CN110491520A - A kind of construction method of the sclerotin status assessment model based on semi-supervised learning - Google Patents
A kind of construction method of the sclerotin status assessment model based on semi-supervised learning Download PDFInfo
- Publication number
- CN110491520A CN110491520A CN201910682966.0A CN201910682966A CN110491520A CN 110491520 A CN110491520 A CN 110491520A CN 201910682966 A CN201910682966 A CN 201910682966A CN 110491520 A CN110491520 A CN 110491520A
- Authority
- CN
- China
- Prior art keywords
- data
- sclerotin
- assessment
- group
- status
- 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
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
- G16H20/30—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to physical therapies or activities, e.g. physiotherapy, acupressure or exercising
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
- G16H20/60—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to nutrition control, e.g. diets
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Nutrition Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Physical Education & Sports Medicine (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The construction method of the invention discloses a kind of sclerotin status assessment model based on semi-supervised learning, it include: the assessment data for extracting the user of multiple groups difference sclerotin state, the assessment packet includes vital sign data, posture data, bone density data and life habit data, and the assessment data are numeric form；The assessment data for pre-processing every group of user obtain multiple groups criterion evaluation data；The multiple groups criterion evaluation data are classified, formed have bone density data labeled data collection and without bone density data without labeled data collection；Based on the labeled data collection and sclerotin status criteria training sorter model, sclerotin status assessment prescheme is obtained；And based on sclerotin status assessment prescheme described in the no continuous repetitive exercise of labeled data collection, sclerotin status assessment model is obtained；The sclerotin status assessment model is used to export the current sclerotin state of user according to the assessment data of user.
Description
Technical field
The present invention relates to health state evaluation technical field, particularly relates to a kind of sclerotin state based on semi-supervised learning and comment
Estimate the construction method of model.
Background technique
Sclerotin state influences personal daily life very big.Sclerotin state if there is exception, such as sclerotin deficiency and
Osteoporosis etc. may all cause bone strength to decline, and generate biggish harm.It is easy to send out in minor trauma or daily routines
Raw fracture, i.e. fragility fractures, this type fracture are difficult to rehabilitation, disable, lethality height, often decline quality of life, labour's funeral
It loses, has aggravated the financial burden of a people and society.
As China human mortality aging aggravates, the state of earlier evaluations bone health, and make evaluated person according to estimating
Skeletal health, by reinforcing the protection of daily bone health and habit of making the life better, for preventing or avoiding fragility fractures
Have great importance.
The appraisal procedure of current bone health state is broadly divided into Traditional measurements method and the assessment based on machine learning
Method.Traditional appraisal procedure has that assessment factor is too simple, does not account for and is mutually related between different risk factors
Effect can not find the potential risk the etc. when combination of weak correlative factor.Appraisal procedure based on machine learning is mostly unsupervised machine
Device study and Supervised machine learning, unsupervised machine learning are mostly qualitative analysis, can not provide assessment result to accurate quantitative analysis
Accuracy；That there are models is relatively simple for Supervised machine learning, can not the feature to data be accurately fitted, and can not
Data characteristics itself is studied, can not accurate evaluation sclerotin state the problem of.
Summary of the invention
In view of this, it is an object of the invention to propose that one kind can construct more perfect sclerotin status assessment model side
Method.
Based on a kind of above-mentioned purpose building side of the sclerotin status assessment model based on semi-supervised learning provided by the invention
Method characterized by comprising
The assessment data of the user of multiple groups difference sclerotin state are extracted, the assessment packet includes vital sign data, body
State data, bone density data and life habit data, the assessment data are numeric form；
The assessment data for pre-processing every group of user obtain multiple groups criterion evaluation data；
The multiple groups criterion evaluation data are classified, the labeled data collection with bone density data is formed and are not had
Bone density data without labeled data collection；
Based on the labeled data collection and sclerotin status criteria training sorter model, the pre- mould of sclerotin status assessment is obtained
Type；And based on sclerotin status assessment prescheme described in the no continuous repetitive exercise of labeled data collection, sclerotin status assessment is obtained
Model；The sclerotin status assessment model is used to export the current sclerotin state of user according to the assessment data of user.
Described in one of the embodiments, that the multiple groups criterion evaluation data are classified, being formed has bone density
The labeled data collection of data and without bone density data include: without the step of labeled data collection
According in every group of criterion evaluation data, if there are bone density data, multiple groups assessment data are divided into close with bone
The labeled data collection of degree and without bone density without labeled data collection；The labeled data concentrates the assessment with first group of number
Data, and the assessment data with second group of number are concentrated without labeled data, the value of first group of number and second group of number
Value is the value of the multiple groups with value.
It is described in one of the embodiments, to be commented based on sclerotin state described in the no continuous repetitive exercise of labeled data collection
Estimating prescheme includes:
By inputting in the sclerotin status assessment prescheme without labeled data collection for second group of number, second group of number is obtained
Assessment sclerotin state without labeled data collection and assessment confidence level, the assessment confidence level and the assessment sclerotin state pair
It answers；
The assessment confidence level is compared with preset confidence level, according to comparison result, by second group of number without mark
Note data set is divided into the data set of the data set of the high confidence level of third group number and the low confidence of the 4th group of number；Wherein, institute
State the value of third group number and the value of the 4th group of number and value be second group of number value；
The data set training assessment prescheme of high confidence level based on the third group number, updates assessment prescheme
Parameter；
The pre- mould of sclerotin status assessment after the data set of the low confidence of the 4th group of number to be inputted to the undated parameter
In type, the assessment sclerotin state and assessment confidence level of the data set of the low confidence of the 4th group of number are obtained, and confidence level will be assessed
It is compared with preset confidence level, according to structure is compared, the data set of the low confidence of the 4th group of number is carried out again
It divides, and updates the sclerotin status assessment prescheme after undated parameter based on the data set of gained high confidence level, until gained is low
The data set of confidence level is empty set.
The assessment confidence level refers to the general of the assessment sclerotin status categories that prediction obtains in one of the embodiments,
Rate；The data set of high confidence level is the data set assessed confidence level and be greater than preset confidence level, and the data set of low confidence is to comment
Estimate the data set that confidence level is less than preset confidence level, the preset confidence level is set as 0.8.
The assessment packet of every group of user of the pretreatment includes in one of the embodiments:
The assessment data of every group of user, the data lacked in every group of assessment data of completion are analyzed, and remove the data of mistake,
Obtain the Pre-Evaluation data of every group of user；
In the Pre-Evaluation data for calculating every group of user, the information gain of single datum, the value for retaining information gain is big
In the data of threshold value, the value for removing information gain is less than the data of threshold value；
It is greater than in the data of threshold value in the value of the information gain of reservation, chooses associated data and be combined, reservation group
It closes the data obtained and removes the data being combined, obtain the criterion evaluation data of every group of user.
Whether the vital sign data includes: gender, age, age of meuopause and suffers from one of the embodiments,
There is the numerical value of disease and Long-term taking medicine；The posture data include: height, weight, bust, bust, waistline, hip circumference, brachium, body
The numerical value that rouge rate, femur are long and shin bone is long；The bone density data include: the number of bone density value and Bone mineral density position
Value；The life habit data include: native place, ancestral home life-time, are engaged in job category, daily sunshine duration, daily exercise
Duration, whether smoke and smoke and daily amount of smoking, whether drink and quantity of drinking, daily beverage preference, whether there is or not poisoning history,
Whether replenish the calcium, whether there is or not the numerical value of bone related surgical history and medical history.
The construction device of the present invention also provides a kind of sclerotin status assessment model based on semi-supervised learning, comprising:
Data extraction module, the assessment data of the user for extracting multiple groups difference sclerotin state, the assessment packet
Vital sign data, posture data, bone density data and life habit data are included, the assessment data are numeric form；
Preprocessing module obtains multiple groups criterion evaluation data for pre-processing the assessment data of every group of user；
Categorization module forms the mark with bone density data for the multiple groups criterion evaluation data to be classified
Data set and without bone density data without labeled data collection；
Training module, for obtaining sclerotin based on the labeled data collection and sclerotin status criteria training sorter model
Status assessment prescheme；And based on sclerotin status assessment prescheme described in the no continuous repetitive exercise of labeled data collection, obtain
Sclerotin status assessment model；The sclerotin status assessment model is used to export the current sclerotin of user according to the assessment data of user
State.
The present invention also provides a kind of sclerotin state evaluating method based on semi-supervised learning, comprising:
Extract the assessment data of user；
The assessment data are inputted into the sclerotin status assessment model pre-established, obtain the current sclerotin state of user；
Export sclerotin condition improvement suggestion corresponding with the sclerotin state outcome that user is current；
The sclerotin status assessment model is constructed using method as described above；
The sclerotin condition improvement suggestion includes movement, diet, the suggestion of health-care and the Bad habits that avoid.
The sclerotin status assessment model construction equipment based on semi-supervised learning that the present invention also provides a kind of, including processor,
Memory, bus and storage are on a memory and the computer program that can run on a processor；
Wherein, the processor, memory complete mutual communication by the bus；
The processor is used to realize sclerotin status assessment model building method as described above when the computer program
Step.
The present invention also provides a kind of computer readable storage medium, sclerotin is stored on the computer readable storage medium
Status assessment model construction program, the sclerotin status assessment model construction program realize bone as described above when being executed by processor
The step of matter status assessment model building method.
From the above it can be seen that the building of the sclerotin status assessment model provided by the invention based on semi-supervised learning
Method, according in every group of criterion evaluation data, if there are bone density data, and multiple groups assessment data are divided into bone density
Labeled data collection and training without labeled data collection, and based on the labeled data collection and sclerotin status criteria without bone density
Sorter model obtains sclerotin status assessment prescheme；And the sclerotin status assessment prescheme will be inputted without labeled data collection
In, assessment sclerotin state and assessment confidence level corresponding with assessment sclerotin state are obtained, by assessment confidence level and preset confidence
Degree compares, and is divided into data set of the assessment confidence level greater than the high confidence level of preset confidence level and assessment confidence level less than pre-
If confidence level low confidence data set, then constantly go update sclerotin status assessment pre- by the data set of high confidence level
The parameter of model forms sclerotin status assessment model until the data set of low confidence is sky.So that final gained sclerotin
Status assessment model, it is more perfect, it can more accurately identify the sclerotin state of the user without bone density data.
Detailed description of the invention
Fig. 1 is the process of the construction method of the sclerotin status assessment model based on semi-supervised learning of the embodiment of the present invention
Figure；
Fig. 2 is the flow chart of the sclerotin state evaluating method based on semi-supervised learning of the embodiment of the present invention.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, below in conjunction with specific embodiment, and reference
Attached drawing, the present invention is described in more detail.
Referring to Fig. 1, the embodiment of the present invention provides a kind of building of sclerotin status assessment model based on semi-supervised learning
Method, comprising:
Step S100, extracts the assessment data of the user of multiple groups difference sclerotin state, and the assessment packet includes life entity
Data, posture data, bone density data and life habit data are levied, the assessment data are numeric form；
Step S200 pre-processes the assessment data of every group of user, obtains multiple groups criterion evaluation data；
Step S300 classifies the multiple groups criterion evaluation data, forms the labeled data with bone density data
Collection and without bone density data without labeled data collection；
Step S400 obtains sclerotin state based on the labeled data collection and sclerotin status criteria training sorter model
Assess prescheme；And based on sclerotin status assessment prescheme described in the no continuous repetitive exercise of labeled data collection, sclerotin is obtained
Status assessment model；The sclerotin status assessment model is used to export the current sclerotin shape of user according to the assessment data of user
State.
It further include design seismic wave questionnaire before step S100, and by propagating the modes such as paper roll under electronics volume and line on line,
The user of different sclerotin states is organized to fill in.Different sclerotin states refer mainly to the differences such as bone health, sclerotin deficiency, osteoporosis
The group of sclerotin state.The questionnaire is mainly designed according to the standard of sclerotin state, combines involved in pertinent literature it
He and the stronger factor of sclerotin state correlation.
Questionnaire acquisition information specifically include that vital sign information, mainly include gender, the age, age of meuopause with
And whether with disease and Long-term taking medicine etc.；Posture information mainly includes height, weight, bust, bust, waistline, hip circumference, arm
Length, body fat rate, femur are long and shin bone is long etc.；Bone density information mainly includes bone density value and Bone mineral density position；Life
Habit information mainly includes native place, ancestral home life-time, is engaged in job category, daily sunshine duration, tempers duration daily, be
No smoking and smoking and daily amount of smoking, whether drink and quantity of drinking, daily beverage preference, whether there is or not poisoning history, whether mend
Calcium, whether there is or not bone related surgical history and medical histories etc..
In step S100, the assessment data of the user for extracting multiple groups difference sclerotin state are referred specifically to, using word
Symbol knows method for distinguishing, identifies the every terms of information in more parts of questionnaires, extracts every assessment data in questionnaire.It should
Illustrate, the form of resulting assessment data is numeric form.One group of assessment data refers to be extracted from a questionnaire
All numerical value.
Specifically, the vital sign data includes: gender, age, age of meuopause and whether with disease and long-term
The numerical value such as medication；The posture data include: that height, weight, bust, bust, waistline, hip circumference, brachium, body fat rate, femur are long
And the numerical value such as shin bone length；The bone density data include: the numerical value such as bone density value and Bone mineral density position；The life is practised
Property data include: native place, ancestral home life-time, be engaged in job category, daily sunshine duration, daily take exercise duration, whether smoke
And smoking and daily amount of smoking, whether drink and quantity of drinking, daily beverage preference, whether there is or not poisoning history, whether replenish the calcium, whether there is or not
The numerical value such as bone related surgical history and medical history.
In step S200, the step of pre-processing the assessment data of every group of user, may include:
S210 analyzes the assessment data of every group of user, searches the missing that whether there is certain item data in assessment data, if
There are certain item datas situations such as filling in mistake.
When discovery exist when the missing of certain item data, to the item data using corresponding numerical value progress completion.For example, when asking
When gender is male in vital sign data in volume, age of meuopause shortage of data；When whether being inhaled in life habit data in questionnaire
When cigarette is no, daily smoking capacity shortage of data；In the case of similar, the data of missing are extended this as into numerical value -1.
When discovery there are certain item datas when filling in mistake, which is removed.For example, when life entity in questionnaire
When gender is male in sign data, age of meuopause data are filled in；When ancestral home life-time is big in life habit data in questionnaire
In the age；In the case of similar, the data of mistake are removed.
S220, will by step S210 processing after the data obtained, by calculate, carry out the removal of data.Predominantly calculate
The information gain of every item data, and all data in every group of questionnaire is subjected to descending arrangement according to the value of information gain,
Retain in list preceding 80~90% data.Specifically, the information gain for calculating every item data calculates separately every item data and deposits
The value of information content when with missing in tune difference Questionnaire systems, and the difference of the two is calculated, which is the value of information gain.
S230 in the data obtained, chooses associated data and is combined, formed new after step S220 processing
Data.After combination, retain new data, remove the data being combined, forms criterion evaluation data.Combining form can be
Being added may be the operation such as multiplication.Such as length of smoking and daily smoking capacity are associated data, by the operation of multiplication, are formed total
The data of smoking capacity.Then in this group of data, length of smoking and daily smoking capacity are removed, total smoking capacity is retained.
After pretreatment, assessment data can be made more accurate, it is more efficient, the efficiency of model construction is improved, with
And the precision of gained model.
In step S300, it will classify through criterion evaluation data obtained by step S200, refer to according to every group of criterion evaluation
In data, if there are bone density data, and multiple groups assessment data are divided into two group data sets.And by the data with bone density
Collection is named as labeled data collection, and the data set for not having bone density is named as no labeled data collection.Labeled data concentrates tool
There are the assessment data of first group of number, and concentrates the assessment data with second group of number without labeled data.That is, described first
Group number value and second group of number value and value for the multiple groups value.
In step S400, described the step of sorter model is trained based on the labeled data collection and sclerotin status criteria
In, LR (this spy of logic) regression model can be used.It should be noted that the sorter model is not unique, as long as can
It is returned, SVM (support vector machines), NN (nerve net with being ok for semi-supervised learning algorithm, such as LR (this spy of logic)
Network), Decision Tree (decision tree), RF (random forest), xgboost (extreme gradient promotion) etc..
The step of this special regression model obtains sclerotin status assessment prescheme using logic specifically includes, in linear regression
On the basis of, all data that the labeled data is concentrated carries out linear combination, then combined result is passed through one layer of sigmoid
Function Mapping at the result is that 1 or 0 probability.Iteration by gradient descent method Jing Guo certain step number obtains the ginseng of LR model
It counts to get sclerotin status assessment prescheme is arrived.It specifically, is on the basis of linear regression, by the phase of labeled data concentration
The data of same type carry out linear combination, then by after linear combination vital sign data, posture data, bone density data and
Life habit data are corresponding with sclerotin state.
Specifically, the expression formula of LR regression model is this number of writing of the logic of parametrization, as shown in formula (I)
Wherein, x is the item data that labeled data is concentrated, and θ is weight vector, that is, needs the parameter of trained x, θTX is
For the linear combination for the item data that labeled data is concentrated.
Specific training process can be with are as follows: by combined result by one layer of sigmoid Function Mapping at the result is that 1 or
0 or probability, given one group of criterion evaluation data are referred to, by shown in objective function shown in formula (II) and formula (III)
Sigmoid function, solve parameter θ, can be obtained the group assessment data prediction result be 1 probability, if probability is greater than
0.5, then show to belong to the category, if showing to be not belonging to the category less than 0.5.Classification represents any in sclerotin state
A kind of state.
That is, gained sclerotin status assessment prescheme can export the current bone of user according to the assessment data of user
Matter state.
It is described to include: based on sclerotin status assessment prescheme described in the no continuous repetitive exercise of labeled data collection
S421 inputs second group of number in the sclerotin status assessment prescheme without labeled data collection, obtains the
The assessment sclerotin state and assessment confidence level without labeled data collection of two groups of numbers.Specifically, every group of number without labeled data collection all
With assessment sclerotin state and assessment confidence level corresponding with the evaluation status.
The assessment confidence level is compared S422 with preset confidence level, according to comparison result, by second group of number
No labeled data collection is divided into the data set of the data set of the high confidence level of third group number and the low confidence of the 4th group of number；Its
In, the value of the value of the third group number and the 4th group of number and value for second group of number value；
Specifically, assessment confidence level refers to the probability for the assessment sclerotin status categories that prediction obtains.The number of high confidence level
It is greater than the data set of preset confidence level for assessment confidence level according to collection, the data set of low confidence is less than default for assessment confidence level
Confidence level data set, preset confidence level can be set to 0.8.S423, the high confidence level based on the third group number
The data set training assessment prescheme, updates the parameter of assessment prescheme；
S424, the sclerotin status assessment after the data set of the low confidence of the 4th group of number to be inputted to the undated parameter
In prescheme, the assessment sclerotin state and assessment confidence level of the data set of the low confidence of the 4th group of number are obtained, and assessment is set
Reliability is compared with preset confidence level, and according to structure is compared, the data set of the low confidence of the 4th group of number is carried out
It divides again, and updates the sclerotin status assessment prescheme after undated parameter based on the data set of gained high confidence level, until institute
The data set for obtaining low confidence is empty set.
Step S421 to S424 is it is to be understood that input the sclerotin status assessment without labeled data collection for second group of number
In prescheme, classify, the data set and assessment sclerotin state for being divided into assessment sclerotin state high confidence level with a high credibility can
Then the data set of the low low confidence of reliability constantly goes to update sclerotin status assessment prescheme by the data set of high confidence level
Parameter, until all assessment sclerotin states all there is high confidence level, that is, form sclerotin status assessment model.
Preferably, in the parameter of more new model, regular terms is introduced in step S423 and S424, it can be to avoid over-fitting
The problem of.Over-fitting is prevented by punishing excessive parameter.
The embodiment of the present invention also provides one kind, the construction device of the sclerotin status assessment model based on semi-supervised learning, packet
It includes:
Data extraction module, the assessment data of the user for extracting multiple groups difference sclerotin state, the assessment packet
Vital sign data, posture data, bone density data and life habit data are included, the assessment data are numeric form；
Preprocessing module obtains multiple groups criterion evaluation data for pre-processing the assessment data of every group of user；
Categorization module forms the mark with bone density data for the multiple groups criterion evaluation data to be classified
Data set and without bone density data without labeled data collection；
Training module, for obtaining sclerotin based on the labeled data collection and sclerotin status criteria training sorter model
Status assessment prescheme；And based on sclerotin status assessment prescheme described in the no continuous repetitive exercise of labeled data collection, obtain
Sclerotin status assessment model；The sclerotin status assessment model is used to export the current sclerotin of user according to the assessment data of user
State.
The construction method of sclerotin status assessment model provided by the invention based on semi-supervised learning, is commented according to every group of standard
Estimate in data, if there are bone density data, and multiple groups assessment data are divided into the labeled data collection with bone density and are not had
Bone density trains sorter model without labeled data collection, and based on the labeled data collection and sclerotin status criteria, obtains bone
Matter status assessment prescheme；And will be inputted in the sclerotin status assessment prescheme without labeled data collection, obtain assessment sclerotin shape
Assessment confidence level is compared with preset confidence level, is divided into and commenting by state and assessment confidence level corresponding with assessment sclerotin state
The data set for estimating confidence level greater than the high confidence level of preset confidence level is set with assessment confidence level less than the low of preset confidence level
Then the data set of reliability constantly goes the parameter for updating sclerotin status assessment prescheme by the data set of high confidence level, until
The data set of low confidence is sky, that is, forms sclerotin status assessment model.So that final gained sclerotin status assessment model, more
It adds kind, can more accurately identify the sclerotin state of the user without bone density data.
Referring to Fig. 2, the embodiment of the present invention also provides a kind of sclerotin state evaluating method based on semi-supervised learning, packet
It includes:
S500 extracts the assessment data of user；
The assessment data are inputted the sclerotin status assessment model pre-established, obtain the current sclerotin of user by S600
State；
S700 exports sclerotin condition improvement suggestion corresponding with the sclerotin state outcome that user is current；
Wherein, in step S600, obtained by the sclerotin status assessment model should be constructed with the aforedescribed process.
In step S700, sclerotin condition improvement suggestion includes movement, diet, the suggestion of health-care and the bad life that avoids
Habit living.
The embodiment of the present invention also provides a kind of building equipment of sclerotin status assessment model based on semi-supervised learning,
Include: processor, memory, bus and storage on a memory and the computer program that can run on a processor；
Wherein, the processor, memory complete mutual communication by the bus；
The processor realizes method as described above when executing the computer program.
The function of each functional module of the building equipment of sclerotin status assessment model can be according to upper described in the embodiment of the present invention
The method specific implementation in embodiment of the method is stated, specific implementation process is referred to the associated description of above method embodiment,
Details are not described herein again.
The embodiment of the present invention also provides a kind of non-transient computer readable storage medium, is stored with bone on the storage medium
Matter status assessment model construction program, the computer program realize method as described above when being executed by processor.
The function of each functional module of computer readable storage medium described in the embodiment of the present invention can be according to the above method
Method specific implementation in embodiment, specific implementation process are referred to the associated description of above method embodiment, herein not
It repeats again.
It should be noted that all statements for using " first " and " second " are for differentiation two in the embodiment of the present invention
The non-equal entity of a same names or non-equal parameter, it is seen that " first " " second " only for the convenience of statement, does not answer
It is interpreted as the restriction to the embodiment of the present invention, subsequent embodiment no longer illustrates this one by one.
It should be understood by those ordinary skilled in the art that: the discussion of any of the above embodiment is exemplary only, not
It is intended to imply that the scope of the present disclosure (including claim) is limited to these examples；Under thinking of the invention, above embodiments
Or can also be combined between the technical characteristic in different embodiments, step can be realized with random order, and be existed such as
Many other variations of the upper different aspect of the invention, for simplicity, they are not provided in details.
In addition, to simplify explanation and discussing, and in order not to obscure the invention, it can in provided attached drawing
It is connect with showing or can not show with the well known power ground of integrated circuit (IC) chip and other components.In addition, can
To show device in block diagram form, to avoid obscuring the invention, and this has also contemplated following facts, i.e., about
The details of the embodiment of these block diagram arrangements be height depend on will implement platform of the invention (that is, these details are answered
When within the scope of the understanding for being completely in those skilled in the art).Elaborating detail (for example, circuit) to describe the present invention
Exemplary embodiment in the case where, it will be apparent to those skilled in the art that can these be specific thin
Implement the present invention in the case where section or in the case that these details change.Therefore, these descriptions should be considered as
Bright property rather than it is restrictive.
Although having been incorporated with specific embodiments of the present invention, invention has been described, according to retouching for front
It states, many replacements of these embodiments, modifications and variations will be apparent for those of ordinary skills.Example
Such as, other machine learning algorithms (for example, LR (logic this special) recurrence, SVM (support vector machines), NN (neural network),
Decision Tree (decision tree), RF (random forest), xgboost (extreme gradient promotion)) discussed reality can be used
Apply example.
The embodiment of the present invention be intended to cover fall into all such replacements within the broad range of appended claims,
Modifications and variations.Therefore, all within the spirits and principles of the present invention, any omission, modification, equivalent replacement, the improvement made
Deng should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of construction method of the sclerotin status assessment model based on semi-supervised learning characterized by comprising
The assessment data of the user of multiple groups difference sclerotin state are extracted, the assessment packet includes vital sign data, posture number
According to, bone density data and life habit data, the assessment data are numeric form；
The assessment data for pre-processing every group of user obtain multiple groups criterion evaluation data；
The multiple groups criterion evaluation data are classified, formed have bone density data labeled data collection and do not have bone it is close
Degree evidence without labeled data collection；
Based on the labeled data collection and sclerotin status criteria training sorter model, sclerotin status assessment prescheme is obtained；And
Based on sclerotin status assessment prescheme described in the no continuous repetitive exercise of labeled data collection, sclerotin status assessment model is obtained；
The sclerotin status assessment model is used to export the current sclerotin state of user according to the assessment data of user.
2. the construction method of the sclerotin status assessment model according to claim 1 based on semi-supervised learning, feature exist
In, it is described that the multiple groups criterion evaluation data are classified, it forms the labeled data collection with bone density data and does not have
Bone density data include: without the step of labeled data collection
According in every group of criterion evaluation data, if there are bone density data, and multiple groups assessment data are divided into bone density
Labeled data collection and without bone density without labeled data collection；The labeled data concentrates the assessment number with first group of number
According to, and the assessment data with second group of number are concentrated without labeled data, the value of the value of first group of number and second group of number
And value be the multiple groups value.
3. the construction method of the sclerotin status assessment model according to claim 2 based on semi-supervised learning, feature exist
In described to include: based on sclerotin status assessment prescheme described in the no continuous repetitive exercise of labeled data collection
By inputting in the sclerotin status assessment prescheme without labeled data collection for second group of number, the nothing of second group of number is obtained
The assessment sclerotin state and assessment confidence level, the assessment confidence level of labeled data collection are corresponding with the assessment sclerotin state；
The assessment confidence level is compared with preset confidence level, according to comparison result, by second group of number without mark number
The data set of the data set of the high confidence level of third group number and the low confidence of the 4th group of number is divided into according to collection；Wherein, described
The value of the value of three groups of numbers and the 4th group of number and value for second group of number value；
The data set training assessment prescheme of high confidence level based on the third group number, updates the ginseng of assessment prescheme
Number；
In sclerotin status assessment prescheme after the data set of the low confidence of the 4th group of number to be inputted to the undated parameter,
The assessment sclerotin state and assessment confidence level of the data set of the low confidence of the 4th group of number are obtained, and confidence level will be assessed and preset
Confidence level be compared, according to compare structure, the data set of the low confidence of the 4th group of number is divided again, and
Data set based on gained high confidence level updates the sclerotin status assessment prescheme after undated parameter, until gained low confidence
Data set is empty set.
4. the construction method of the sclerotin status assessment model according to claim 3 based on semi-supervised learning, feature exist
In the assessment confidence level refers to the probability for the assessment sclerotin status categories that prediction obtains；The data set of high confidence level is to comment
Estimate the data set that confidence level is greater than preset confidence level, the data set of low confidence is that assessment confidence level is less than preset confidence level
Data set, the preset confidence level is set as 0.8.
5. the construction method of the sclerotin status assessment model according to claim 1 based on semi-supervised learning, feature exist
In the assessment packet of every group of user of the pretreatment includes:
The assessment data of every group of user, the data lacked in every group of assessment data of completion are analyzed, and remove the data of mistake, are obtained
The Pre-Evaluation data of every group of user；
In the Pre-Evaluation data for calculating every group of user, the information gain of single datum, and will be each in every group of assessment data
Item data carries out descending arrangement according to the value of information gain, retains in sorted lists preceding 80~90% data；
It in the data of reservation, chooses associated data and is combined, retain combination the data obtained and remove and be combined
Data obtain the criterion evaluation data of every group of user.
6. the construction method of the sclerotin status assessment model according to claim 1 based on semi-supervised learning, feature exist
In the vital sign data includes: gender, age, age of meuopause and the numerical value for whether suffering from disease and Long-term taking medicine；Institute
Stating posture data includes: height, weight, bust, bust, waistline, hip circumference, brachium, body fat rate, the number that femur is long and shin bone is long
Value；The bone density data include: the numerical value of bone density value and Bone mineral density position；The life habit data include: nationality
It passes through, ancestral home life-time, is engaged in job category, daily sunshine duration, tempers duration daily, whether smoke and smoke and inhale daily
Cigarette quantity, whether drink and quantity of drinking, daily beverage preference, whether there is or not poisoning history, whether replenish the calcium, whether there is or not bone related surgical histories
And the numerical value of medical history.
7. a kind of construction device of the sclerotin status assessment model based on semi-supervised learning characterized by comprising
Data extraction module, the assessment data of the user for extracting multiple groups difference sclerotin state, the assessment packet include life
Sign data, posture data, bone density data and life habit data are ordered, the assessment data are numeric form；
Preprocessing module obtains multiple groups criterion evaluation data for pre-processing the assessment data of every group of user；
Categorization module forms the labeled data with bone density data for the multiple groups criterion evaluation data to be classified
Collection and without bone density data without labeled data collection；
Training module, for obtaining sclerotin state based on the labeled data collection and sclerotin status criteria training sorter model
Assess prescheme；And based on sclerotin status assessment prescheme described in the no continuous repetitive exercise of labeled data collection, sclerotin is obtained
Status assessment model；The sclerotin status assessment model is used to export the current sclerotin shape of user according to the assessment data of user
State.
8. a kind of sclerotin state evaluating method based on semi-supervised learning characterized by comprising
Extract the assessment data of user；
The assessment data are inputted into the sclerotin status assessment model pre-established, obtain the current sclerotin state of user；
Export sclerotin condition improvement suggestion corresponding with the sclerotin state outcome that user is current；
Sclerotin status assessment model application any one of claims 1 to 6 method is constructed；
The sclerotin state health Improving advice include movement, diet, health-care suggestion and avoid Bad habits.
9. a kind of sclerotin status assessment model construction equipment based on semi-supervised learning, which is characterized in that including processor, storage
Device, bus and storage are on a memory and the computer program that can run on a processor；
Wherein, the processor, memory complete mutual communication by the bus；
The processor is used to realize the sclerotin state model structure as described in any one of claim 1 to 6 when the computer program
The step of construction method.
10. a kind of computer readable storage medium, which is characterized in that be stored with sclerotin shape on the computer readable storage medium
State assessment models construction procedures realize such as claim 1 when the sclerotin status assessment model construction program is executed by processor
The step of to any one of 6 sclerotin status assessment model building method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910682966.0A CN110491520A (en) | 2019-07-26 | 2019-07-26 | A kind of construction method of the sclerotin status assessment model based on semi-supervised learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910682966.0A CN110491520A (en) | 2019-07-26 | 2019-07-26 | A kind of construction method of the sclerotin status assessment model based on semi-supervised learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110491520A true CN110491520A (en) | 2019-11-22 |
Family
ID=68548528
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910682966.0A Pending CN110491520A (en) | 2019-07-26 | 2019-07-26 | A kind of construction method of the sclerotin status assessment model based on semi-supervised learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110491520A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111402230A (en) * | 2020-03-16 | 2020-07-10 | 徐州医科大学 | Femoral medullary cavity morphological feature analysis method based on Monte Carlo method |
| CN111755129A (en) * | 2020-06-30 | 2020-10-09 | 山东大学 | Multi-mode osteoporosis layering early warning method and system |
| CN112700880A (en) * | 2020-12-31 | 2021-04-23 | 杭州依图医疗技术有限公司 | Optimization method, training method, model, processing device, and storage medium |
| WO2021217867A1 (en) * | 2020-04-29 | 2021-11-04 | 平安科技（深圳）有限公司 | Xgboost-based data classification method and apparatus, computer device, and storage medium |
| US20220059188A1 (en) * | 2020-08-24 | 2022-02-24 | Kpn Innovations, Llc. | Method of and system for identifying and ameliorating body degradations |
| CN116564526A (en) * | 2023-07-11 | 2023-08-08 | 天津市滨海新区中医医院 | Bone health assessment method, device and storage medium |
| CN116631618A (en) * | 2023-04-28 | 2023-08-22 | 南方医科大学南方医院 | Osteoporosis risk monitoring system, osteoporosis risk monitoring method, electronic equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104318242A (en) * | 2014-10-08 | 2015-01-28 | 中国人民解放军空军工程大学 | High-efficiency SVM active half-supervision learning algorithm |
| CN106228183A (en) * | 2016-07-18 | 2016-12-14 | 北京邮电大学 | A kind of semi-supervised learning sorting technique and device |
| CN109256212A (en) * | 2018-08-17 | 2019-01-22 | 上海米因医疗器械科技有限公司 | Bone health assessment models construction method, device, equipment, medium and appraisal procedure |
| CN109460914A (en) * | 2018-11-05 | 2019-03-12 | 云南大学 | Method is determined based on the bridge health grade of semi-supervised error correction study |
-
2019
- 2019-07-26 CN CN201910682966.0A patent/CN110491520A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104318242A (en) * | 2014-10-08 | 2015-01-28 | 中国人民解放军空军工程大学 | High-efficiency SVM active half-supervision learning algorithm |
| CN106228183A (en) * | 2016-07-18 | 2016-12-14 | 北京邮电大学 | A kind of semi-supervised learning sorting technique and device |
| CN109256212A (en) * | 2018-08-17 | 2019-01-22 | 上海米因医疗器械科技有限公司 | Bone health assessment models construction method, device, equipment, medium and appraisal procedure |
| CN109460914A (en) * | 2018-11-05 | 2019-03-12 | 云南大学 | Method is determined based on the bridge health grade of semi-supervised error correction study |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111402230A (en) * | 2020-03-16 | 2020-07-10 | 徐州医科大学 | Femoral medullary cavity morphological feature analysis method based on Monte Carlo method |
| WO2021217867A1 (en) * | 2020-04-29 | 2021-11-04 | 平安科技（深圳）有限公司 | Xgboost-based data classification method and apparatus, computer device, and storage medium |
| CN111755129A (en) * | 2020-06-30 | 2020-10-09 | 山东大学 | Multi-mode osteoporosis layering early warning method and system |
| CN111755129B (en) * | 2020-06-30 | 2023-12-19 | 山东大学 | Multi-mode osteoporosis layering early warning method and system |
| US20220059188A1 (en) * | 2020-08-24 | 2022-02-24 | Kpn Innovations, Llc. | Method of and system for identifying and ameliorating body degradations |
| US11798652B2 (en) * | 2020-08-24 | 2023-10-24 | Kpn Innovations, Llc. | Method of and system for identifying and ameliorating body degradations |
| CN112700880A (en) * | 2020-12-31 | 2021-04-23 | 杭州依图医疗技术有限公司 | Optimization method, training method, model, processing device, and storage medium |
| CN116631618A (en) * | 2023-04-28 | 2023-08-22 | 南方医科大学南方医院 | Osteoporosis risk monitoring system, osteoporosis risk monitoring method, electronic equipment and storage medium |
| CN116564526A (en) * | 2023-07-11 | 2023-08-08 | 天津市滨海新区中医医院 | Bone health assessment method, device and storage medium |
| CN116564526B (en) * | 2023-07-11 | 2023-09-19 | 天津市滨海新区中医医院 | Bone health assessment method, device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110491520A (en) | A kind of construction method of the sclerotin status assessment model based on semi-supervised learning | |
| CN106682616B (en) | Method for recognizing neonatal pain expression based on two-channel feature deep learning | |
| Sk et al. | Coronary Heart Disease Prediction and Classification using Hybrid Machine Learning Algorithms | |
| CN107977671A (en) | A kind of tongue picture sorting technique based on multitask convolutional neural networks | |
| CN112890816A (en) | Health index scoring method and device for individual user | |
| KR101701114B1 (en) | Body composition measuring apparatus and server providing a personalized information | |
| CN110491479A (en) | A kind of construction method of sclerotin status assessment model neural network based | |
| CN110956626B (en) | Image-based prognosis evaluation method and device | |
| CN110706822B (en) | Health management method based on logistic regression model and decision tree model | |
| CN108304887A (en) | Naive Bayesian data processing system and method based on the synthesis of minority class sample | |
| CN107180283A (en) | A kind of behavior prediction system and method for being in hospital again combined based on optimal characteristics | |
| CN106682412A (en) | Diabetes prediction method based on medical examination data | |
| CN108717693A (en) | A kind of optic disk localization method based on RPN | |
| CN106529110A (en) | Classification method and equipment of user data | |
| CN112614590A (en) | Machine learning-based elderly disability risk prediction method and system | |
| CN109410074A (en) | Intelligent core protects method and system | |
| Popkes et al. | Interpretable outcome prediction with sparse Bayesian neural networks in intensive care | |
| Prabu et al. | Grid Search for Predicting Coronary Heart Disease by Tuning Hyper-Parameters. | |
| CN108922621A (en) | A kind of appraisal procedure and device of the grade that disables | |
| CN114943629A (en) | Health management and health care service system and health management method thereof | |
| CN109934179A (en) | Human motion recognition method based on automated characterization selection and Ensemble Learning Algorithms | |
| CN112037888B (en) | Physiological health characteristic data monitoring method, device, equipment and storage medium | |
| CN109583272B (en) | Footprint system capable of acquiring living state of human body | |
| Ahmed et al. | A Review on the Detection Techniques of Polycystic Ovary Syndrome Using Machine Learning | |
| Moudani et al. | Intelligent decision support system for osteoporosis prediction |
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: 20191122 |