of generation is forwarded to by the training program aid decision-making system of computer end
Gait training device.
7. a kind of auxiliary patient Parkinson gait rehabilitation training device according to claim 6, it is characterised in that: Parkinson
Patient information includes: age, gender, sick age, illness grade.
8. a kind of auxiliary patient Parkinson gait rehabilitation training device according to claim 6, it is characterised in that: the change
Dividing self-encoding encoder algorithm is to generate model algorithm, including three parts: encoder, priori, decoder；Specific step is as follows:
(1) model by hidden variable Z generation target data X, the setting of variation self-encoding encoder are constructed using variation self-encoding encoder
Hidden variable Z Normal Distribution；
(2) using Parkinson's patients clinical gait rehabilitation training sample of statistics, remember sample data are as follows: { X1,…,Xn, sample number
It being indicated according to entirety with X, p (X) indicates the distribution of X, specially formula (1), wherein setting hidden variable Z obeys standardized normal distribution,
That is p (Z)=N (0,1),
(3) setting Posterior distrbutionp p (Z | X) is normal distribution, for given sample Xk, there are one to be specific to X for settingkAfter
Test distribution p (Z | Xk), from p (Z | Xk) profile samples go out hidden variable Z, then hidden variable Z is reduced into Xk；
(4) it finds and is specific to XkNormal distribution p (Z | Xk) two groups of parameters: mean μ and variances sigma2, the two is equal in the present invention
For vector；
(5) two neural network μ are constructedk=f (Xk) and log σ2=f2(Xk) be fitted, calculating is specific to sample XkNormal state
Distribution p (Z | Xk) mean value and variance, acquire and be specific to sample XkMean value and variance i.e. obtain normal distribution, from normal distribution
One hidden variable Z of middle samplingk, utilize generatorIt obtainsThen generation is minimizedIt is corresponding with its
Original sample XkDifference, use formulaIt calculates；
(6) variation self-encoding encoder allows all p (Z | X) all to see to standardized normal distribution during variation self-encoding encoder encodes
Together, sampling data can also be generated from standardized normal distribution N (0,1) in decoding process；
(7) variation self-encoding encoder model allows all p (Z | X) to be all to calculate each isolated component normal state to the process that N (0,1) is dressed
KL divergence KL (N (μ, the σ of distribution and standardized normal distribution2) | | N (0,1)) between least disadvantage, least disadvantage is denoted as loss,
As shown in formula (2)；
(8) from normal distribution N (μ, σ2) one hidden variable Z of middle sampling, it is equivalent to the sampling one from standardized normal distribution N (0,1)
Value, indicates the value with ε, then utilizes formula Z=μ+ε × σ, hidden variable Z is calculated, wherein the square root σ of mean μ and variance
It is obtained for model training.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910052596.2A CN109793645B (en) | 2019-01-21 | 2019-01-21 | Supplementary recovered trainer of parkinsonism people gait |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910052596.2A CN109793645B (en) | 2019-01-21 | 2019-01-21 | Supplementary recovered trainer of parkinsonism people gait |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109793645A true CN109793645A (en) | 2019-05-24 |
| CN109793645B CN109793645B (en) | 2021-07-13 |
Family
ID=66559826
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910052596.2A Active CN109793645B (en) | 2019-01-21 | 2019-01-21 | Supplementary recovered trainer of parkinsonism people gait |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109793645B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111899844A (en) * | 2020-09-28 | 2020-11-06 | 平安科技（深圳）有限公司 | Sample generation method and device, server and storage medium |
| CN114366557A (en) * | 2021-12-31 | 2022-04-19 | 华南理工大学 | Man-machine interaction system and method for lower limb rehabilitation robot |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW201503882A (en) * | 2013-07-18 | 2015-02-01 | Stone & Resource Ind R & D Ct | Indicating device and method for gait |
| CN105534678A (en) * | 2015-12-02 | 2016-05-04 | 华馨伊 | Rehabilitation training system based on internet data management |
| CN205251971U (en) * | 2015-10-19 | 2016-05-25 | 陈新红 | Guide stick that guide parkinson disease and dyskinesia patient walked |
| CN106693280A (en) * | 2016-12-29 | 2017-05-24 | 深圳市臻络科技有限公司 | Virtual-reality-based Parkinsonism training method, system and device |
| JP2017148594A (en) * | 2017-05-08 | 2017-08-31 | 有限会社ホームケア渡部建築 | Walking assist device |
| CN107485844A (en) * | 2017-09-27 | 2017-12-19 | 广东工业大学 | A kind of limb rehabilitation training method, system and embedded device |
| CN108098736A (en) * | 2016-11-24 | 2018-06-01 | 广州映博智能科技有限公司 | A kind of exoskeleton robot auxiliary device and method based on new perception |
| CN108392795A (en) * | 2018-02-05 | 2018-08-14 | 哈尔滨工程大学 | A kind of healing robot Multimode Controlling Method based on Multi-information acquisition |
| CN110300542A (en) * | 2016-07-25 | 2019-10-01 | 开创拉布斯公司 | Use the method and apparatus of wearable automated sensor prediction muscle skeleton location information |
-
2019
- 2019-01-21 CN CN201910052596.2A patent/CN109793645B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW201503882A (en) * | 2013-07-18 | 2015-02-01 | Stone & Resource Ind R & D Ct | Indicating device and method for gait |
| CN205251971U (en) * | 2015-10-19 | 2016-05-25 | 陈新红 | Guide stick that guide parkinson disease and dyskinesia patient walked |
| CN105534678A (en) * | 2015-12-02 | 2016-05-04 | 华馨伊 | Rehabilitation training system based on internet data management |
| CN110300542A (en) * | 2016-07-25 | 2019-10-01 | 开创拉布斯公司 | Use the method and apparatus of wearable automated sensor prediction muscle skeleton location information |
| CN108098736A (en) * | 2016-11-24 | 2018-06-01 | 广州映博智能科技有限公司 | A kind of exoskeleton robot auxiliary device and method based on new perception |
| CN106693280A (en) * | 2016-12-29 | 2017-05-24 | 深圳市臻络科技有限公司 | Virtual-reality-based Parkinsonism training method, system and device |
| JP2017148594A (en) * | 2017-05-08 | 2017-08-31 | 有限会社ホームケア渡部建築 | Walking assist device |
| CN107485844A (en) * | 2017-09-27 | 2017-12-19 | 广东工业大学 | A kind of limb rehabilitation training method, system and embedded device |
| CN108392795A (en) * | 2018-02-05 | 2018-08-14 | 哈尔滨工程大学 | A kind of healing robot Multimode Controlling Method based on Multi-information acquisition |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111899844A (en) * | 2020-09-28 | 2020-11-06 | 平安科技（深圳）有限公司 | Sample generation method and device, server and storage medium |
| WO2021159740A1 (en) * | 2020-09-28 | 2021-08-19 | 平安科技（深圳）有限公司 | Sample generation method and apparatus, and server and storage medium |
| CN114366557A (en) * | 2021-12-31 | 2022-04-19 | 华南理工大学 | Man-machine interaction system and method for lower limb rehabilitation robot |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109793645B (en) | 2021-07-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Young et al. | A classification method for user-independent intent recognition for transfemoral amputees using powered lower limb prostheses | |
| CN104606868B (en) | A kind of Intelligent bracelet for alleviating Parkinsonian's freezing of gait | |
| WO2017023864A1 (en) | Systems, devices, and method for the treatment of osteoarthritis | |
| CN204426918U (en) | A kind of Intelligent bracelet for alleviating Parkinsonian's freezing of gait | |
| CN109793645A (en) | A kind of auxiliary patient Parkinson gait rehabilitation training device | |
| CN103356160A (en) | Blink detection system for electronic ophthalmic lens | |
| CN106267694B (en) | Fracture of lower limb rehabilitation system and its control method | |
| CN1846805B (en) | Functional electric stimulation system and method | |
| US11426098B2 (en) | System and method for gait monitoring and improvement | |
| CN109567812A (en) | Gait analysis system based on Intelligent insole | |
| CN111588597A (en) | Intelligent interactive walking training system and implementation method thereof | |
| CN110338952B (en) | Device for training normal gait and application thereof | |
| CN107411750A (en) | A kind of intelligence step appearance antidote and system | |
| CN205507686U (en) | Wearable device of foot with virtual reality control function | |
| CN205126247U (en) | Intelligence shoe -pad with gait analysis function | |
| CN207804991U (en) | Postural training correcting device | |
| Lee et al. | Development of a novel 2-dimensional neck haptic device for gait balance training | |
| CN110102037A (en) | Anti- Prevention of fall Sex Rehabilitation system and method applied to patients with vertigo rehabilitation | |
| Figueiredo et al. | Instrumented insole system for ambulatory and robotic walking assistance: First advances | |
| CN108703848A (en) | A kind of Intelligent low-oxygen health care system based on physical training scheme | |
| CN205409871U (en) | Be used for disinfecting and heat retaining intelligent health shoes | |
| Sun et al. | Programmable neural processing on a smartdust for brain-computer interfaces | |
| CN114529985A (en) | Intention identification method for daily complex terrain movement of old people | |
| CN209361244U (en) | A kind of five degree of freedom information exchange ankle rehabilitation parallel robot | |
| CN209376807U (en) | One kind being based on pressure sensing intelligent shoe |
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 |