CN109717833A - A kind of neurological disease assistant diagnosis system based on human motion posture - Google Patents
A kind of neurological disease assistant diagnosis system based on human motion posture Download PDFInfo
- Publication number
- CN109717833A CN109717833A CN201811416598.7A CN201811416598A CN109717833A CN 109717833 A CN109717833 A CN 109717833A CN 201811416598 A CN201811416598 A CN 201811416598A CN 109717833 A CN109717833 A CN 109717833A
- Authority
- CN
- China
- Prior art keywords
- data
- gait
- variance
- mean value
- human motion
- 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.)
- Pending
Links
Landscapes
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The present invention discloses a kind of neurological disease assistant diagnosis system based on human motion posture, belong to intelligent medical field, by the way that the athletic posture of examiner is quantified, 23 dimension gait correlated characteristics are extracted from human motion attitude data, it is input in classification prediction model, to diagnose to examiner, the result diagnosed to examiner is generated into visualization motor function audit report, and the suggestion that provides assistance in diagnosis.
Description
Technical field
The invention belongs to intelligent medical fields, and in particular to a kind of neural aided disease diagnosis based on human motion posture
System.
Background technique
In recent years, the trend (bibliography: " 2015 risen year by year is being presented in 60 years old of China and the above size of population
Year community service statistical communique of development " .2016.).Therefore, parkinsonism, cerebral small vessels disease etc. are apt to occur in the elderly
Cerebral nervous system disease increasingly receive social concerns.It is main with the clinic for the neurodegenerative disease that Parkinson's disease (PD) is representative
Want feature for dyskinesia, including progressive bradykinesia, myotonia, static tremor and posture abnormal gait etc., additionally
It may be with a large amount of non-motor symptoms (NMS), such as hyposphresia, constipation, depression, sleep disturbance.With the progress of the course of disease, fortune
Dynamic symptom and non-motor symptoms gradually aggravate, until often there is motor complication in end-stage disease, including curative effect of medication decline, " open-
Close " phenomenon, unusual fluctuation disease etc..End-stage disease examiner is often because of disequilibrium, tumble, freezing of gait, dysphagia and aphasis
Etc. causing life that can not take care of oneself or even long-term bed, quality of life degradation (bibliography: Liu Shuying and Chen Biao " pa gold
Gloomy disease prevalence situation " contemporary Chinese neurological disease magazine .2016,16 (2), 98-101.).The full prevalence of Parkinson's disease is about
For 0.3% (bibliography: de Lau, L.M., and M.M.Breteler. " Epidemiology of Parkinson's
disease."Neurologic Clinics.2007,5(6),525-535.).As a kind of typical senile chronic disease, pa
Gold it is gloomy disease in elderly population illness rate be multiplied, over-65s elderly population illness rate be 1%~2%, 85 years old the above are
3%~5% [4].And full age bracket disease incidence be 8~18/,100,000 man-years, 50/,100,000 man-year of over-65s age bracket, 75 years old with
Upper age bracket is 1,50/,100,000 man-years, 85 years old or more age bracket is 4,00/,100,000 man-years (bibliography: de Lau, L.M., and
M.M.Breteler."Epidemiology of Parkinson's disease."Neurologic Clinics.2007,5
(6),525-535.).According to age heaping incidence it is found that the risk that 60 years old the elderly suffered from Parkinson's disease at 80 years old is about
2.5% (bibliography: Elbaz, A, et al. " Risk tables for parkinsonism and Parkinson's
Disease. " Journal of Clinical Epidemiology.2002,55 (1), 25-31.), and the early stage of Parkinson's disease
Diagnosis can control to the state of an illness and alleviation play an important role.Clinically mainly pass through the method pair of scale et al. work marking
The nervous function of patient is evaluated, while being equipped with the inspections such as PET and further being made a definite diagnosis.But these methods are dependent on doctor's
The inspection equipment of professional knowledge and valuableness.
To solve the above-mentioned problems, researchers largely grind to the detection of the nervous system disease and auxiliary diagnosis
Study carefully.The designs such as Jiang Chao word realize one and check system based on the electronic cognitive function of wacom Digitizing plate and sound pick-up outfit, can
By the interaction data of user's input, data under voice gets off, and (bibliography: Jiang Chao word " one is towards the nervous system disease
Multichannel cognitive function check system research " .Diss. Chinese Academy of Sciences university .2014.)；Wherein, the data that pen interacts are not
Only include fine trace information, also includes pressure, inclination angle, the corner etc. of pen.Examiner with nervous function disease
Also with the obstacle of somatic movement function, therefore, the body of user is obtained using means such as motion sensor or Image Acquisition
Motion state is analyzed.Method based on gait also has for neurological disease diagnostic assistance.Document (bibliography: Wren,
T.A.,et al. "Differences in implementation of gait analysis recommendations
based on affiliation with a gait laboratory."Gait&Posture.2013,37(2),206-
209.) reliability of three-dimensional gait analysis method is studied, and experimental verification has been carried out to 30 healthy adults.
For the examiner of apoplexy sequelae, the gait parameter that researcher utilizes gait scale in Wisconsin to obtain, to parameter when
Empty consistency is studied (bibliography: Guzik, Agnieszka, et al. " Analysis of consistency
between temporospatial gait parameters and gait assessment with the use of
Wisconsin Gait Scale in post-stroke patients."Neurologia I Neurochirurgia
Polska.2017, 51(1),60-65.).And it is directed to Parkinson examiner, there is researcher using the method for determining study to it
Gait feature is classified (bibliography: Wei, Zeng, et al. " Parkinson's disease
classification using gait analysis via deterministic learning."Neuroscience
Letters.2016,663,268-278.).Four kind gait analysis models of the researchers such as Hans Kainz to medical institutions' outpatient service
Reliability carried out comparative analysis (bibliography: Kainz, H, et al. " Reliability of four models for
clinical gait analysis."Gait&Posture.2017,54,325.).Document (bibliography: Song, Yin,
et al.On Discovering the Correlated Relationship between Static and Dynamic
Data in Clinical Gait Analysis.Machine Learning and Knowledge Discovery in
Databases.Springer Berlin Heidelberg.2013,8190,563-578.) in, researcher proposes a kind of needle
To the probability likelihood model of the correlation of static data and behavioral characteristics in gait analysis.Belinda Bilney etc. is utilized
GAITRite gait acquisition system (bibliography: Givon, U, G.Zeilig, and A.Achiron. " Gait analysis
in multiple sclerosis:Characterization of temporal–spatial parameters using
GAITRite functional ambulation system. " Gait&Posture.2009,29 (1), 138-142.),
Herman, T. etc. use the quick insole of power (bibliography: Herman, T, et al. " Gait instability and fractal
dynamics of older adults with a"cautious"gait:why do certain older adults
Walk fearfully? " Gait&Posture.2005,21 (2), 178-185.), Macko, R.F. and Pearson, O.R. etc.
With the motion sensor (bibliography: Macko, R.F., et al. " Microprocessor-based for being attached to foot
ambulatory activity monitoring in stroke patients."Medicine&Science in
Sports&Exercise.2002,34 (3), 94-9.) (bibliography: Pearson, O.R., et al. " Quantification
of walking mobility in neurological disorders."Qjm Monthly Journal of the
Association of Physicians.2004,97 (8), 463.), Gabel, M. etc. are acquired using Kinect with visual manner
The parameters such as the rhythm and pace of moving things, leg speed, the stride of walking process are used to analyze limb motion situation and contacting for the nervous system disease (refers to text
It offers: Gabel, M, et al. " Full body gait analysis with Kinect. " Engineering in
Medicine& Biology Society Conf Proc IEEE Eng Med Biol Soc.2012,2012(4),
1964.)。
Machine learning and visual technology also apply among disease detection and auxiliary diagnosis in large quantities.H.R.Roth etc.
Detection (the bibliography: Roth, Holger R., et al. " A New of lymph node has been carried out using convolutional neural networks technology
2.5D Representation for Lymph Node Detection Using Random Sets of Deep
Convolutional Neural Network Observations."Med Image Comput Comput Assist
Interv.2014,17 (1), 520-527.), H. Shin etc. devises the CNN net for computer aided detection medical image
Network structure and transfer learning method (bibliography: Hoochang, Shin, et al. " Deep Convolutional Neural
Networks for Computer-Aided Detection:CNN Architectures,Dataset
Characteristics and Transfer Learning."IEEE Transactions on Medical
Imaging.2016,35(5),1285-1298.).S.Kiranyaz et al. is using convolutional neural networks to the difference of electrocardiosignal
Pulse frequency type is classified (bibliography: Kiranyaz, Serkan, T.Ince, and M.Gabbouj. " Real-Time
Patient-Specific ECG Classification by 1-D Convolutional Neural Networks."
IEEE Transactions on Biomedical Engineering.2016,63 (3), 664-675.), it is led by Wu Enda
Machine learning group, Stanford University, develop a kind of new deep learning algorithm, the rhythm of the heart that can diagnose 14 seed types loses
Often (bibliography: Rajpurkar, Pranav, et al. " Cardiologist-Level Arrhythmia Detection
with Convolutional Neural Networks."2017.).(bibliography: the " such as Chen Yuke are based on GPUs can for document
Depending on heart assistance diagnostic system research " health care equipment .2011,32 (10) of change technology, 16-18.) it then realizes and is based on
The Accurate Segmentation and three-dimensional visualization of the cardiac tomogram image of GPUs, complete the design of heart assistance diagnostic system.Document (ginseng
Examine document: Zhang Yu " lung CT image blood vessel segmentation algorithm is studied with three-dimensional visualization " .Diss. Northeastern University .2011.) right
On the basis of blood vessel segmentation and three-dimensional reconstruction theory analysis, in conjunction with image segmentation algorithm tool ITK and visualization toolkit
VTK devises a kind of blood vessel segmentation platform, allows and carry out three-dimensional reconstruction and human-computer interaction to segmentation result, and achieve
Good effect of visualization.In document (bibliography: Xiang Nan " head of pancreas and the flat conjunction structure of pancreatic neoplasms three-dimensional visualization diagnosis and treatment
Build and clinical application research " .Diss. Nanfang Medical Univ .2016.) in researcher demonstrate with MI-3DVS system, use
Three-dimensional visualization instructs the lower continuous pancreas end to side intestinal anastomosis of single line simple and easy to do, damages less to pancreatic tissue, meets pancreatic tissue
Anatomical features, suitable for all residual pancreas situations and solid and reliable, hence it is evident that reduce the incidence of postoperative pancreatic fistula.Document (ginseng
Examine document: Wang Shengjun " coronary artery computer-aided diagnosis system Core Technology Research and realization based on heart CTA volume data "
.Diss. Northeastern University .2011.) it proposes a kind of spinning polygon and writes music face reconstruction technique, so that preferably adjuvant clinical works
Person observes tiny disease, to reduce the rate of missed diagnosis and misdiagnosis rate of doctor.In terms of the nervous system disease, Li Yongming et al. is utilized
Language sample and random forest grader devise the diagnosis algorithm (bibliography: Li, Y., et al. " of Parkinson's disease
[Research on Diagnosis Algorithm of Parkinson’s Disease Based on Speech
Sample Multi-edit and Random Forest]."Journal of Biomedical Engineering.2016,
33(6),1053-1059.)。
Summary of the invention
The purpose of the present invention is to propose to a kind of neurological disease assistant diagnosis system based on human motion posture, acquisition is checked
The human body attitude exercise data of person simultaneously quantifies, and extracts gait correlated characteristic, is input in trained classification prediction model,
Examiner is diagnosed, and provides audit report.
To achieve the goals above, the invention adopts the following technical scheme:
A kind of neurological disease assistant diagnosis system based on human motion posture, comprising:
One data acquisition module, for acquiring the human motion attitude data of examiner；
One data preprocessing module, for being pre-processed to human motion attitude data；
One characteristic extracting module, for being mentioned from the human body attitude feature that pretreated human motion attitude data includes
Take multidimensional gait correlated characteristic；
One classification prediction module, including the classification prediction model for gait correlated characteristic, which is used
It is input in the classification prediction model in by the multidimensional gait correlated characteristic, to be diagnosed to examiner；
One audit report generation module, for visualization motor function inspection report will to be generated to the result that examiner diagnoses
It accuses.
Further, the data acquisition module acquires human motion attitude data, including coloured silk by a depth camera
Color information flow, depth information stream, three kinds of flow datas of bone information stream, i.e. source data.
Further, the action data of four groups of inspections of data collecting module collected examiner movement, four groups of inspections
Movement includes hand function and lower limb test, seat stands up, erect position balances, walk test.
Further, it is described pretreatment include data are compressed, except make an uproar, smoothing processing.Wherein, the compress technique
Solve the problems, such as data storage, it is described that the interference of extraneous factor is eliminated with smoothing technique except making an uproar, improve the quality of data.
Further, the smoothing processing, which refers to, carries out uniform interleave processing to the non-uniform video of frame per second, makes its holding
30 frame per second, while mirror image switch is carried out to image.
Further, when the data preprocessing module pre-processes data, the timing convolution mould of triple channel is constructed
Plate, is slided in timing by the template and real-time perfoming convolutional calculation is generated new timing and believed by Gaussian smoothing
Breath reduces white Gaussian noise.
Further, the classification prediction model is related to the multidimensional gait special using linear discriminent analysis (LDA)
Sign carries out dimension-reduction treatment.
Further, the classification prediction model passes through training data training based on the model of machine learning algorithm by a kind of
It obtains, which is the extracted 23 dimension gait correlated characteristic of human motion attitude data gathered in advance.
Further, the multidimensional gait correlated characteristic is 23 dimension gait correlated characteristics, the 23 dimension gait correlated characteristic packet
Include leg speed, right leg speed mean value, right leg speed variance, left leg speed mean value, left leg speed variance, Zhou Qi Walk speed mean value, Zhou Qi Walk speed variance,
Right step-length mean value, right step-length variance, Zuo Buchang mean value, Zuo Buchang variance, left and right step size coordinating, the high mean value of right step, right step are high
Variance, Zuo Bugao mean value, Zuo Bugao variance, left and right walk high harmony, step width mean value, step width variance, step pitch mean value, step pitch side
Difference, upper body part part Z angle mean value, upper body part divide Z angle variance.
Further, the model based on machine learning algorithm is the vector machine (SVM supported based on gaussian kernel function
(RBF-kernel)), gradient boosted tree (GBDT), K- neighbour (KNN), random forest (RF), logistical regression (LR) this five
One of them model of kind of algorithm.
Further, two classification based trainings are carried out to the model based on above-mentioned five kinds of algorithms, using the side of 10 folding cross validations
Formula obtains the classification accuracy (accuracy) of each model, accuracy (precision), recall rate (recall), F1 degree
Amount, ROC area under a curve (AUC), in conjunction with the generalization ability of each algorithm, selection is suitable for human motion attitude data collection
Best model is as classification prediction model.
Further, the visualization motor function audit report is directed to three meter Bu Hang, refers to that nose test, lower limb harmony are surveyed
It tries these three inspections movement and draws correlation graph, and finally giving the suggestion of an auxiliary diagnosis in the form of text.
The deep video of present invention human-computer interaction technology and the extremity motor function of equipment acquisition examiner, therefrom will inspection
The athletic posture for the person of looking into is quantified, and provides data supporting for further analysis and diagnosis.Then it is extracted from initial data
Characteristic feature, and calculated using the machine learning such as support vector machines, logistical regression, random forest, K- neighbour, gradient boosted tree
Method is trained, and constructs the classification prediction model for the suggestion that can provide assistance in diagnosis for doctor.Finally by the fortune function function of examiner
Energy data visualization forms a more complete motor function audit report, hospital is facilitated to be put on record and consulted.
Compared with prior art, the present invention has the advantage that as follows with good effect: the present invention is based on computer visions
Method collects human body attitude data, does not need examiner and dresses additional equipment；The present invention is extracted human body fortune according to examiner
Dynamic posture feature, and classification prediction model is established using wherein 23 dimension gait correlated characteristics, to identify neurological disease, and to this
A little features carry out pathological analysis；Visual analyzing is carried out to the motor function quantized data of the inspection movement of emphasis in the present invention,
A motor function audit report is formed, enables a physician to more intuitively obtain inspection result, is analyzed and examined convenient for it
It is disconnected.
Detailed description of the invention
Fig. 1 is a kind of neurological disease assistant diagnosis system structural schematic diagram based on human motion posture of embodiment.
Fig. 2 is motor function overhaul flow chart.
Fig. 3 is the timing Gaussian convolution template schematic diagram for three-dimensional coordinate.
Fig. 4 is that visualization auxiliary diagnosis reports schematic diagram.
Specific embodiment
To enable features described above and advantage of the invention to be clearer and more comprehensible, special embodiment below, and institute's attached drawing is cooperated to make
Detailed description are as follows.
The present embodiment provides a kind of neurological disease auxiliary diagnostic equipments based on human motion posture, and the apparatus structure is as schemed
Shown in 1, it is described as follows:
1) data acquisition module, for acquiring the human motion attitude data of examiner；
2) data preprocessing module, for being pre-processed to above-mentioned data；
3) characteristic extracting module, for being mentioned from the human body attitude feature that pretreated human motion attitude data includes
Take 23 dimension gait correlated characteristics；
4) classify prediction module, it is trained for gait correlated characteristic for being input to 23 dimension gait correlated characteristics
Classify in prediction model, the person that comes deagnostic test；
5) audit report generation module, for generating visualization motor function audit report according to modal analysis results.
In step 1), the human body attitude data acquisition acquires four groups based on traditional motor function evaluation charter
Inspection movement action data, including hand function and lower limb test, seat stands up, erect position balances, walk test.It is specifically examined
It is as shown in Figure 2 to look into process.Wherein, the human body attitude data of acquisition include colour information stream, depth information stream, bone information stream three
Kind flow data.
In step 2), since the video that depth camera is recorded contains above-mentioned a large amount of information, therefore its depositing of occupying
It is larger to store up volume, lossless storage information above one second, about needs 250 Mbytes of memory space.Therefore, the present invention is to volume
Colour information in huge source XEF format video information, which carries out dump, becomes small volume, and more general AVI format letter
Breath.And in this course, uniform interleave has been carried out to the non-uniform situation of frame per second to handle, it is made to keep 30 frame per second, guaranteed
The play quality of video, while mirror image switch also has been carried out to image, so that it is more in line with the observation habit of doctor.
Due in the acquisition recording process of data, being limited to ground, often equally it is difficult to ensure only in camera coverage
The bone information of one people of examiner occurs.Therefore, movement of the present invention using target examiner when carrying out four groups of inspections and acting
The feature short time depth information of stage (such as three meters change), the relative coordinate of subject present position and camera optical center with
And the confidence level of each artis of personage's bone information stream is filtered the bone information of irrelevant personnel, so that it is determined that target
Examiner.
Since the artis that the feature calculation of the classification prediction model is mainly based in bone information stream is imaging
Three dimensional space coordinate in machine coordinate system.And bone information stream is the timing information of 30 frame per second, needs to construct a time window
Mouthful, and consider the Gaussian smoothing situation of three dimensions.Therefore, the present invention constructs timing convolution mask (such as Fig. 3 of triple channel
It is shown), which slides in timing and real-time perfoming convolutional calculation, generates the timing information of new skeletal joint point, the letter
Breath is the three-dimensional coordinate information after Gaussian smoothing, and the operation is so that white Gaussian noise in original bone body joint point coordinate information
Substantially reduce.
It in step 3), is acted for four groups of above-mentioned inspections, in conjunction with kinematics and statistics parameter, obtains 68 dimension human bodies
Athletic posture feature.The present invention is the significance analysis that each feature carries out feature, is examined using generalized linear model to 475
It looks into data and carries out multiplicity, and the age and gender of examiner are adjusted, wherein being diagnosed as the nervous system disease
(CNS disorders) is 72, and control group is 303,23 dimension gait correlated characteristics is therefrom extracted, as being input to
The data of classification and Detection model, the 23 dimension gait correlated characteristic are as shown in table 1.
Reason that there are two 23 dimension gait correlated characteristics of selection: first is that, gait feature is significant in all correlated characteristics
Property it is most strong, the leg speed of patient with nervous system disease, step-length, step are high, step pitch is substantially less than control group；And its step width is then significant big
In the control group of normal person；In the correlated characteristic of ability of posture control, that turns round in the walking of patient with nervous system disease group is used
Time is significantly higher than normal controls group.Second is that 4 in acquisition organize greatly in 16 kinds of motor function inspections movement, three meter Bu Hang and
Turning round is common actions in daily life, and refers to nose test, both hands inspections movement is then dynamic for the inspection of task formula in turn etc.
Make, is based on this, one general assessment models monitored for patient examination and the daily state of an illness of building, using as described in Table 1
23 dimension gait correlated characteristics as the training data of model be optimal.
Table 1 23 ties up gait correlated characteristic and its calculation method
Above-mentioned table 1 gives the calculation method of 23 dimension gait correlated characteristics simultaneously, wherein being Descartes's three-dimensional seat with human body
The origin of system is marked, X is the positive side direction of body, and Y is vertical direction, and Z is the positive front-rear direction of body.
Due to that over-fitting situation can occur in machine learning modeling, therefore, the model utilizes linear discriminent analysis pair
23 dimensional features have carried out dimension-reduction treatment, and model acquirement is allowed to possess better generalization ability.
In step 4), classification prediction model is the general assessment monitored for patient examination and the daily state of an illness
Model, can be based on five kinds of different machine in normal service learning algorithms, by from the original human body athletic posture number acquired in the past
According to 23 dimension gait correlated characteristics of middle extraction as training data, which is trained.Selected machine learning algorithm
Including five kinds of SVM (RBF-kernel), GBDT, KNN, RF, LR.
Then, the model of above-mentioned five kinds of algorithms carries out two classification based trainings, and to each by the way of 10 folding cross validations
Classification accuracy, accuracy, recall rate and the F1 of a model are measured, the area under ROC curve has carried out performance evaluation, consider number
The characteristics of according to smaller, positive and negative sample imbalance is collected, using better performances and suitable for the model of data set.As shown in table 2.
2 results of performance analysis of table
When selecting the model based on above-mentioned any algorithm, the forward of the performance seniority among brothers and sisters of above-mentioned five kinds of indexs is considered first
Person, next generalization ability for comprehensively considering model are appropriate for data set.By upper table data it is found that SVM (RBF-kernel)
Reached highest accuracy rate, and accurate rate, recall rate, F1 value ranked second, be only second to KNN, but in view of data set compared with
The generalization ability of small, positive and negative imbalanced training sets and model, the SVM (RBF-kernel) for possessing slack variable, which is directed to, has the spy
The data set of point has outstanding generalization ability, and KNN model is too simple and does not have good generalization ability, therefore is based on
The model performance of vector machine algorithm is best.
In step 5), the method that the visualization motor function audit report utilizes data visualization, the inspection to emphasis
The motor function quantized data for looking into movement carries out visual analyzing, forms a motor function audit report, enable a physician to compared with
Intuitively to obtain inspection result, analyzed and diagnosed convenient for it.Diagnosis report schematic diagram as shown in Figure 4.
Wherein, the report is directed to three meter Bu Hang, refers to nose test, these three inspections of lower limb coordination testing movement, respectively
Give timing variations curve, the gradation histogram threshold value division figure of cyclical action, variation in movement check process
Harmony bias figure is spent, and the two dimension of classification prediction model shows scatter plot.And one is finally being given in the form of text
The suggestion of a auxiliary diagnosis.
It is auxiliary to a kind of nerve based on human motion posture of the present invention above by form expression and case study on implementation
Assisted methods for diagnosing diseases is described in detail, but specific implementation form of the invention is not limited thereto.The one of this field
As technical staff, can it is carried out without departing substantially from the spirit of the method for the invention and principle in the case where it is various apparent
Variation and modification.Protection scope of the present invention should be subject to described in claims.
Claims (10)
1. a kind of neurological disease assistant diagnosis system based on human motion posture, comprising:
One data acquisition module, for acquiring the human motion attitude data of examiner；
One data preprocessing module, for being pre-processed to human motion attitude data；
One characteristic extracting module, for extracting step from the human body attitude feature that pretreated human motion attitude data includes
State correlated characteristic, the gait correlated characteristic include；
One classification prediction module, including the classification prediction model for gait correlated characteristic, which is used for will
The multidimensional gait correlated characteristic is input in the classification prediction model, to diagnose to examiner；
One audit report generation module, for visualization motor function audit report will to be generated to the result that examiner diagnoses.
2. the system as claimed in claim 1, which is characterized in that the data acquisition module acquires people by a depth camera
Body athletic posture data, the data include three kinds of colour information stream, depth information stream, bone information stream flow datas.
3. the system as claimed in claim 1, which is characterized in that four groups of inspection movements of the data collecting module collected examiner
Action data, which includes hand function and lower limb test, seat stands up, erect position balances, walk test.
4. the system as claimed in claim 1, which is characterized in that the pretreatment includes being compressed to data, except making an uproar, smoothly
Processing；In pretreatment, the timing convolution mask of triple channel is constructed, slides simultaneously real-time perfoming convolution in timing by the template
It calculates, by Gaussian smoothing, generates new timing information, reduce white Gaussian noise；The smoothing processing refers to frame per second
Non-uniform video carries out uniform interleave processing, so that it is kept 30 frame per second, while carrying out mirror image switch to image.
5. the system as claimed in claim 1, which is characterized in that the classification prediction model is analyzed using linear discriminent to institute
It states multidimensional gait correlated characteristic and carries out dimension-reduction treatment.
6. the system as claimed in claim 1, which is characterized in that the classification prediction model is based on machine learning algorithm by one kind
Model by training data training obtain, the training data be human motion attitude data gathered in advance it is extracted 23 dimension
Gait correlated characteristic.
7. system as described in claim 1 or 6, which is characterized in that the multidimensional gait correlated characteristic is that 23 dimension gaits are related
Feature；The 23 dimension gait correlated characteristic includes leg speed, right leg speed mean value, right leg speed variance, left leg speed mean value, left leg speed side
Difference, Zhou Qi Walk speed mean value, Zhou Qi Walk speed variance, right step-length mean value, right step-length variance, Zuo Buchang mean value, Zuo Buchang variance, left and right
Step size coordinating, the high mean value of right step, the high variance of right step, Zuo Bugao mean value, Zuo Bugao variance, the high harmony of left and right step, step width are equal
Value, step width variance, step pitch mean value, step pitch variance, upper body part part Z angle mean value, upper body part divide Z angle variance.
8. system as claimed in claim 6, which is characterized in that the model based on machine learning algorithm is based on Gaussian kernel
Vector machine that function is supported, gradient boosted tree, K- neighbour, random forest, logistical regression this five kinds of algorithms one of them
Model.
9. system as claimed in claim 8, which is characterized in that carry out two classification to the model based on above-mentioned five kinds of algorithms and instruct
Practice, classification accuracy, the accuracy, recall rate, F1 measurement, ROC of each model are obtained by the way of 10 folding cross validations
Area under a curve, in conjunction with the generalization ability of each algorithm, select to be suitable for the best model of human motion attitude data collection as
Classification prediction model.
10. the system as claimed in claim 1, which is characterized in that the visualization motor function audit report by three meter Bu Hang,
Refer to that nose test, these three inspections of lower limb coordination testing movement are depicted as correlation graph, auxiliary diagnosis is provided in the form of text
It is recommended that.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811416598.7A CN109717833A (en) | 2018-11-26 | 2018-11-26 | A kind of neurological disease assistant diagnosis system based on human motion posture |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811416598.7A CN109717833A (en) | 2018-11-26 | 2018-11-26 | A kind of neurological disease assistant diagnosis system based on human motion posture |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109717833A true CN109717833A (en) | 2019-05-07 |
Family
ID=66294641
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811416598.7A Pending CN109717833A (en) | 2018-11-26 | 2018-11-26 | A kind of neurological disease assistant diagnosis system based on human motion posture |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109717833A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109805898A (en) * | 2019-03-22 | 2019-05-28 | 中国科学院重庆绿色智能技术研究院 | Critical illness Mortality Prediction method based on attention mechanism timing convolutional network algorithm |
| CN110215202A (en) * | 2019-05-14 | 2019-09-10 | 杭州电子科技大学 | The pre- measuring/correlation method in Cardiac RR interval based on gait nonlinear characteristic |
| CN110236523A (en) * | 2019-06-17 | 2019-09-17 | 杭州电子科技大学 | Gait-Cardiac RR interval the correlating method returned based on Gauss |
| CN110390298A (en) * | 2019-07-23 | 2019-10-29 | 曲彦隆 | A kind of gait simulation and forecast system and simulating and predicting method |
| CN110561399A (en) * | 2019-09-16 | 2019-12-13 | 腾讯科技（深圳）有限公司 | Auxiliary shooting device for dyskinesia condition analysis, control method and device |
| CN112233800A (en) * | 2020-11-19 | 2021-01-15 | 吾征智能技术(北京)有限公司 | Disease prediction system based on abnormal behaviors of children |
| CN112401834A (en) * | 2020-10-19 | 2021-02-26 | 南方科技大学 | Movement-obstructing disease diagnosis device |
| WO2021063935A1 (en) | 2019-09-30 | 2021-04-08 | F. Hoffmann-La Roche Ag | Prediction of disease status |
| CN112932663A (en) * | 2021-03-02 | 2021-06-11 | 成都与睿创新科技有限公司 | Intelligent auxiliary method and system for improving safety of laparoscopic cholecystectomy |
| CN113545771A (en) * | 2021-07-12 | 2021-10-26 | 西安交通大学 | Integrated K-nearest neighbor quantitative diagnosis system for Parkinson disease based on plantar pressure |
| CN114098714A (en) * | 2021-11-12 | 2022-03-01 | 深圳市臻络科技有限公司 | Method for establishing frozen gait recognition model based on machine vision |
| CN114532986A (en) * | 2022-02-09 | 2022-05-27 | 北京中科睿医信息科技有限公司 | Human body balance measurement method and system based on three-dimensional space motion capture |
| CN116259405A (en) * | 2021-12-09 | 2023-06-13 | 凝动万生医疗科技(武汉)有限公司 | Robotic Procedure Automation (RPA) system and method for dyskinesia disease |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103956171A (en) * | 2014-04-01 | 2014-07-30 | 中国科学院软件研究所 | Multi-channel mini-mental state examination system |
| CN107092861A (en) * | 2017-03-15 | 2017-08-25 | 华南理工大学 | Lower limb movement recognition methods based on pressure and acceleration transducer |
| CN107273677A (en) * | 2017-06-08 | 2017-10-20 | 中国科学院软件研究所 | A kind of multi-channel nerve function quantitative evaluation system |
| US20180184948A1 (en) * | 2016-12-30 | 2018-07-05 | Mindmaze Holding Sa | System, method and apparatus for diagnosis and therapy of neuromuscular or neurological deficits |
-
2018
- 2018-11-26 CN CN201811416598.7A patent/CN109717833A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103956171A (en) * | 2014-04-01 | 2014-07-30 | 中国科学院软件研究所 | Multi-channel mini-mental state examination system |
| US20180184948A1 (en) * | 2016-12-30 | 2018-07-05 | Mindmaze Holding Sa | System, method and apparatus for diagnosis and therapy of neuromuscular or neurological deficits |
| CN107092861A (en) * | 2017-03-15 | 2017-08-25 | 华南理工大学 | Lower limb movement recognition methods based on pressure and acceleration transducer |
| CN107273677A (en) * | 2017-06-08 | 2017-10-20 | 中国科学院软件研究所 | A kind of multi-channel nerve function quantitative evaluation system |
Non-Patent Citations (2)
| Title |
|---|
| 李茜楠: "基于Kinect的异常步态检测", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 黄建平 等: "《帕金森病诊疗与康复》", 30 June 2015 * |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109805898A (en) * | 2019-03-22 | 2019-05-28 | 中国科学院重庆绿色智能技术研究院 | Critical illness Mortality Prediction method based on attention mechanism timing convolutional network algorithm |
| CN109805898B (en) * | 2019-03-22 | 2024-04-05 | 中国科学院重庆绿色智能技术研究院 | Critical death prediction method based on attention mechanism time sequence convolution network algorithm |
| CN110215202A (en) * | 2019-05-14 | 2019-09-10 | 杭州电子科技大学 | The pre- measuring/correlation method in Cardiac RR interval based on gait nonlinear characteristic |
| CN110236523A (en) * | 2019-06-17 | 2019-09-17 | 杭州电子科技大学 | Gait-Cardiac RR interval the correlating method returned based on Gauss |
| CN110390298A (en) * | 2019-07-23 | 2019-10-29 | 曲彦隆 | A kind of gait simulation and forecast system and simulating and predicting method |
| CN110390298B (en) * | 2019-07-23 | 2022-01-11 | 曲彦隆 | Gait simulation prediction system and simulation prediction method |
| CN110561399A (en) * | 2019-09-16 | 2019-12-13 | 腾讯科技（深圳）有限公司 | Auxiliary shooting device for dyskinesia condition analysis, control method and device |
| US11945125B2 (en) | 2019-09-16 | 2024-04-02 | Tencent Technology (Shenzhen) Company Limited | Auxiliary photographing device for dyskinesia analysis, and control method and apparatus for auxiliary photographing device for dyskinesia analysis |
| WO2021063935A1 (en) | 2019-09-30 | 2021-04-08 | F. Hoffmann-La Roche Ag | Prediction of disease status |
| CN112401834A (en) * | 2020-10-19 | 2021-02-26 | 南方科技大学 | Movement-obstructing disease diagnosis device |
| CN112233800B (en) * | 2020-11-19 | 2024-06-14 | 吾征智能技术(北京)有限公司 | Disease prediction system based on abnormal behaviors of children |
| CN112233800A (en) * | 2020-11-19 | 2021-01-15 | 吾征智能技术(北京)有限公司 | Disease prediction system based on abnormal behaviors of children |
| CN112932663B (en) * | 2021-03-02 | 2021-10-22 | 成都与睿创新科技有限公司 | Intelligent auxiliary system for improving safety of laparoscopic cholecystectomy |
| CN112932663A (en) * | 2021-03-02 | 2021-06-11 | 成都与睿创新科技有限公司 | Intelligent auxiliary method and system for improving safety of laparoscopic cholecystectomy |
| CN113545771A (en) * | 2021-07-12 | 2021-10-26 | 西安交通大学 | Integrated K-nearest neighbor quantitative diagnosis system for Parkinson disease based on plantar pressure |
| CN113545771B (en) * | 2021-07-12 | 2022-10-28 | 西安交通大学 | Integrated K-neighbor quantitative Parkinson disease diagnosis system based on plantar pressure |
| CN114098714B (en) * | 2021-11-12 | 2024-06-07 | 深圳市臻络科技有限公司 | Method for establishing frozen gait recognition model based on machine vision |
| CN114098714A (en) * | 2021-11-12 | 2022-03-01 | 深圳市臻络科技有限公司 | Method for establishing frozen gait recognition model based on machine vision |
| CN116259405A (en) * | 2021-12-09 | 2023-06-13 | 凝动万生医疗科技(武汉)有限公司 | Robotic Procedure Automation (RPA) system and method for dyskinesia disease |
| CN114532986A (en) * | 2022-02-09 | 2022-05-27 | 北京中科睿医信息科技有限公司 | Human body balance measurement method and system based on three-dimensional space motion capture |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109717833A (en) | A kind of neurological disease assistant diagnosis system based on human motion posture | |
| US20230181077A1 (en) | Machine differentiation of abnormalities in bioelectromagnetic fields | |
| JP7519045B2 (en) | Method and system for assessing disease using dynamic analysis of cardiac and photoplethysmographic signals - Patents.com | |
| US9101276B2 (en) | Analysis of brain patterns using temporal measures | |
| Shin et al. | Decision boundary-based anomaly detection model using improved AnoGAN from ECG data | |
| Wu et al. | A novel method to detect multiple arrhythmias based on time-frequency analysis and convolutional neural networks | |
| Klug et al. | The BeMoBIL Pipeline for automated analyses of multimodal mobile brain and body imaging data | |
| Cimr et al. | Automatic detection of breathing disorder from ballistocardiography signals | |
| Fernando et al. | Deep Learning for Medical Anomaly Detection-A Survey. | |
| Torres-García et al. | Biosignal processing and classification using computational learning and intelligence: principles, algorithms, and applications | |
| CN115691794A (en) | Auxiliary analysis method and system for neural diagnosis | |
| Abbod et al. | Survey on the use of smart and adaptive engineering systems in medicine | |
| JP2024534131A (en) | Method and system for engineering conduction deviation features from biophysical signals for use in characterizing physiological systems - Patents.com | |
| Ahamad | System architecture for brain-computer interface based on machine learning and internet of things | |
| Erin et al. | Spectral Analysis of Cardiogenic Vibrations to Distinguish Between Valvular Heart Diseases. | |
| Poon et al. | Special issue on health informatics and personalized medicine | |
| WO2021031155A1 (en) | Method and device for multi-scale characteristic extraction based on ecg | |
| Patil et al. | Prediction and analysis of heart disease using SVM algorithm | |
| Sanamdikar et al. | Classification of ECG Signal for Cardiac Arrhythmia Detection Using GAN Method | |
| Ramirez-Bautista et al. | Artificial intelligence approaches to physiological parameter analysis in the monitoring and treatment of non-communicable diseases: A review | |
| Cipresso et al. | Big data in preclinical ECG alterations research | |
| Bao et al. | Processing of Cardiac Signals for Health Monitoring and Early Detection of Heart Diseases | |
| Shankar et al. | Wavelet based Machine Learning Approaches towards Precision Medicine in Diabetes Mellitus. | |
| TWI708589B (en) | System for predicting cardiovascular and brain function in combination with physiological detection device and method thereof | |
| US20230076069A1 (en) | Methods and Systems for Engineering Cardiac Waveform Features From Biophysical Signals for Use in Characterizing Physiological Systems |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190507 |