CN109920551A - Autism children social action performance characteristic analysis system based on machine learning - Google Patents
Autism children social action performance characteristic analysis system based on machine learning Download PDFInfo
- Publication number
- CN109920551A CN109920551A CN201910068956.8A CN201910068956A CN109920551A CN 109920551 A CN109920551 A CN 109920551A CN 201910068956 A CN201910068956 A CN 201910068956A CN 109920551 A CN109920551 A CN 109920551A
- Authority
- CN
- China
- Prior art keywords
- analysis
- autism children
- performance characteristic
- machine learning
- action performance
- 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
- 238000004458 analytical method Methods 0.000 title claims abstract description 87
- 206010003805 Autism Diseases 0.000 title claims abstract description 83
- 208000020706 Autistic disease Diseases 0.000 title claims abstract description 83
- 238000010801 machine learning Methods 0.000 title claims abstract description 37
- 230000009471 action Effects 0.000 title claims description 49
- 238000000034 method Methods 0.000 claims abstract description 20
- 230000006870 function Effects 0.000 claims abstract description 14
- 230000000694 effects Effects 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims description 22
- 238000003066 decision tree Methods 0.000 claims description 14
- 230000007547 defect Effects 0.000 claims description 11
- 238000011156 evaluation Methods 0.000 claims description 8
- 230000010354 integration Effects 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 7
- 230000001186 cumulative effect Effects 0.000 claims description 4
- 238000011161 development Methods 0.000 claims description 4
- 201000010099 disease Diseases 0.000 claims description 4
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims description 4
- 238000013473 artificial intelligence Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000004891 communication Methods 0.000 claims description 2
- 230000008451 emotion Effects 0.000 claims 1
- 208000024891 symptom Diseases 0.000 abstract description 19
- 208000030251 communication disease Diseases 0.000 abstract 1
- 230000011273 social behavior Effects 0.000 abstract 1
- 230000006399 behavior Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 238000003745 diagnosis Methods 0.000 description 3
- 238000011835 investigation Methods 0.000 description 3
- 238000012417 linear regression Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 230000003997 social interaction Effects 0.000 description 2
- 238000013179 statistical model Methods 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000008131 children development Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000003032 molecular docking Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000001225 therapeutic effect Effects 0.000 description 1
- 230000001755 vocal effect Effects 0.000 description 1
- 230000003442 weekly effect Effects 0.000 description 1
Landscapes
- Electrically Operated Instructional Devices (AREA)
Abstract
The method for the autism children human communication disorders symptom analysis based on machine learning that the invention discloses a kind of comprising the steps of: obtain all types of autism children social behavior feature set;Obtain the analysis of cases report of previous autism children;System learns the information of typing;Update fitting function;New analysis of cases is carried out using updated fitting function.The autism children analysis of cases system based on machine learning that the invention also discloses a kind of.The present invention is based on the analysis methods of machine learning can be according to the analysis of cases information of existing children, and therefrom study is regular to relevant signature analysis automatically;And in later analytic activity, according to the characteristic parameter that newly inputs and study to analysis method be automatically that rehabilitation personnel recommends diagnostic result to be with a wide range of applications to greatly improve the efficiency and accuracy of rehabilitation.
Description
Technical field
The invention belongs to field of artificial intelligence, and in particular to a kind of autism children society row based on machine learning
For the analysis method and system of performance characteristic.
Background technique
Symptom analysis is the important ring in autism children rehabilitation.Currently, in the rehabilitation link of autism children,
Symptom analysis is all to be showed by rehabilitation teacher according to feature, based on the symptom analysis in the past with similar behavior, is generated new
Symptom analysis feeds back to autism children.
Rehabilitation teacher is generated new based on the symptom analysis in the past with similar behavior according to previous performance characteristic
Symptom analysis is to autism children.This analysis method is dependent in the past with the symptom analysis of similar behavior.If currently certainly
The type or quantity for closing disease children's Symptoms change, and certainly will influence whether corresponding symptom analysis.For rehabilitation teacher
For, if difference all occurs for the symptom combination and previous any successful analysis case after changing, it is difficult accurate judgement and works as
What kind of the change of preceding performance, which generates to symptom analysis, influences.Therefore it very likely will lead to symptom analysis to go wrong, thus
Mistaken diagnosis is caused, autism children cannot train accordingly.
Summary of the invention
To overcome and solving the above-mentioned problems in the prior art, the present invention provides a kind of pairs based on machine learning
The method and system of autism children social action performance characteristic progress intellectual analysis.
The analysis method of autism children social action performance characteristic proposed by the present invention based on machine learning, comprising with
Lower step:
A, all kinds of autism children social action performance characteristic set are obtained, by sociability evaluation questionnaire to the spy
Collection is closed and is handled, and the autism children social action performance characteristic set being corrected is obtained;
B, the analysis of cases report information of existing previous autism children social action performance characteristic is obtained;
C, analysis of cases acquired in the set of the behavior expression feature in abovementioned steps A and the step B is reported
Machine learning training system of the information input based on XGBoost algorithm, the analysis of cases report of the training system to typing
It accuses information to be learnt, the study refers to the social basic skill training mesh of study 47 (that is, estimation items mesh number summation in Fig. 4)
Mark and 6 big autism children essential informations and specific intervention stratege therefrom extract characteristic parameter, export each node of decision tree
Weight is divided, corresponding decision tree parameter is obtained by calculating, finally obtains and meet expected decision tree, i.e. learning model;
D, analysis of cases is carried out to autism children social action performance characteristic using learning model;
E, the learning model is updated, is updated in the autism children social action performance characteristic and step B in step A
Analysis of cases report information;
F, new analysis of cases is carried out using updated learning model, and next autism children social action is showed
Feature carries out analysis of cases.
In the step A, the performance characteristic type that different types of social action performance symptom is included is different.Such as Fig. 3
Shown, the autism children social action performance characteristic includes: social attention defect, self-consciousness defect, social affective
Defect and communication exchange defect.
In the step A, the set of the autism children social action performance characteristic is as shown in Figure 4: reference of the present invention
" autism child development assesses table (social field) " of China Disabled Persons Federation, by basic capacity, society before social activity in original social interaction
Three skill, etiquette columns is handed over to carry out splitting and reorganizing, and according to autism children sociability development characteristic, by social energy
Power splitting and reorganizing is social attention, self-consciousness, non-spoken language interpersonal skill, spoken interpersonal skill, greeting, farewell, phone gift
Instrument, eight sub- columns of high-order ceremony include several sub-projects in every sub- column;Obtain autism children performance as shown in Figure 4
Characteristic set has used unique method of the invention in this " set " technical treatment, has been different from existing site assessment and obtains
The method that must gather:
Sociability evaluation questionnaire
Capability evaluation questionnaire is a questionnaire for being directed to autism children parent, and questionnaire passes through setting 47 social energy
The relevant speciality problem in power field collects the sociability information of autism children, and problem broad covered area, difficulty are low, are easy to discriminate
It not, while being the artificial method of investigation and study for being different from Traditional Rehabilitation teacher non-programming, systematization;
About characteristic set collection method in invention:
(1) method with explanation: capability evaluation questionnaire be issued to online by APP and wechat public platform it is self-closing
Disease Parents carry out information and include, without going to proving ground under line can be completed, to improve the whole effect of diagnosis rehabilitation
Rate.
(2) feature collection mode and the shown relationship gathered of Fig. 4: questionnaire design is complete with shown 47 sociability indexs
Complete corresponding, the information for directly extracting respective capabilities index field carries out division arrangement, and frame diagram shown in Fig. 4 can be obtained;
(3) accuracy and normative, the investigation side of rehabilitation teacher the technical advantage compared with existing manual method: are improved
Formula is relatively subjective and arbitrarily, it cannot be guaranteed that the mode investigated every time is precisely in place;All standing, and the questionnaire of template type to investigate
Process is in the efficient of same standard and specification always;Efficiency is improved, the table and chart automatically generated by computer is convenient for
Mutually conversion and data processing, shorten Data Integration time and process;It ensure that the uniformity of investigation method and data preparation,
The present invention is the questionnaire assessment and scene assessment established based on same set of sociability evaluation system, so whole from investigating to collecting
Reason result be not in observation variable can not tandem docking the case where;
In the step B, the analysis of cases report information refer to before according to rehabilitation teacher come the self-closing disease evaluated
Children's symptomatic diagnosis and corresponding rehabilitation suggestion;The case of the autism children social action performance characteristic refered in particular in the present invention
Analysis report information include the test results such as children's sociability development orientation map, social skill interpersonal skill etiquette,
Corresponding test result analysis and rehabilitation suggestion, the rehabilitation frequency are recommended;
Wherein, the step C is specifically:
The data (case described in behavior expression characteristic set described in step A and step B of researcher's input
Analysis report information) it is used as learning sample, utilize sample data (the 47 social basic skills got by the step B
Training objective and 6 big autism children essential informations and specific intervention stratege), characteristic parameter is therefrom extracted, decision is exported
Each node split weight is set, is obtained by calculation and obtains corresponding decision tree parameter, finally obtain and meet expected decision tree, that is, learn
Practise model.
Wherein, the characteristic parameter refers to social attention defect, self-consciousness defect, social affective defect and links up friendship
Flow defect and 6 big autism children essential informations;
Wherein, described 6 big autism children essential informations refer to age, language competence, learning ability, gender, nurse shape
Condition and family background;
Wherein, it is obtained by calculation and obtains in corresponding decision tree parameter, the calculating refers to given for one have n a
The data set of sample and m feature, input set are as follows:
D=(xi, yi) (| D |=n, xi∈Rm, yi∈R);
And it predicts to export using K cumulative functions with a tree ensemble model:
Wherein, determine index using the common R^2 goodness of fit to judge learning outcome, in statistics for measure because
Ratio shared by part can be explained as independent variable in the variation of variable, the explanation strengths of statistical model is judged with this, and in this reality
In the linear regression tested, the coefficient of determination is square of sample correlation coefficient;After other regressors are added, coefficient of determination phase
Become square of coefficient of multiple correlation with answering.
There is the variable y1 of n value for one group, y2....yn, (such as big essential information of autism children 6 is defeated as variable
Enter) have:
Then total sum of squares are as follows:
Regression sum of square are as follows:
The coefficient of determination can be write as:
For the coefficient of determination closer to 1, the degree of fitting of model is higher;As shown in figure 8, substituting into random data obtains goodness of fit system
Number is 0.7828, if goodness of fit coefficient is closer to 1, model accuracy will be higher.The calculating side proposed through the invention
Method can greatly improve the efficiency and accuracy of rehabilitation, be with a wide range of applications.
Wherein, step D carries out analysis of cases to autism children social action performance characteristic, the offer in analytic process
Sociability master rate judges and sociability rehabilitation suggestion is peculiar as the present invention;
Wherein, the step E renewal learning model specifically includes the step for updating new performance characteristic value and update typing
The related data of rapid A, B;
Based on above method, the autism children social action based on machine learning that the invention also provides a kind of shows special
The analysis system of sign, the system comprises consisting of:
Data inputting module, including front-end A PP and wechat public platform, for being responsible for typing autism children social action table
The set of existing feature and the analysis of cases report information of the existing autism children social action performance characteristic;
Server module is responsible for the data received in data inputting module and carries out preliminary screening;
Learning training module, for the machine learning training system based on XGBoost algorithm, to the set and case point
It analyses report information and carries out autism children characteristic information and diagnostic message study, obtain learning model;The machine learning training
Module includes the end PC program, for carrying out Data Integration and model training;
Analysis module, for carrying out the analysis of autism children social action performance characteristic by the learning model;Institute
Stating analysis module includes doc class text tool, is responsible for integration and final data is presented and is presented to front-end A PP and wechat public platform.
Compared with prior art, the beneficial effect comprise that the present invention is based on the intelligent analysis methods of machine learning
And system, it can therefrom be learnt to relevant automatically according to the analysis of cases information of existing autism children autism children
Signature analysis rule;And in later analytic process, according to the feature performance (mark sheet of autism children to be measured newly inputted
It is existing) and it is suitable for rehabilitation's teacher recommendation by the analysis rule (the analysis of cases information of existing autism children) that study obtains
Therapeutic scheme recommends diagnostic result for rehabilitation personnel, to greatly improve the symptom analysis to autism children to be measured
Accuracy and efficiency also improves the efficiency and accuracy of rehabilitation.
Detailed description of the invention
Fig. 1 is machine learning model schematic diagram in the present invention.
Fig. 2 is training overall flow in the present invention.
Fig. 3 is autism children social action performance characteristic in the present invention.
Fig. 4 is the set of autism children social action performance characteristic in the present invention.
Fig. 5 is that the present invention is based on the signals of the analysis system of the autism children social action performance characteristic of machine learning
Figure.
Fig. 6 is the schematic diagram in the embodiment of the present invention.
Fig. 7 is the schematic diagram of assessment result in the embodiment of the present invention.
Fig. 8 is each node split weight example of decision tree in learning model in the present invention.
Fig. 9 is the schematic diagram of analysis system in the present invention.
Specific embodiment
In conjunction with following specific embodiments and attached drawing, the present invention is described in further detail.Implement process of the invention,
Condition, experimental method etc. are among the general principles and common general knowledge in the art, this hair in addition to what is specifically mentioned below
It is bright that there are no special restrictions to content.
Embodiment 1
As shown in Figure 1, the present invention is based on the analysis method of the autism children social action performance characteristic of machine learning, packet
Containing following steps (as shown in Figure 2):
A, the social action performance symptom set of certain type autism children autism children is obtained.Different types of society
The performance characteristic type that behavior expression symptom is included is different.For example, need exist for obtaining is that certain type of symptom is wrapped
The complete or collected works of the performance characteristic contained;
B, the performance characteristic of previous symptom is obtained.Pervious performance characteristic is obtained, (includes children society by passing characteristic information
Hand over test results, corresponding test result analysis and the health such as ability development positioning map, social skill interpersonal skill etiquette
Rebuild view, the rehabilitation frequency is recommended) it is used as input sample,And provide corresponding rehabilitation teacher by conventional methodAnalysis of cases knot
Fruit is as output sample, for the training sample set in machine learning;
C, the information of typing is learnt.The existing autism children society acquired using the step B
The analysis of cases report information of meeting behavior expression feature therefrom extracts characteristic parameter, output is determined as machine learning sample data
Each node split weight of plan tree is obtained by calculation and obtains corresponding decision tree parameter, finally obtains and meets expected decision tree i.e.
Learning model;
D, analysis of cases is carried out to autism children social action performance characteristic.
E, renewal learning model all updates its learning model for every a kind of product;
F, new analysis is carried out using updated learning model.For new performance characteristic value, it can use and obtained
To learning model calculated, to recommend suitable scheme to rehabilitation teacher.
Learning model in the embodiment of the present invention:
According to questionnaire data, assessment report is generated, carries out artificial intelligence deduction, by mass data operation, forms assessment
It is recommended that.
Wherein,
(1) 47 social interaction evaluation items are combined, XGBoost is introduced and carries out factors assessment
XGBoost algorithm, full name eXtreme Gradient Boosting are that Gradient Tree Boosting is calculated
One branch of method is the expansible machine learning system of Tree Boosting a kind of.XGBoost is the distribution of a update
Formula grad enhancement library, it is intended to it realizes efficiently, it is flexibly and portable.It realizes machine learning under Gradient Boosting frame
Algorithm.XGBoost provides parallel tree and promotes (also referred to as GBDT, GBM), can rapidly and accurately solve many data sciences and ask
Topic.Identical code is run on main distributed environment (Hadoop, SGE, MPI), and can solve more than billions of
The problem of a sample.
1) model is summarized
The core concept of XGBoost as being is promoted according to the negative gradient direction of loss function with GBDT, be into
It has gone the GBDT algorithm of Taylor's second outspread, has added some regular terms.It is given for one to have n sample and m feature
Data set
D=(xi, yi) (| D |=n, xi∈Rm, yi∈R)
One tree ensemble model predicts to export using K cumulative functions:
WhereinIt is the space of CART (regression tree).Wherein q represents the knot of each tree
Each sample can be mapped in corresponding leaf node by structure, and T is the number of leaf node in tree.It is each to correspond to one
Independent tree construction q and leaf weight w.Different from decision tree, each regression tree is continuous comprising one on each leaf node
Fractional value represents the score of i-th of node.It is the marking to sample x, i.e. model predication value.For each sample, will use more
It is categorized into leaf node by decision rule in a tree, and final to obtain by the score w in cumulative corresponding leaf
Prediction (summation that the prediction result of each sample is exactly each tree prediction score).For collection of functions used in learning model
It closes, minimizes following regularization target:
Wherein l be one can dimpling loss function, the difference between predictive metrics value and target value.Section 2 punishes model
Complexity (the sum of the complexities of all regression trees).Contain two parts in this, one be leaf node sum, one
It is the L2 regularization term that leaf node obtains.This additional regularization term is capable of the study weight of smooth each leaf node to keep away
Exempt from over-fitting.Intuitively, the target of regularization will tend to selection using simple and anticipation function model.Work as regularization parameter
When being zero, this function just becomes traditional GDBT.
2)GDBT
Aggregation model is set using function as parameter, so cannot directly be updated using traditional update method.But
Using addition mode of learning (Additve training) training, start from the prediction of constant, every time increased one it is new
The current tree of function learning finds current optimal tree-model and is added in band overall model:
Therefore, key is to learn the t tree, finds optimal ft, increase ftAnd objective function is minimized, whereinIt is
The predicted value of sample i in the t times iteration:
After experience error, objective function can be rewritten as:
After experience error the second Taylor series:
So final target is after eliminating constant term:
(2) index is determined by the R^2 goodness of fit to judge learning outcome
The present invention will carry out the Structure learning of assessment completeness using XGBoost algorithm and 1000+ parts of effective sample data,
Suitable intervention stratege is finally calculated using linear regression.It can be by becoming certainly in variation in statistics for measuring dependent variable
Amount explains ratio shared by part, and the explanation strengths of statistical model is judged with this.For simple linear regression, the coefficient of determination
For square of sample correlation coefficient.After other regressors are added, the coefficient of determination correspondingly becomes coefficient of multiple correlation
Square.
There are the variable y1, y2....yn of n value for one group, have:
Then total sum of squares are as follows:
Regression sum of square are as follows:
The coefficient of determination can be write as:
For the coefficient of determination closer to 1, the degree of fitting of model is higher.
The step D specifically: as shown in fig. 6, choosing first example, conceal crucial personal information, logging data generates
Autism children social contact ability assessment report:
One, assessment result
Assessment result is as shown in Figure 7.
The specific data of table 1: Fig. 7 are published
Note: test result is score/total score (accuracy) in table
Interpretation of result part is the detailed analysis of table 1.
Two, interpretation of result
The monthly age of the children is 99 months, and basic skills master rate is 57.14% before social activity, and interpersonal skill master rate is
50.00%, etiquette master rate is 48.00%.
As shown in Figure 7, zero Xiang Dabiao of children, social attention, self-consciousness, non-spoken interpersonal skill, spoken social activity
Skill, greeting, farewell, phone ceremony, high-order ceremony eight it is below standard.Basic skills (society before on the whole the children are social
Property pay attention to, self-consciousness) generally, interpersonal skill (non-spoken language interpersonal skill, spoken interpersonal skill) generally, etiquette (greet,
Farewell, phone ceremony, high-order ceremony) it is general.
Three, rehabilitation suggestion
According to the monthly age of the children and sociability developmental sequence, it is proposed that combine reality scene in daily life, pass through
Relevant nursery rhymes, story and life activity are first intervened from following five contents.
Project 1 actively causes caretaker's concern
Intervene target:
Actively cause caretaker's concern
Intervention Strategy:
1) actively cause caretaker's concern
2) actively cause caretaker's concern
Project 2 smiles and responds the greeting of stranger
Intervene target:
It smiles and responds the greeting of stranger
Intervention Strategy:
1) it smiles and responds the greeting of stranger
2) it smiles and responds the greeting of stranger
Project 3, stranger close to when have appropriate dodge
Intervene target:
Stranger is looked at when stranger goes in face of oneself, and starts to take a step to go ahead around stranger.
Intervention Strategy:
1) by relevant children's song and children's story, allowing children to understand will be far from stranger.
2) band children outside encounters the unacquainted people of children close to children and when attempting to have extremity in daily life, uses
Do verbal cue children: you recognize Ta? can shake hands/allow with Ta Ta armfuls? and it is dodged (such as: double with movement signal children
It after the back of the hand, spaces out with stranger).
Project 4, caretaker have the meaning for chasing caretaker when leaving
Intervene target:
There is the meaning for chasing caretaker when caretaker leaves
Intervention Strategy:
1) there is the meaning for chasing caretaker when caretaker leaves
2) there is the meaning for chasing caretaker when caretaker leaves
Project 5, the article for recognizing oneself
Intervene target:
Recognize the article of oneself
Intervention Strategy:
1) recognize the article of oneself
2) recognize the article of oneself
Four, intervene the frequency
It is no less than 3 times weekly;
No less than 30 minutes every time;
Each node split weight example of decision tree is as shown in figure 8, the weighted value at age is about 0.23 language in learning model
The weighted value of ability is about 0.21, and the weighted value of learning ability is about 0.19, and gender male and woman's weighted value respectively may be about 0.19
With 0.04, the weighted value of parent's nurse and ancestors' nurse respectively may be about 0.09 and 0.03, the weight of average family and intellectual with a senior professional title family
Value respectively may be about 0.07 and 0.02;Fig. 8 has carried out descending arrangement to the weighted value of each index while providing data, thus may be used
To see in six essential informations, the age is maximum for the learning model influence factor of machine, and language competence and learning ability are taken second place,
Mean square error, mean absolute error and interpretable variance below histogram are used to individually auxiliary verifying and are observed total scale of construction and a
The scale relationship of the corresponding desired value of the scale of construction, thus whether updated according to numerical values recited judgment models, and the goodness of fit
Coefficient then can totally find out the gap between model and absolute ideal model (regressand value 1).
As shown in figure 9, the analysis of the autism children social action performance characteristic based on machine learning in the present embodiment
System, comprising:
Data inputting module, including front-end A PP and wechat public platform, for being responsible for typing autism children social action table
The set of existing feature and the analysis of cases report information of existing autism children social action performance characteristic;
Server module is responsible for the data received in data inputting module and carries out preliminary screening;
Learning training module, for the machine learning training system based on XGBoost algorithm, to set and analysis of cases report
It accuses information and carries out autism children characteristic information and diagnostic message study, obtain learning model;Machine learning training module includes
The end PC program, for carrying out Data Integration and model training;
Analysis module, for carrying out the analysis of autism children social action performance characteristic by learning model;Analyze mould
Block includes doc class text tool, is responsible for integration and final data is presented and is presented to front-end A PP and wechat public platform.
Protection content of the invention is not limited to above embodiments.Under the spirit and scope without departing substantially from present inventive concept,
Various changes and advantages that will be apparent to those skilled in the art are all included in the present invention, and are with appended claims
Protection scope.
Claims (10)
1. a kind of analysis method of the autism children social action performance characteristic based on machine learning, which is characterized in that described
Method the following steps are included:
A, the set of autism children social action performance characteristic is obtained;
B, the analysis of cases report information of autism children social action performance characteristic is obtained;
C, based on the machine learning training system of XGBoost algorithm, the set and analysis of cases report information are carried out self-closing
Disease children's character information and diagnostic message study, the study refer to the social basic skills of study autism children 47 and 6 greatly
Autism children essential information obtains learning model;The learning model is that a gradient promotes tree-model;
D, the analysis of autism children social action performance characteristic is carried out by the learning model.
2. the analysis method of the autism children social action performance characteristic according to claim 1 based on machine learning,
It is characterized in that, further, being updated to the learning model, comprising:
E, the learning model is updated, the case in the autism children social action performance characteristic and step B in step A is updated
Example analysis report information;
F, the analysis of next autism children social action performance characteristic is carried out using updated learning model.
3. the analysis method of the autism children social action performance characteristic according to claim 1 based on machine learning,
It is characterized in that, the autism children social action performance characteristic includes: social attention defect, self-consciousness defect, society
Hand over emotion flaw and communication exchange defect.
4. the analysis method of the autism children social action performance characteristic according to claim 1 based on machine learning,
It is characterized in that, using sociability evaluation questionnaire to be gathered in the step A: the sociability evaluation questionnaire is needle
To the questionnaire of autism children parent, questionnaire collects the social activity of autism children by setting 47 social basic skills
Ability information.
5. the analysis method of the autism children social action performance characteristic according to claim 4 based on machine learning,
It is characterized in that, the analysis of the autism children social action performance characteristic refers to the learning model according to the social energy
Force estimation questionnaire data generates assessment report, carries out artificial intelligence deduction, by data operation, forms assessment and suggests.
6. the analysis method of the autism children social action performance characteristic according to claim 1 based on machine learning,
It is characterized in that, in the step B, the analysis of cases report information refer to comprising children's sociability development orientation map,
Social skill interpersonal skill etiquette test result, corresponding test result analysis and rehabilitation suggestion and the rehabilitation frequency are recommended.
7. the analysis method of the autism children social action performance characteristic according to claim 1 based on machine learning,
It is characterized in that, in the step C, machine learning training system using the set and the analysis of cases report information as
Sample data extracts characteristic parameter from the sample data, exports each node split weight of decision tree, and calculate decision tree
Parameter obtains the learning model.
8. the analysis method of the autism children social action performance characteristic according to claim 7 based on machine learning,
It is characterized in that, the learning model judges learning outcome by R^2 goodness of fit index of discrimination.
9. the analysis method of the autism children social action performance characteristic according to claim 7 based on machine learning,
It is characterized in that, the calculating of the decision tree parameter refers to the data set that has n sample and m feature given for one,
Input set are as follows: D=(xi, yi) (| D |=n, xi∈Rm, yi∈ R), and it is cumulative using K with a tree ensemble model
Function predict to export:
Wherein, the coefficient of determination is square of sample correlation coefficient;After regressor is added, the coefficient of determination correspondingly becomes more
Square of recorrelation coefficient;There are the variable y1, y2....yn of n value for one group, then the coefficient of determination are as follows:
Wherein,
10. a kind of analysis system of the autism children social action performance characteristic based on machine learning, which is characterized in that use
Such as described in any item analysis methods of claim 1-9, the system comprises following:
Data inputting module, including front-end A PP and wechat public platform, it is special for being responsible for typing autism children social action performance
The set of sign and the analysis of cases report information of the existing autism children social action performance characteristic;
Server module is responsible for the data received in data inputting module and carries out preliminary screening;
Learning training module, for the machine learning training system based on XGBoost algorithm, to the set and analysis of cases report
It accuses information and carries out autism children characteristic information and diagnostic message study, obtain learning model;The machine learning training module
Including the end PC program, for carrying out Data Integration and model training;
Analysis module, for carrying out the analysis of autism children social action performance characteristic by the learning model;Described point
Analysing module includes doc class text tool, is responsible for integration and final data is presented and is presented to front-end A PP and wechat public platform.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910068956.8A CN109920551A (en) | 2019-01-24 | 2019-01-24 | Autism children social action performance characteristic analysis system based on machine learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910068956.8A CN109920551A (en) | 2019-01-24 | 2019-01-24 | Autism children social action performance characteristic analysis system based on machine learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109920551A true CN109920551A (en) | 2019-06-21 |
Family
ID=66960661
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910068956.8A Pending CN109920551A (en) | 2019-01-24 | 2019-01-24 | Autism children social action performance characteristic analysis system based on machine learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109920551A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110237397A (en) * | 2019-07-11 | 2019-09-17 | 重庆市中医院 | A kind of brain cognitive training system |
CN110674875A (en) * | 2019-09-25 | 2020-01-10 | 电子科技大学 | Pedestrian motion mode identification method based on deep hybrid model |
CN111009321A (en) * | 2019-08-14 | 2020-04-14 | 电子科技大学 | Application method of machine learning classification model in juvenile autism auxiliary diagnosis |
CN111461937A (en) * | 2020-03-31 | 2020-07-28 | 北京复米教育科技有限公司 | Automatic tracking method and system for autism intervention teaching course |
CN111524577A (en) * | 2020-04-17 | 2020-08-11 | 广东医博荟健康管理有限公司 | Autism integral intervention treatment auxiliary system and method |
CN112668809A (en) * | 2020-10-30 | 2021-04-16 | 华东师范大学 | Method for establishing autism child rehabilitation effect prediction model and method and system for predicting autism child rehabilitation effect |
US20210174933A1 (en) * | 2019-12-09 | 2021-06-10 | Social Skills Training Pty Ltd | Social-Emotional Skills Improvement |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2003075187A1 (en) * | 2002-03-02 | 2003-09-12 | Bl Systems Inc. | Apparatus and method for constructing data mining model using ensemble machines |
US20040107413A1 (en) * | 2002-11-22 | 2004-06-03 | Bixler John D. | Autism treatment system and method |
US20160179786A1 (en) * | 2014-12-19 | 2016-06-23 | International Business Machines Corporation | Diagnosing autism spectrum disorder using natural language processing |
US20160180735A1 (en) * | 2014-12-19 | 2016-06-23 | International Business Machines Corporation | Coaching a participant in a conversation |
US20170069216A1 (en) * | 2014-04-24 | 2017-03-09 | Cognoa, Inc. | Methods and apparatus to determine developmental progress with artificial intelligence and user input |
CN107748736A (en) * | 2017-10-13 | 2018-03-02 | 河海大学 | A kind of multiple-factor Remote Sensing temperature space NO emissions reduction method based on random forest |
CN107910070A (en) * | 2017-11-06 | 2018-04-13 | 广州优涵信息技术有限公司 | A kind of expert recommendation method based on expert's case library |
CN108010568A (en) * | 2017-03-24 | 2018-05-08 | 华东师范大学 | A kind of rehabilitation training system based on more course plot modules |
CN108346107A (en) * | 2017-12-28 | 2018-07-31 | 阿里巴巴集团控股有限公司 | A kind of social content Risk Identification Method, device and equipment |
CN108538365A (en) * | 2017-03-24 | 2018-09-14 | 华东师范大学 | A kind of self-closing disease sociability assessment system based on data analysis technique |
CN109256207A (en) * | 2018-08-29 | 2019-01-22 | 王雁 | A method of based on XGBoost+SVM hybrid machine Learner diagnosis keratoconus case |
-
2019
- 2019-01-24 CN CN201910068956.8A patent/CN109920551A/en active Pending
Patent Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2003075187A1 (en) * | 2002-03-02 | 2003-09-12 | Bl Systems Inc. | Apparatus and method for constructing data mining model using ensemble machines |
US20040107413A1 (en) * | 2002-11-22 | 2004-06-03 | Bixler John D. | Autism treatment system and method |
US20170069216A1 (en) * | 2014-04-24 | 2017-03-09 | Cognoa, Inc. | Methods and apparatus to determine developmental progress with artificial intelligence and user input |
US20160179786A1 (en) * | 2014-12-19 | 2016-06-23 | International Business Machines Corporation | Diagnosing autism spectrum disorder using natural language processing |
US20160180735A1 (en) * | 2014-12-19 | 2016-06-23 | International Business Machines Corporation | Coaching a participant in a conversation |
CN108010568A (en) * | 2017-03-24 | 2018-05-08 | 华东师范大学 | A kind of rehabilitation training system based on more course plot modules |
CN108538365A (en) * | 2017-03-24 | 2018-09-14 | 华东师范大学 | A kind of self-closing disease sociability assessment system based on data analysis technique |
CN107748736A (en) * | 2017-10-13 | 2018-03-02 | 河海大学 | A kind of multiple-factor Remote Sensing temperature space NO emissions reduction method based on random forest |
CN107910070A (en) * | 2017-11-06 | 2018-04-13 | 广州优涵信息技术有限公司 | A kind of expert recommendation method based on expert's case library |
CN108346107A (en) * | 2017-12-28 | 2018-07-31 | 阿里巴巴集团控股有限公司 | A kind of social content Risk Identification Method, device and equipment |
CN109256207A (en) * | 2018-08-29 | 2019-01-22 | 王雁 | A method of based on XGBoost+SVM hybrid machine Learner diagnosis keratoconus case |
Non-Patent Citations (1)
Title |
---|
车少辉: "基于神经网络方法的杉木人工林林分生长模拟研究", 《中国博士学位论文全文数据库农业科技辑》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110237397A (en) * | 2019-07-11 | 2019-09-17 | 重庆市中医院 | A kind of brain cognitive training system |
CN111009321A (en) * | 2019-08-14 | 2020-04-14 | 电子科技大学 | Application method of machine learning classification model in juvenile autism auxiliary diagnosis |
CN110674875A (en) * | 2019-09-25 | 2020-01-10 | 电子科技大学 | Pedestrian motion mode identification method based on deep hybrid model |
US20210174933A1 (en) * | 2019-12-09 | 2021-06-10 | Social Skills Training Pty Ltd | Social-Emotional Skills Improvement |
CN111461937A (en) * | 2020-03-31 | 2020-07-28 | 北京复米教育科技有限公司 | Automatic tracking method and system for autism intervention teaching course |
CN111524577A (en) * | 2020-04-17 | 2020-08-11 | 广东医博荟健康管理有限公司 | Autism integral intervention treatment auxiliary system and method |
CN112668809A (en) * | 2020-10-30 | 2021-04-16 | 华东师范大学 | Method for establishing autism child rehabilitation effect prediction model and method and system for predicting autism child rehabilitation effect |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109920551A (en) | Autism children social action performance characteristic analysis system based on machine learning | |
Marselle et al. | Does perceived restorativeness mediate the effects of perceived biodiversity and perceived naturalness on emotional well-being following group walks in nature? | |
ȚĂRANU | Data mining in healthcare: decision making and precision | |
CN105528526B (en) | A kind of life cycle dynamical health management system excavated based on traditional Chinese medicine succession with big data | |
Kenett et al. | Investigating the structure of semantic networks in low and high creative persons | |
Shaffer | Q-squared: Combining qualitative and quantitative approaches in poverty analysis | |
Kaiser et al. | Little jiffy, mark IV | |
Amelia et al. | Meta-analysis of student performance assessment using fuzzy logic | |
CN107516110A (en) | A kind of medical question and answer Semantic Clustering method based on integrated convolutional encoding | |
Borghi et al. | The role of emotions in the consumer meaning-making of interactions with social robots | |
CN107315775A (en) | A kind of index calculating platform and method | |
Ismail et al. | Behavioural features for mushroom classification | |
CN110046757B (en) | Outpatient clinic volume prediction system and prediction method based on LightGBM algorithm | |
CN106202377A (en) | A kind of online collaborative sort method based on stochastic gradient descent | |
CN106980769A (en) | A kind of destination client category system | |
CN113705725A (en) | User personality characteristic prediction method and device based on multi-mode information fusion | |
Pandiaraj et al. | Effective heart disease prediction using hybridmachine learning | |
CN105893748A (en) | Happiness index evaluation method and system based on rough set theory | |
Stanković et al. | Artificial neural network model for prediction of students’ success in learning programming | |
KR20110098286A (en) | Self health diagnosis system of oriental medicine using fuzzy inference method | |
Kasemthaweesab et al. | Association analysis of diabetes mellitus (DM) with complication states based on association rules | |
CN106055875A (en) | Dermatoglyph analysis and processing apparatus based on big data | |
Lan et al. | Evaluation system of music art instructional quality based on convolutional neural networks and big data analysis | |
CN117350905A (en) | Sports teaching management system based on big data | |
Graco et al. | Toward knowledge-driven data mining |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190621 |