CN109102899A - Chinese medicine intelligent assistance system and method based on machine learning and big data - Google Patents
Chinese medicine intelligent assistance system and method based on machine learning and big data Download PDFInfo
- Publication number
- CN109102899A CN109102899A CN201810805208.9A CN201810805208A CN109102899A CN 109102899 A CN109102899 A CN 109102899A CN 201810805208 A CN201810805208 A CN 201810805208A CN 109102899 A CN109102899 A CN 109102899A
- Authority
- CN
- China
- Prior art keywords
- symptom
- disease
- disease type
- module
- diagnosis
- 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
- 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
-
- 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
- G16H20/10—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- Primary Health Care (AREA)
- Epidemiology (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Chemical & Material Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Medicinal Chemistry (AREA)
- Medical Treatment And Welfare Office Work (AREA)
- Medical Preparation Storing Or Oral Administration Devices (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention discloses a kind of Chinese medicine intelligent assistance system and method based on machine learning and big data, the system comprises: four methods of diagnosis information collection module: for collecting the related four methods of diagnosis data information of patient main suit；Four methods of diagnosis info conversion module: the four methods of diagnosis information for inputting doctor is cut into descriptive symptom language N；Symptom language matching module: for symptom language N to be converted to standardized sympotomatic set；Symptom weight identification module: for calculating the identification weighted value of various symptom according to magnanimity case data；Disease type matching module: for analyzing the characteristic of disease disease bit vector of main disease type and diagnosis；Prescription recommending module: for recommending prescription according to disease type.The strength of association of chief complaint language disease of tcm of the present invention by storage in the database, the possibility Chinese medicine disease type of patient can be quickly determined by a series of analysis method, then treatment prescription out is quickly recommended according to Chinese medicine disease type symptom, doctor is facilitated more accurately to diagnose, reliability is higher.
Description
Technical field
The present invention relates to Chinese medicine technical field of information processing, and in particular to a kind of Chinese medicine based on machine learning and big data
Intelligent assistance system and method.
Background technique
With the fast development based on data and information system, how to pass through information-based means and quickly improve primary care
The quality of medical care of unit by be the following basic medical unit and patient core demand.The four methods of diagnosis that system is acquired by realizing doctor
Use of the degree of association and algorithm of information and Chinese medical disease etc. solves the limitation of clinic doctor's self-technique, and with suggestion at
Side releases patient's pain.Currently, predominantly staying in theoretic or these technologies for the correlation research of main suit and disease
It can only be solved the problems, such as with method some portion of.
Summary of the invention
The purpose of the present invention is to provide a kind of Chinese medicine intelligent assistance system and method based on machine learning and big data,
Solve the problems, such as that how to be directed to disease of tcm by information-based means for different medical unit quickly recommends treatment prescription out.
In order to solve the above technical problems, the invention adopts the following technical scheme:
A kind of Chinese medicine intelligent assistance system based on machine learning and big data, the system comprises:
Four methods of diagnosis information collection module: for collecting the related four methods of diagnosis data information of patient main suit；
Four methods of diagnosis info conversion module: for passing through StandardSyndrom component after receiving patient's four methods of diagnosis information according to book
The comma or blank character write execute the movement of natural language word cutting, and the four methods of diagnosis information that doctor inputs is cut into descriptive symptom
Language N；
Symptom language matching module: for being converted four methods of diagnosis information using neural network algorithm by StandardSyndrom component
The obtained symptom language N of module is matched with symptom repertorie, is converted to standardized sympotomatic set；
Symptom weight identification module: for what is passed in the reception of DiagnosisSystem component from StandardSyndrom
After standardized sympotomatic set, the identification weighted value of various symptom is calculated according to magnanimity case data；
Disease type matching module: the Chinese medicine disease type for being learnt from magnanimity case data by DiagnosisSystem component
The identification weighted value progress integration calculating of the various symptom obtained with the weighted value of classical symptom with symptom weight identification module, and
Ranking is carried out according to the weight summation of the symptom in each disease type, main disease type is obtained and utilizes characteristic of disease disease position neural network model meter
Calculate the characteristic of disease disease bit vector of diagnosis；
Prescription recommending module: for obtained by PrescriptionAnalysis Assembly calculation the characteristic of disease disease position of each prescription to
Amount, then the disease of each prescription is passed through into PrescriptionDecision component for the characteristic of disease disease bit vector of each prescription and disease type again
The characteristic of disease disease bit vector of the diagnosis obtained with module goes to carry out similarity mode, finally using the highest prescription of matching degree as pushing away
It recommends.
Preferably, the system also includes data memory module, for storing magnanimity case data, the magnanimity case
Data include the characteristic of disease disease bit vector of the relation data of disease type and symptom, the candidate prescription under each disease type and each candidate prescription.
Preferably, the disease type matching module, is also used to according to primary symptom disease type matched in the module and various symptom
Identification weight analysis go out the primary symptom disease type appearance minor symptom disease type.
A kind of Chinese medicine intelligence householder method based on machine learning and big data, comprising the following steps:
S1: the related four methods of diagnosis data information of four methods of diagnosis information collection module collection patient main suit；
S2: four methods of diagnosis info conversion module is by StandardSyndrom component according to writing after receiving patient's four methods of diagnosis information
Comma or blank character execute natural language word cutting movement, by doctor input four methods of diagnosis information be cut into descriptive symptom language
N；
S3: symptom language matching module utilizes neural network algorithm by four methods of diagnosis information modulus of conversion by StandardSyndrom component
The obtained symptom language N of block is matched with symptom repertorie, is converted to standardized sympotomatic set；
S4: symptom weight identification module receives the mark passed over from StandardSyndrom in DiagnosisSystem component
After the sympotomatic set of standardization, the identification weighted value of various symptom is calculated according to magnanimity case data；
S5: Chinese medicine disease type that disease type matching module learn from magnanimity case data by DiagnosisSystem component and
The identification weighted value for the various symptom that the weighted value of classical symptom is obtained with symptom weight identification module carries out integration calculating, and presses
Ranking is carried out according to the weight summation of the symptom in each disease type, obtain main disease type and is calculated using characteristic of disease disease position neural network model
The characteristic of disease disease bit vector of the diagnosis out；
S6: prescription recommending module obtains the characteristic of disease disease bit vector of each prescription by PrescriptionAnalysis Assembly calculation,
The diagnosis for again being obtained the characteristic of disease disease bit vector of each prescription and disease type matching module by PrescriptionDecision component
Characteristic of disease disease bit vector go carry out similarity mode, finally using the highest prescription of matching degree as recommendation.
Preferably, the calculating of the identification weighted value of various symptom is calculated in the S4 step according to magnanimity case data
Mode is: the image that the relationship of symptom and disease type in magnanimity case data is characterized according to symptom, disease type is label is carried out
Storage, then calculates the matrix of weighted value according to TF-IDF algorithm.
Preferably, the calculation formula of the matrix of weighted value is calculated in the S4 step according to TF-IDF algorithm are as follows: TF-
IDF (symptom, disease type)=(symptom comes across the frequency/card type of the disease type in the Symptomatic total frequency of institute) (1+ is total by * LOG
Total disease type number/symptom appears in how many disease types).
Preferably, the S5 step further includes disease type matching module according to primary symptom disease type matched in the module and various
The identification weight analysis of symptom goes out the minor symptom disease type of primary symptom disease type appearance.
Compared with prior art, the beneficial effects of the present invention are:
The strength of association of chief complaint language disease of tcm of the present invention by storage in the database, passes through a series of analysis side
Method can quickly determine the possibility Chinese medicine disease type symptom of patient, then quickly recommend treatment prescription out according to Chinese medicine disease type symptom,
The insufficient doctor of experience is facilitated more accurately to diagnose, reliability is higher.
Detailed description of the invention
Fig. 1 is flow chart of the invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
For one embodiment of system of the invention, a kind of based on the Chinese medicine of machine learning and big data, intelligently auxiliary is
System, the system comprises: four methods of diagnosis information collection module: for collecting the related four methods of diagnosis data information of patient main suit；Four methods of diagnosis information
Conversion module: for passing through StandardSyndrom component after receiving patient's four methods of diagnosis information according to the comma or sky of writing
White symbol executes the movement of natural language word cutting, and the four methods of diagnosis information that doctor inputs is cut into descriptive symptom language N；Symptom language
With module: for being obtained four methods of diagnosis info conversion modules using neural network algorithm by StandardSyndrom component
Symptom language N is matched with symptom repertorie, is converted to standardized sympotomatic set；Symptom weight identification module: it is used for
After DiagnosisSystem component receives the standardized sympotomatic set passed over from StandardSyndrom, according to magnanimity
Case data calculate the identification weighted value of various symptom；Disease type matching module: for by DiagnosisSystem component from
The weighted value of the Chinese medicine disease type learnt and classical symptom obtains various with symptom weight identification module in magnanimity case data
The identification weighted value of symptom carries out integration calculating, and carries out ranking according to the weight summation of the symptom in each disease type, obtains main
Disease type and the characteristic of disease disease bit vector that diagnosis is calculated using characteristic of disease disease position neural network model；Prescription recommending module: for passing through
The characteristic of disease disease bit vector to each prescription of PrescriptionAnalysis Assembly calculation, then the disease of each prescription is passed through again
The characteristic of disease for the diagnosis that PrescriptionDecision component obtains the characteristic of disease disease bit vector of each prescription and disease type matching module
Sick bit vector goes to carry out similarity mode, finally using the highest prescription of matching degree as recommendation.In addition each component in the program
It can be carried out according to the function of realization customized.
Further, for another embodiment of system of the invention, the system also includes data memory modules, are used for
Magnanimity case data are stored, the magnanimity case data include the candidate prescription under the relation data of disease type and symptom, each disease type
And the characteristic of disease disease bit vector of candidate prescription.These data are all to carry out analysis and arrangement by a large amount of case data to be stored in
In database.
Further, for another embodiment of system of the invention, the disease type matching module is also used to according to the mould
The identification weight analysis of matched primary symptom disease type and various symptom goes out the minor symptom disease type of primary symptom disease type appearance in block.
Referring to Fig. 1, for one embodiment of method of the invention, a kind of Chinese medicine intelligence based on machine learning and big data
Energy householder method, comprising the following steps:
S1: the related four methods of diagnosis data information of four methods of diagnosis information collection module collection patient main suit；It include patient main suit in data information
One set.
S2: four methods of diagnosis info conversion module passes through StandardSyndrom component basis after receiving patient's four methods of diagnosis information
The comma or blank character of writing execute the movement of natural language word cutting, and the four methods of diagnosis information that doctor inputs is cut into descriptive disease
Adverbial modifier N；
S3: symptom language matching module utilizes neural network algorithm by four methods of diagnosis information modulus of conversion by StandardSyndrom component
The obtained symptom language N of block is matched with symptom repertorie, is converted to standardized sympotomatic set；.In traditional Chinese medicine, because of position
And property, it is all important index, therefore can all need to do similar Intelligent treatment.
S4: symptom weight identification module is received from StandardSyndrom in DiagnosisSystem component and is passed over
Standardized sympotomatic set after, the identification weighted value of various symptom is calculated according to magnanimity case data；Its object is to energy
Enough in several hundred kinds of possible disease types, most probable primary symptom (the first diagnosis of similar doctor trained in Western medicine) is found out
S5: Chinese medicine disease type that disease type matching module learn from magnanimity case data by DiagnosisSystem component and
The identification weighted value for the various symptom that the weighted value of classical symptom is obtained with symptom weight identification module carries out integration calculating, and presses
Ranking is carried out according to the weight summation of the symptom in each disease type, obtain main disease type and is calculated using characteristic of disease disease position neural network model
The characteristic of disease disease bit vector of the diagnosis out；DiagnosisSystem component can also diagnose other than card type, good using prior learning
Neural network model, go will likely characteristic of disease and sick position also determine and, the characteristic of disease disease bit vector referred to as diagnosed is done so as to subsequent
Precisely used when fine tuning.
S6: prescription recommending module by PrescriptionAnalysis Assembly calculation obtain the characteristic of disease disease position of each prescription to
Amount, then examined the characteristic of disease disease bit vector of each prescription with what disease type matching module obtained by PrescriptionDecision component
Disconnected characteristic of disease disease bit vector goes to carry out similarity mode, finally using the highest prescription of matching degree as recommendation.
PrescriptionAnalysis component is responsible for carrying out the quantitative analysis of all distinguished veteran doctors of TCM correlation prescriptions in advance, to prop up
PrescriptionDecision component is held at the side of choosing, there are enough quantitative values as foundation.
PrescriptionAnalysis component can try each first prescription characteristic of disease related to effect originally clinically and sick position is all first
Standardization is got off.PrescriptionAnalysis component uses the algorithm of neural network, goes to be fitted these standardized characteristic of diseases
After the completion of sick position study, this component be have go the drug composition within arbitrary prescription to analyze here side it is possible
Purpose effect.
Further, for another embodiment of method of the invention, according to magnanimity case data in the S4 step
The calculation for calculating the identification weighted value of various symptom is: by magnanimity case data symptom and disease type relationship according to
The image that symptom is characterized, disease type is label is stored, and the matrix of weighted value is then calculated according to TF-IDF algorithm.
Further, for another embodiment of method of the invention, according to TF-IDF algorithm meter in the S4 step
Calculate the calculation formula of the matrix of weighted value are as follows: TF-IDF (symptom, disease type)=(frequency that the symptom comes across the disease type/it should
Card type Symptomatic total frequency) * LOG (the disease type number/symptom of 1+ in total appears in how many disease types)；The starting point exists
Using the formula of TFIDF, inherently set out but because being calculated with this non-linear formula to simulate the so-called degree of discrimination
Come value, in range, have it is exponential, therefore can with calculate after TF-IDF size, make ranking results, and
And limitation range, as a kind of means for standardizing Normalization, can finally obtain this and identify weight between 0 to 1
The matrix of value.
Further, for another embodiment of method of the invention, the S5 step further includes disease type matching module
Go out the minor symptom disease type of primary symptom disease type appearance according to the identification weight analysis of primary symptom disease type matched in the module and various symptom.
When the clinical more complicated state of an illness, DiagnosisSyste component can then find the diagnosis of only primary symptom, and being can not be fine
Ground goes to explain the sympotomatic set that fully enters into, and at this moment component can try to find a minor symptom again that (the second of similar doctor trained in Western medicine is examined
It is disconnected), when to use primary symptom and minor symptom at the same time, it can preferably explain faced clinical state.
Primary symptom and minor symptom can cover one group of complete this symptom group of symptom group clinically jointly and can be updated to instructs in advance
The neural network for the symptom verification element perfected, can go and meet again to obtain characteristic of disease disease bit vector this vector
Each prescription vector in candidate prescription library within the scope of main symptom does the similarity mode based on COS value so as to can be in candidate
In prescription, after filtering out most suitable prescription.In fact be exactly two inner product values with dimensional vector, i.e., cos value be inner product divided by
Span from.
In use, be finely adjusted for the ease of doctor to prescription, system can be according to obtained recommendation prescription from database
Middle lookup obtains the characteristic of disease disease bit vector of the prescription, then by the characteristic of disease disease bit vector of the characteristic of disease disease bit vector of the prescription and diagnosis
It compares, and the characteristic of disease disease position to differ greatly between two kinds of characteristic of disease disease bit vectors is prompted (disparity range can be set, surpassed
Then indicate to differ greatly out), it is finely adjusted for diagnostician.It, can be according to the experience of doctor to place by way of this fine tuning
Side is finely adjusted, and can continue to optimize prescription during continuous fine tuning.
Application examples one: by one group of symptom, judge that chill violates the Syndrome in TCM of lung by intelligent auxiliary diagnosis system.
S1: user (doctor) issues one group of four methods of diagnosis information of starting: [cough, coughs up a small amount of dilute white phlegm, and asthma is micro- to have aversion to cold hair
Heat, nasal obstruction, watery nasal discharge, larynx is itched, or sees that bodily pain is lossless, thin white fur of tongue, floating and tense pulse].
S2:StandardSyndrom component after receiving patient's four methods of diagnosis information, by by step S1 Lai doctor input letter
Breath is cut into symptom language descriptive one by one [(cough) (coughing up a small amount of dilute white phlegm) (asthma) (micro- to have fever with aversion to cold) (nose
Plug) (watery nasal discharge) (larynx is itched) (or seeing that bodily pain is lossless) (thin white fur of tongue) (floating and tense pulse)].
Next S3:StandardSyndrom component can utilize neural network algorithm, go to calibrate most suitable matched standard
Symptom word: [(cough) (it is dilute that phlegm lacks matter) (phlegm color is white) (asthma) (aversion to cold) (fever) (nasal obstruction) (thin nasal discharge) (larynx is itched)
(bodily pain) (lossless) (tongue fur is thin) (whitish tongue) (floating pulse) (arteries and veins is tight)].
S4:DiagnosisSystem component receive the standardized sympotomatic set that is passed over from StandardSyndrom it
Afterwards, the identification weighted value of various symptom is calculated according to magnanimity case data.
S5:DiagnosisSystem component: 5.1 this component can be according in advance from the distinguished veteran doctors of TCM case data of magnanimity
In, learn the weighted value to various Syndrome in TCM and each classical symptom, goes to do integration with the identification weighted value of the various symptom of S4
Calculate at this moment, more than can identify in system nearly 400 cards can be gone to make the ranking on the basis of summation weight this
When, the ranking of the Top3 of system is respectively card of the 1. syndrome of wind-cold attacking lung .2. syndrome of superficies tightened by wind-cold .3. cold closed lungs card because of Chinese medicine
Type is also relevance from each other, therefore system finds out most probable three cards, and meaning clinically can be closer to, sometimes
It is the difference in degree, as is sometimes a shade of difference fallen ill on sick position, such as chill between syndrome of wind-cold attacking lung and cold closed lung card
Attack 5.2 this case between lung card and syndrome of superficies tightened by wind-cold because single main symptom can explain well it is nearly all clinically
The symptom seen, so to also stay at a main symptom i.e. enough for conclusion.5.3 this component utilize characteristic of disease disease position neural network model
The vector come is determined, can be (' cold ', 0.97), (' exogenous wind ', 0.92), (' table ', 0.86), (' lung ', 0.80),
(' close ', 0.75), (' phlegm ', 0.64), (' the circulation of vital energy in the wrong direction ', 0.49).
Master of the S6:PrescriptionDecision component on DiagnosisSystem component clearly good Chinese medicine
Card look for for after syndrome of wind-cold attacking lung, this component can be candidate in the available prescription under this Chinese medicine main symptom；Master is belonged to all
When card is the prescription library row search of syndrome of wind-cold attacking lung, PrescriptionDecision component can go to utilize this group of four methods of diagnosis information
The characteristic of disease disease bit vector of representative diagnosis is done with the characteristic of disease disease bit vector of prescription effect of all available prescription candidates and is matched
It calculates, finally by the highest prescription of matching degree；To this recommend prescription, Sanao tang plus-minus, vector calculate, being will be below
Drug forms (aster 9g, Fructus Aurantii 9g, dried orange peel 9g, almond 9g, Chinese ephedra 9g, tuber of pinellia 9g, campanulaceae 9g, Radix Glycyrrhizae 9g), is updated to and is based on
Composition and effect of tens of thousands of head prescriptions and the neural network model after having learnt, and obtain prescription effect characteristic of disease disease bit vector "
Exogenous wind ": 0.98, " closing ": 0.52, " stagnation of the circulation of vital energy ": 0.39, " lung ": 0.82, " phlegm ": 0.77, " cold ": 0.92 }.
Application examples two: by one group of symptom, the clinically complicated state of an illness is judged by intelligent auxiliary diagnosis system, needs main symptom
And after secondary card merges, the case of ability complete diagnosis.
S1: user (doctor) issues one group of four methods of diagnosis information of starting, and [glossolalia, dizzy, headache, tired, tongue is strong
Language is not smoothgoing, hemiplegia, nausea and vomiting, and feeling of fullness uncomfortable in chest, dry, bitter taste is indigestion and loss of appetite, dry and hard excrement, and right side upper and lower extremities muscular strength weakens, right
Side upper and lower extremities limitation of activity, the right skew of tongue body, red tongue body, tongue is yellow thick greasy, and body is fat]
S2:StandardSyndrom component after receiving patient's four methods of diagnosis information, by by step 1 doctor input information, cut
It is cut into symptom language descriptive one by one.
Next S 3:StandardSyndrom component can utilize neural network algorithm, remove to calibrate most suitable matched mark
Quasi- symptom word [(dislike by (glossolalia), (dizziness), (headache), (out of strength), (the strong language of tongue is stuttering), (hemiplegia)
The heart), (vomiting), (chest gastral cavity ruffian), (dry), (bitter taste), (loss of appetite), (dry and hard excrement), (tongue is askew), (tongue
It is red), (yellow tongue fur), (tongue fur is greasy), (tongue fur is thick), (body is fat)].
S4:DiagnosisSystem component receive the standardized sympotomatic set that is passed over from StandardSyndrom it
Afterwards, the identification weighted value of various symptom is calculated according to magnanimity case data.
S5:DiagnosisSystem component -5.1 at this moment, system Integrated Summary input come in classical symptom group and
Various card with after the weighted value of symptom, can find Syndrome Scale that single main symptom can be explained be inadequate at this moment, system
It can try to look for for each main symptom in the top possible card so as to the main symptom and secondary card that make collocation get up,
Can be high on comprehensive weight, and common construable Syndrome Scale be enough with this case study on implementation for, the row of Top3
Name is respectively 1. syndrome of wind-phlegm invading collaterals syndrome of accumulated dampness-toxicity, 2. syndrome of phlegm-heat attacking internally, 3. syndrome of wind-phlegm invading upward syndrome of hyperactivity of fire due to yin deficiency system
First recommendation and third are recommended, and are all on diagnosis, with main symptom and secondary card, this is because system is using a master
Card, when such as syndrome of wind-phlegm invading collaterals, can find some and wet or malicious relevant symptom group, can not effectively be explained, therefore can be again
It tries to find out syndrome of accumulated dampness-toxicity, this card is made collocation this component of 5.2 out and sentenced using characteristic of disease disease position neural network model
The vector fixed, can be [(' phlegm ', 0.71), (' heat ', 0.47), (' large intestine ', 0.36), (' liver ', 0.27), ('
Dynamic wind ', 0.24), (' head ', 0.14), (' blood stasis ', 0.09), (' wet ', 0.08), (' the circulation of vital energy in the wrong direction ', 0.08)].
Master of the S6:PrescriptionDecision component on DiagnosisSystem component clearly good Chinese medicine
Card look for for after syndrome of wind-phlegm invading collaterals, this component can be candidate in the available prescription under this Chinese medicine main symptom；In search,
PrescriptionDecision component can utilize same algorithm again, finally by the highest prescription of matching degree, to this recommendation
Prescription, Wendan Tang, from quasi- square source: the vector that Guo Zhen ball name cures case calculates, be by drug below composition (calamus 8g,
Rhizoma pinellinae praeparata 8g, Radix Curcumae 10g, Exocarpium Citri Rubrum 10g, Fructus Aurantii 10g, Poria cocos 12g, caulis bambusae in taenian 12g, bombyx batryticatus 12g, Caulis Spatholobi 12g, lycopodium calvatum 12g,
Radix Angelicae Sinensis 15g), it is updated to composition and effect based on tens of thousands of first prescriptions and the neural network model after having learnt, and obtain prescription
Characteristic of disease disease bit vector " dynamic wind ": 0.46, " closing ": 0.17, " stagnation of the circulation of vital energy ": 0.13, " heat ": 0.46, " phlegm ": 0.94, " wet ":
0.28, " blood stasis ": 0.22, " liver ": 0.12 }.
" one embodiment ", " another embodiment ", " embodiment ", " preferred implementation spoken of in the present specification
Example " etc., referring to combining specific features, structure or the feature of embodiment description includes describing extremely in the application generality
In few one embodiment.It is not centainly to refer to the same embodiment that statement of the same race, which occur, in multiple places in the description.Into one
For step, when describing a specific features, structure or feature in conjunction with any embodiment, what is advocated is to combine other implementations
Example realizes that this feature, structure or feature are also fallen within the scope of the present invention.
Although reference be made herein to invention has been described for multiple explanatory embodiments of the invention, however, it is to be understood that
Those skilled in the art can be designed that a lot of other modification and implementations, these modifications and implementations will fall in this Shen
It please be within disclosed scope and spirit.More specifically, disclose in the application, drawings and claims in the range of, can
With the building block and/or a variety of variations and modifications of layout progress to theme combination layout.In addition to building block and/or layout
Outside the modification and improvement of progress, to those skilled in the art, other purposes also be will be apparent.
Claims (7)
1. a kind of Chinese medicine intelligent assistance system based on machine learning and big data, it is characterised in that: the system comprises:
Four methods of diagnosis information collection module: for collecting the related four methods of diagnosis data information of patient main suit；
Four methods of diagnosis info conversion module: for passing through StandardSyndrom component after receiving patient's four methods of diagnosis information according to book
The comma or blank character write execute the movement of natural language word cutting, and the four methods of diagnosis information that doctor inputs is cut into descriptive symptom
Language N；
Symptom language matching module: for being converted four methods of diagnosis information using neural network algorithm by StandardSyndrom component
The obtained symptom language N of module is matched with symptom repertorie, is converted to standardized sympotomatic set；
Symptom weight identification module: for what is passed in the reception of DiagnosisSystem component from StandardSyndrom
After standardized sympotomatic set, the identification weighted value of various symptom is calculated according to magnanimity case data；
Disease type matching module: the Chinese medicine disease type for being learnt from magnanimity case data by DiagnosisSystem component
The identification weighted value progress integration calculating of the various symptom obtained with the weighted value of classical symptom with symptom weight identification module, and
Ranking is carried out according to the weight summation of the symptom in each disease type, main disease type is obtained and utilizes characteristic of disease disease position neural network model meter
Calculate the characteristic of disease disease bit vector of diagnosis；
Prescription recommending module: for obtained by PrescriptionAnalysis Assembly calculation the characteristic of disease disease position of each prescription to
Amount, then examined the characteristic of disease disease bit vector of each prescription with what disease type matching module obtained by PrescriptionDecision component
Disconnected characteristic of disease disease bit vector goes to carry out similarity mode, finally using the highest prescription of matching degree as recommendation.
2. the Chinese medicine intelligent assistance system according to claim 1 based on machine learning and big data, it is characterised in that: institute
The system of stating further includes data memory module, and for storing magnanimity case data, the magnanimity case data include disease type and symptom
Relation data, the candidate prescription under each disease type.
3. the Chinese medicine intelligent assistance system based on machine learning and big data according to weighing and require 1, it is characterised in that: described
Disease type matching module is also used to go out the primary symptom according to the identification weight analysis of primary symptom disease type matched in the module and various symptom
The minor symptom disease type that disease type occurs.
4. a kind of Chinese medicine intelligence householder method based on machine learning and big data, it is characterised in that: the following steps are included:
S1: the related four methods of diagnosis data information of four methods of diagnosis information collection module collection patient main suit；
S2: four methods of diagnosis info conversion module is by StandardSyndrom component according to writing after receiving patient's four methods of diagnosis information
Comma or blank character execute natural language word cutting movement, by doctor input four methods of diagnosis information be cut into descriptive symptom language
N；
S3: symptom language matching module utilizes neural network algorithm by four methods of diagnosis information modulus of conversion by StandardSyndrom component
The obtained symptom language N of block is matched with symptom repertorie, is converted to standardized sympotomatic set；
S4: symptom weight identification module receives the mark passed over from StandardSyndrom in DiagnosisSystem component
After the sympotomatic set of standardization, the identification weighted value of various symptom is calculated according to magnanimity case data；
S5: Chinese medicine disease type that disease type matching module learn from magnanimity case data by DiagnosisSystem component and
The identification weighted value for the various symptom that the weighted value of classical symptom is obtained with symptom weight identification module carries out integration calculating, and presses
Ranking is carried out according to the weight summation of the symptom in each disease type, obtain main disease type and is calculated using characteristic of disease disease position neural network model
The characteristic of disease disease bit vector of the diagnosis out；
S6: prescription recommending module obtains the characteristic of disease disease bit vector of each prescription by PrescriptionAnalysis Assembly calculation,
The diagnosis for again being obtained the characteristic of disease disease bit vector of each prescription and disease type matching module by PrescriptionDecision component
Characteristic of disease disease bit vector go carry out similarity mode, finally using the highest prescription of matching degree as recommendation.
5. the Chinese medicine intelligent assistance system according to claim 4 based on machine learning and big data, it is characterised in that: institute
State in S4 step is according to the calculation that magnanimity case data calculate the identification weighted value of various symptom: by magnanimity case number
The image that the relationship of symptom and disease type in is characterized according to symptom, disease type is label is stored, then according to TF-IDF
Algorithm calculates the matrix of weighted value.
6. the Chinese medicine intelligent assistance system according to claim 5 based on machine learning and big data, it is characterised in that: institute
State the calculation formula for calculating the matrix of weighted value in S4 step according to TF-IDF algorithm are as follows: TF-IDF (symptom, disease type)=
(symptom come across the frequency/card type of the disease type Symptomatic total frequency) * LOG (the disease type number/disease of 1+ in total
Shape appears in how many disease types).
7. the Chinese medicine intelligent assistance system according to claim 4 based on machine learning and big data, it is characterised in that: institute
Stating S5 step further includes identification weight analysis of the disease type matching module according to primary symptom disease type matched in the module and various symptom
The minor symptom disease type that the primary symptom disease type occurs out.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810805208.9A CN109102899A (en) | 2018-07-20 | 2018-07-20 | Chinese medicine intelligent assistance system and method based on machine learning and big data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810805208.9A CN109102899A (en) | 2018-07-20 | 2018-07-20 | Chinese medicine intelligent assistance system and method based on machine learning and big data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109102899A true CN109102899A (en) | 2018-12-28 |
Family
ID=64846994
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810805208.9A Pending CN109102899A (en) | 2018-07-20 | 2018-07-20 | Chinese medicine intelligent assistance system and method based on machine learning and big data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109102899A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109949940A (en) * | 2019-03-11 | 2019-06-28 | 金力 | A kind of big data digging system based on diabetes data library |
| CN110021427A (en) * | 2019-03-29 | 2019-07-16 | 北京医鸣技术有限公司 | A method of for carrying out automatic screening and diagnosis and treatment to patients with heart failure |
| CN110335684A (en) * | 2019-06-14 | 2019-10-15 | 电子科技大学 | The intelligent dialectical aid decision-making method of Chinese medicine based on topic model technology |
| CN111599486A (en) * | 2020-05-12 | 2020-08-28 | 成都睿明医疗信息技术有限公司 | Traditional Chinese medicine prescription recommendation sorting method based on data matching |
| CN111863239A (en) * | 2020-07-06 | 2020-10-30 | 南京大经中医药信息技术有限公司 | Recommendation method and terminal |
| CN112735583A (en) * | 2020-12-25 | 2021-04-30 | 山东众阳健康科技集团有限公司 | Traditional Chinese medicine health preserving robot and method |
| CN113140310A (en) * | 2021-05-06 | 2021-07-20 | 王耘 | Traditional Chinese medicine intelligent diagnosis and treatment system and creation method |
| CN113241173A (en) * | 2021-05-12 | 2021-08-10 | 华中科技大学 | Traditional Chinese medicine auxiliary diagnosis and treatment method and system for chronic obstructive pulmonary disease |
| CN115579132A (en) * | 2022-12-09 | 2023-01-06 | 智慧中医科技(广东)有限公司 | Verification method, device and system based on traditional Chinese medicine diagnosis and treatment experience file format |
| CN116246762A (en) * | 2023-05-11 | 2023-06-09 | 深圳汇医必达医疗科技有限公司 | Prescription dose optimization method, device, equipment and medium based on artificial intelligence |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105825064A (en) * | 2016-03-22 | 2016-08-03 | 施弘 | Daily traditional Chinese medicine therapy intelligent consultancy system |
| CN107578818A (en) * | 2017-08-25 | 2018-01-12 | 广州宝荣科技应用有限公司 | A kind of auxiliary evolution method and device based on deep learning |
-
2018
- 2018-07-20 CN CN201810805208.9A patent/CN109102899A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105825064A (en) * | 2016-03-22 | 2016-08-03 | 施弘 | Daily traditional Chinese medicine therapy intelligent consultancy system |
| CN107578818A (en) * | 2017-08-25 | 2018-01-12 | 广州宝荣科技应用有限公司 | A kind of auxiliary evolution method and device based on deep learning |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109949940A (en) * | 2019-03-11 | 2019-06-28 | 金力 | A kind of big data digging system based on diabetes data library |
| CN110021427A (en) * | 2019-03-29 | 2019-07-16 | 北京医鸣技术有限公司 | A method of for carrying out automatic screening and diagnosis and treatment to patients with heart failure |
| CN110335684A (en) * | 2019-06-14 | 2019-10-15 | 电子科技大学 | The intelligent dialectical aid decision-making method of Chinese medicine based on topic model technology |
| CN111599486A (en) * | 2020-05-12 | 2020-08-28 | 成都睿明医疗信息技术有限公司 | Traditional Chinese medicine prescription recommendation sorting method based on data matching |
| CN111863239A (en) * | 2020-07-06 | 2020-10-30 | 南京大经中医药信息技术有限公司 | Recommendation method and terminal |
| CN112735583A (en) * | 2020-12-25 | 2021-04-30 | 山东众阳健康科技集团有限公司 | Traditional Chinese medicine health preserving robot and method |
| CN113140310A (en) * | 2021-05-06 | 2021-07-20 | 王耘 | Traditional Chinese medicine intelligent diagnosis and treatment system and creation method |
| CN113140310B (en) * | 2021-05-06 | 2024-02-27 | 王耘 | Intelligent diagnosis and treatment system for traditional Chinese medicine and creation method |
| CN113241173A (en) * | 2021-05-12 | 2021-08-10 | 华中科技大学 | Traditional Chinese medicine auxiliary diagnosis and treatment method and system for chronic obstructive pulmonary disease |
| CN115579132A (en) * | 2022-12-09 | 2023-01-06 | 智慧中医科技(广东)有限公司 | Verification method, device and system based on traditional Chinese medicine diagnosis and treatment experience file format |
| CN115579132B (en) * | 2022-12-09 | 2023-08-22 | 智慧中医科技(广东)有限公司 | Verification method, device and system based on traditional Chinese medicine diagnosis and treatment experience file format |
| CN116246762A (en) * | 2023-05-11 | 2023-06-09 | 深圳汇医必达医疗科技有限公司 | Prescription dose optimization method, device, equipment and medium based on artificial intelligence |
| CN116246762B (en) * | 2023-05-11 | 2023-07-21 | 深圳汇医必达医疗科技有限公司 | Prescription dose optimization method, device, equipment and medium based on artificial intelligence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109102899A (en) | Chinese medicine intelligent assistance system and method based on machine learning and big data | |
| CN109166619A (en) | Chinese medicine intelligent diagnostics auxiliary system and method based on neural network algorithm | |
| CN109669994B (en) | Construction method and system of health knowledge map | |
| CN107016438B (en) | System based on traditional Chinese medicine syndrome differentiation artificial neural network algorithm model | |
| CN103678281B (en) | The method and apparatus that text is carried out automatic marking | |
| CN104915561A (en) | Intelligent disease attribute matching method | |
| CN110838368A (en) | Robot active inquiry method based on traditional Chinese medicine clinical knowledge graph | |
| CN109670179A (en) | Case history text based on iteration expansion convolutional neural networks names entity recognition method | |
| CN106776711A (en) | A kind of Chinese medical knowledge mapping construction method based on deep learning | |
| CN108986912A (en) | Chinese medicine stomach trouble tongue based on deep learning is as information intelligent processing method | |
| CN110335665A (en) | It is a kind of applied to medical image auxiliary diagnosis analysis to scheme to search drawing method and system | |
| Huan et al. | Deep convolutional neural networks for classifying body constitution based on face image | |
| CN109378066A (en) | A kind of control method and control device for realizing disease forecasting based on feature vector | |
| CN110459321A (en) | A kind of aided diagnosis of traditional Chinese medicine system based on card element | |
| CN109947901B (en) | Prescription efficacy prediction method based on multilayer perceptron and natural language processing technology | |
| CN108647203B (en) | Method for calculating text similarity of traditional Chinese medicine disease conditions | |
| CN111430025B (en) | Disease diagnosis model training method based on medical image data augmentation | |
| CN108509419A (en) | Ancient TCM books document participle and part of speech indexing method and system | |
| CN110083708A (en) | A kind of medical bodies association analysis method of knowledge based map | |
| CN107705853A (en) | Clinical nutrition risk screening method and system | |
| CN111651991A (en) | Medical named entity identification method utilizing multi-model fusion strategy | |
| Huan et al. | Transfer learning with deep convolutional neural network for constitution classification with face image | |
| CN114255851A (en) | Chinese herbal medicine prescription recommendation method based on hypergraph neural network | |
| Huan et al. | Multilevel and multiscale feature aggregation in deep networks for facial constitution classification | |
| Yuan et al. | Design and implementation of the traditional Chinese medicine constitution system based on the diagnosis of tongue and consultation |
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 |
Application publication date: 20181228 |
|
| RJ01 | Rejection of invention patent application after publication |