CN101408912A - Method for automatically extracting characteristic function of traditional Chinese medicine pulse manifestation - Google Patents
Method for automatically extracting characteristic function of traditional Chinese medicine pulse manifestation Download PDFInfo
- Publication number
- CN101408912A CN101408912A CNA2008101531750A CN200810153175A CN101408912A CN 101408912 A CN101408912 A CN 101408912A CN A2008101531750 A CNA2008101531750 A CN A2008101531750A CN 200810153175 A CN200810153175 A CN 200810153175A CN 101408912 A CN101408912 A CN 101408912A
- Authority
- CN
- China
- Prior art keywords
- pulse
- profile data
- pulse profile
- data
- chinese medicine
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses a method for automatically extracting a traditional Chinese medicine pulse characteristic function. The method comprises the following steps: obtaining periodic data related to time series from source beyond a computer; selecting data from the periodic data, normalizing and automatically building a mathematical model for the selected data; extracting a characteristic function which meets requirements and characterizes the data characteristic. The method is different from the common methods such as multiple regression method, wavelet analysis method or neural network method and the like. The characteristic function extracted by the method has an unlimited concrete form, an characteristic function extraction program is based on an evolutionary computation method, and the evolutionary computation is a self-organizing and self-adapting artificial intelligence technology which simulates the biological evolution process and a problem-solving mechanism, and is a searching algorithm having a 'generation-detection' iterative process, and the extracted characteristic function can assist doctors of traditional Chinese medicine in objectivizing and quantizing the diagnosis work.
Description
Technical field
The present invention is with to handle exterior technology data computing machine program relevant, relates to the traditional Chinese medical science quantification diagnostic field that objectifies, and is a kind of method of utilizing computer technology to realize extraction automatically and quantizing the Chinese medicine pulse feature.
Background technology
Diagnosis by feeling the pulse is one of diagnostic method of tool tcm characteristic, is to distinguish syndrome, the evaluation state of an illness, and prediction disease progression trend is judged the important evidence of result of treatment.The doctor is depended on subjective sensation under referring to the judgement of a certain pulse condition when clinical feeling the pulse.Because the difference of everyone physiology physique, cause the pulse condition that shows under the biological property of its basic pulse condition and the pathologic condition also different, study its pulse condition Changing Pattern based on personalized speciality, it is significant to find out the specificity quantizating index that can be used as the clinical syndrome differentiation foundation.
Arteries and veins figure is meant one group of oscillogram that the pulse condition of feeling the pulse gained is traced out through pulse instrument, can reflect some feature of pulse condition to a certain extent.The arteries and veins figure that obtains has sphygmogram, differential map etc. multiple, and is most widely used with sphygmogram.The appearance of arteries and veins figure makes in the Chinese medicine pulse research and has introduced objective factor, helps the analysis of pulse condition and syndrome.The expression of pulse condition feature and extraction problem are that the traditional Chinese medical science objectifies and quantizes the basic problem of diagnosis among the arteries and veins figure.This paper mainly pays close attention to the data of the sign pulse condition information that obtains from pulse instrument or this kind equipment of electropulsograph, therefore mainly in this article pays close attention to relevant data, and what difference whether not strict differentiation pulse instrument or electropulsograph have, explanation hereby.
At present relevant with pulse instrument or electropulsograph device patent and achievement are more, but pulse condition or sphygmogram characteristic extract automatically and the achievement of quantitative evaluation is less, wherein disclosed on June 28th, 2006, China publication number CN 1792319A, application number 200510061394.2, Chinese medicine pulse characteristic parameter automatic testing method, " probabilistic reasoning model based on Bayesian network is the basis; utilization modern signal processing technology; image recognition technology is extracted tongue picture and pulse condition feature ", " it is characterized in that: the pulse condition collecting device of being made up of pulse diagnosis sensor and pulse-tracing collection circuit obtains pulse signal; and transfer to computing machine, finishes the input pulse signal removed in computing machine and makes an uproar; flex point detects and the detection of arteries and veins graph parameter.”。Wherein utilized technology such as multiple small echo that arteries and veins figure is analyzed.
Document: " Zhong Qi; Xiao Suigui. based on the Chinese medicine pulse Study on Classification and Recognition [J] of evolutionary computation. computer utility and software; 2007; 24 (5): 158-159; 174 " in from three kinds of spectrogram (power spectrum of pulse signal, cepstrum and transport function) analyze in 15 characteristic parameters that obtained, choose the characteristic component of 7 characteristic parameters as pattern classification, respectively to 6 kinds of pulse condition (normal pulses, taut pulse, smooth pulse, thready pulse, slow arteries and veins, rapid pulse) data use document " Ferreira; C..Gene Expression Programming:A New Adaptive Algorithm for Solving Problems[J] .Complex Systems; 2001; 13 (2): 87-129 " in disclosed GEP basic skills excavate the classification functional expression of various pulse conditions, the purpose of excavation functional expression concentrates on classifies to six kinds of pulse conditions.The dissimilar pulse condition of how many kinds of is arranged, and its method is just excavated the how many kinds of function, and 6 kinds of pulse conditions need be excavated 6 functional expressions, and for certain 1 arteries and veins figure to be measured, it is non-judgement that wherein per 1 functional expression can be done corresponding to certain a kind of pulse condition.The function of excavating in the document only is used to define the classification work of quantity, it defines the classification number, and only be used to predict big classification under the new data, but might not represent the feature of feature someone, some time, certain class pulse condition or certain concrete arteries and veins figure well.In fact because there are various individual differences in everyone, and for same class pulse condition, the details of different people pulse condition feature also will inevitably be variant, the document and the not mentioned disposal route that how characterizes the pulse condition minutia with functional expression.In addition, how method is not mentioned in the document can be according to the form of real needs qualified function formula.
The method that other current research mostly is Functional Analysis such as using neural network, wavelet analysis or approximation of function quantizes the pulse condition feature, and these methods generally all define the concrete form of pulse condition character representation.
Summary of the invention
The objective of the invention is expression and extraction problem, overcome the deficiency that above-mentioned prior art exists, propose a kind of automatic extraction and quantize the method for certain traditional Chinese medical science sphygmogram characteristic, realize effectively that the traditional Chinese medical science objectifies and quantize diagnosis with auxiliary at the pulse condition feature.
Provide a kind of Chinese medicine pulse fundamental function extraction method in order to reach purpose of the present invention, comprise that the pulse condition collecting device that the method for obtaining pulse profile data is made up of pulse diagnosis sensor and pulse-tracing collection circuit obtains pulse signal, transfer to computing machine, perhaps employing itself can directly be stored as pulse profile data the electropulsograph of the discernible data of computing machine, can directly pass through USB interface, com interface, or network interface reads in the pulse profile data storehouse with pulse profile data, essence of the present invention is, after obtaining the periodic pulse profile data relevant from the computer system outside with time series, utilize these data of computer programs process, and obtain obtaining to meet the technical data treatment effect of the natural law.Key step comprises, from the pulse profile data that the outside obtains, chooses data, standardization, and it is set up mathematical model automatically, what extract satisfactory this data characteristics of sign does not limit the fundamental function that embodies the formula form, the output and the auxiliary traditional Chinese medical science quantification diagnostic work that objectifies.
Chinese medicine pulse fundamental function extraction method is to realize by following steps:
1, obtain pulse profile data from electropulsograph:
The pulse signal that is obtained by electropulsograph can be an also simulating signal of digital signal.No matter which kind of electropulsograph device, the pulse signal of its collection can be divided into 2 kinds of digital signal and simulating signals.Can be directly it be quantized for digital signal, thereby collect the manageable pulse profile data of computing machine according to the intensity level of signal; Similar with the digitized principle of audio or video, always can adopt pulse code modulation (PCM) or alternate manner that its digitizing is converted to the manageable pulse profile data of computing machine for simulating signal.These pulse profile data of gathering from electropulsograph are the object of postorder work disposal just.
Pulse signal can be the real-time pulse signal of electropulsograph device, also can be the digital signal of having changed of gathering in advance well and leaving in the computing machine.
Need to prove, be without loss of generality, the electropulsograph that we require to obtain pulse profile data has passed through strict adjustment, the pulse profile data of obtaining or almost do not have noise, noise all produces by intrinsic rule when each the measurement, be that ground unrest is stable, in view of aspect ratio to the time what pay close attention to is the difference of feature, stable noise can the effect characteristics function the expression ability, therefore the present invention has foundation focus is concentrated on how to extract feature from pulse profile data, and needn't consider the noise processed problem.
2, determine the pulse profile data starting point:
The waveform details is had nothing in common with each other on cycle although each is beated, and the heartbeat of living person or living animal or pulsation are periodic on the whole, thereby the wave function that forms corresponding to pulse condition also must have the some cycles feature.
For the pulse condition fundamental function with periodicity characteristics, the phase place of the function of the waveform of the pulse condition feature of different pulse profile data starting point correspondences is with different, and its concrete functional expression also can be different.Simpler and clearer more consistent for the final pulse condition fundamental function that extracts, be convenient to aspect ratio to and practical application, technical solution of the present invention is wanted the specified data starting point when handling pulse profile data.This step is finished in computing machine, the pulse profile data that adopts computer programs process to obtain from the outside.
3, determine the pulse profile data terminal point:
According to mathematical knowledge, the phase-independent of data endpoint and periodic function, if but from starting point to the terminal point less than function cycle, then the fundamental characteristics of this periodic function is incomplete in data, on the other hand, too much data can influence the efficient of extracting the pulse condition fundamental function, and therefore for can quick and precisely extract the pulse condition fundamental function, technical solution of the present invention proposes also to want the specified data terminal point when handling pulse profile data.This step is finished in computing machine, the pulse profile data that adopts computer programs process to obtain from the outside.
4, standardization pulse profile data:
According to mathematical knowledge, the amplitude difference of oscillogram, the mathematic(al) representation of wave function also can be different, and the mathematic(al) representation of pulse condition fundamental function is similar with it, and the data that may cause when reducing different electropulsographs and digitizing are inconsistent, the consistance of the final pulse condition fundamental function of standard, improve the widely-used performance of fundamental function, technical solution of the present invention proposes to want the normalized number certificate, and the numerical value standard of the quantification of all pulse profile data is arrived [1,1] or do next step processing between [0,1] again.Normalized method can be a bi-directional scaling etc.This step is finished in computing machine, the pulse profile data that adopts computer programs process to obtain from the outside.
5, heredity inspires evolution algorithm to extract the pulse condition fundamental function from pulse profile data based on biological immune:
The present invention's software engineering that uses a computer is handled the pulse profile data of handling through preceding step, therefrom extracts the pulse condition fundamental function.The fundamental function extraction procedure of the embodiment of the invention is based on a kind of evolutionary computation method.Evolutionary computation is self-organization, the self-adaptation artificial intelligence technology that an analoglike biological evolution process and mechanism are found the solution problem, is a kind of searching algorithm with the iterative process of " generation-detection ".
Fundamental function extraction procedure of the present invention is subjected to the biological immune genetic mechanism to inspire proposition.Immune System can be crossed by immunocyte continuous clonal vaviation in the immune genetic evolutionary process, and immunocyte can be found the material that external foreign matter, bacterium or virus etc. may be harmful to body, to guarantee body health.Immune System has the advantage of all many-sides such as diversity, tolerance, immunological memory, distributed parallel processing, self-organization, self study, self-adaptation and robustness.At present the simulation biology immunity principle to find the solution in the computer realm Artificial Immune Algorithm of problem many, because separately based on different immunity principles, various specific implementation methods differ greatly, simultaneously, for solving different particular problems, structure individual in the algorithm also has than big-difference.
Fundamental function extraction procedure of the present invention aims at based on the biological immune genetics principle and extracts the design of pulse condition fundamental function, be more suitable for extracting and have periodic characteristic, require the functional expression of output versatile and flexible as far as possible, the while is the minutia of all kinds of Wave datas of quantificational description more accurately.In the present invention, the vector of the character string vector that the individuality of participate in evolving is made up of by certain rule functor, operational symbol, variable symbol, constant symbol, one or more mathematical function formulas can be changed and represent to each individuality by the mode of decoding; The degree of closeness fitness numeric representation of individual and the best target signature function of being asked gets by fitness function calculating; Individuality by the initialization some constitutes colony, constantly to use evolutionary operator to make colony constantly evolve based on the mode of simulation biological immune evolution strategy, the individuality that characterizes the pulse condition feature request in colony, occurs meeting, will just can extract the pulse condition fundamental function that characterizes the pulse condition feature after this individuality decoding.For not leading to misunderstanding, explanation here, alleged functional expression of this paper or mathematical function formula are meant the mathematic(al) representation that can describe funtcional relationship between a plurality of variablees.This step is finished in computing machine, the pulse profile data that adopts computer programs process to obtain from the outside, the auxiliary traditional Chinese medical science of fundamental function that obtains characterizing the pulse condition feature quantification diagnostic work that objectifies.
The inventive method can be extracted the pulse condition feature among the arteries and veins figure automatically, the pulse condition feature is represented in the fundamental function mode, be different from common methods such as multiple regression, wavelet analysis or the neural net method form of qualified function mathematic(al) representation in advance, this method can be extracted and automatically concisely not limit the mathematical function formula of concrete form to represent dissimilar pulse condition features, realize accurate, objective, the quantization means of Chinese medicine pulse feature and extract automatically, extract the pulse condition fundamental function subprogram evolution algorithm that heredity inspires based on biological immune.This method aims to be extracted the pulse condition fundamental function and proposes, and is suitable for the extracting cycle function, and functional form is flexible, can be flexibly the minutia of all kinds of Wave datas of quantificational description accurately, also can be used for extracting the fundamental function of other periodic time series data.
Description of drawings
Fig. 1 is the process flow diagram of the Chinese medicine pulse fundamental function extraction method of the embodiment of the invention.
Fig. 2 is that the Chinese medicine pulse data are obtained synoptic diagram.
Fig. 3 is the signal of pulse condition oscillogram.
Fig. 4 is the process flow diagram of definite pulse profile data starting point of the embodiment of the invention.
Fig. 5 is the process flow diagram of definite pulse profile data terminal point of the embodiment of the invention.
Fig. 6 is the process flow diagram of the standardization pulse profile data of the embodiment of the invention.
Fig. 7 is the synoptic diagram of individual configurations in the fundamental function extraction procedure.
Fig. 8 is that the word string vector decoding is the synoptic diagram of mathematic(al) representation in the individual configurations.
The synoptic diagram that changes of word string vector wherein when Fig. 9 is the variation of individual single-point.
The synoptic diagram that changes of word string vector wherein when Figure 10 is individual cross and variation.
Embodiment
Relate generally to the solution that the computing machine external object is controlled or handled, describe thought of the present invention in detail below in conjunction with accompanying drawing.
Fig. 1 is the process flow diagram of the Chinese medicine pulse fundamental function extraction method of the embodiment of the invention.Fig. 1 program 100 has provided the basic step of automatic extraction Chinese medicine pulse fundamental function:
Wherein 102 for to obtain pulse profile data from electropulsograph, and this step relates to from the computer system outside obtains data, and disposal route is seen among Fig. 3 synoptic diagram and the explanation to step 300 in the specific embodiment;
104 for determining the pulse profile data starting point, and concrete disposal route is seen the process flow diagram of program 400 among Fig. 4;
106 for determining the pulse profile data terminal point, and concrete disposal route is seen the process flow diagram of program 500 among Fig. 5;
108 are the standardization pulse profile data, and concrete disposal route is seen the process flow diagram of program 600 among Fig. 6;
110 for calling pulse condition fundamental function extraction procedure step, and the specific procedure of calling is seen the remainder of program 100 among Fig. 2.
Step 112-130 is the process flow diagram of the fundamental function extraction procedure part of the embodiment of the invention, the process of its simulation biological immune genetic evolution.
112 steps are for generating colony randomly, the G of colony is the set of x individuality (the individual detailed description in detail sees also Fig. 7 explanation), individual by n=2 (quantity can be set according to practical application request among other embodiment) word string vector formation, connector is "+", word string vector dimension is m=20;
The rule that generates the word string vector and document " Ferreira; C..Gene ExpressionProgramming:A New Adaptive Algorithm for Solving Problems[J] .Complex Systems; 2001,13 (2): 87-129 " in the method essence of generation K expression formula identical;
If character set F={ ' is sin ', ' tan ', ' sinh ', ' tanh ', ' arcsin ', ' ln ' }, O={ '+', ' ', ' * ', '/', ' ^ ' }, T={ ' t ' }, C={ ' 1 ', ' 3 ', ' 9 ', ' 27 ', ' 81 ', ' 243 ' }, please note: can adjust character set F, O, T according to system requirements in other embodiments, the particular content of C, identical with requiring in the GEP algorithm, as long as the word string among the F is the discernible function string of system, among the O is the operational character string, among the T is the terminal symbol string, getting final product for the constant string among the C; 112 process embodiment are from F, O, T chooses character randomly among the C, insert the preceding 1-9 dimension position of word string vector, from T, choose character among the C randomly, insert word string vector 10-20 dimension position, generate n word string vector, constitute 1 individuality, so operation generates x individuality, constitutes 1 colony.
If immunocyte is to having than high affinity from body in the living organism in the Immune System, then this cell may " manslaughtering " autogenous cell, and produce the immune response of a series of harm self, normal immune system all needs behind the immunocyte that clonal vaviation makes new advances through the self-tolerance process, and having only has the immunocyte of tolerance can be ripe and participate in follow-up immunity circulation to autogenous cell.Immunocyte self-tolerance process in the 114 steps simulation Immune System is by the type of tolerance step control individual in population.For previous step rapid 112 or the 130 new individualities that generate, immunocyte is corresponding to the individuality in this program, if should the decoded functional expression of individuality with 116 in the body storehouse functional expression after certain individual decoding (decoding sees also Fig. 8 explanation) identical, then carry out 118 steps, wither away and promptly delete this individuality, otherwise change 120 steps over to.
The 116th, from the body storehouse, be a kind of data storage, what wherein deposit is the group of individuals that should not appear in the net result; Certainly, since final output is decoded functional expression, in the body storehouse, can directly deposit functional expression; These functional expressions are not meet the functional expression that system needs; Program of the present invention can be controlled every individuality that is decoded as the functional expression in the body storehouse and need not occur in again at the round-robin further part, like this can be so that program can be controlled the particular content and the form of the final output of pulse condition fundamental function, also can reduce unnecessary fitness evaluation work, improve executing efficiency; Need indicate, can also dynamically be regulated and control when 124 steps according to the program implementation situation from the body storehouse; Here with traditional GEP method obvious difference is arranged.
The 120th, ripe colony's step is promptly the ripe G ' of group of individuals that joins through the successful individuality of individual tolerance after 114 steps.
The 122nd, check the population size decision operation whether compliance with system requires, k is one might as well make k=20 here greater than zero integer, if individual number (being population size) reaches requirement among the G ', G ' .size () 〉=k then jumps to 124, otherwise jumps to 130.Can dynamically adjust the size threshold k of mature groups G ' among other embodiment according to system requirements, control the scale that individual number that each stage need test and regulation and control are evolved.
124 steps each individuality in the colony of maturation is decoded respectively (decoding see also Fig. 8 explanation) be functional expression, might as well establish the decoded functional expression of a certain individuality is f (x), and each functional expression is calculated corresponding fitness on 126 pulse profile data storehouses; Fitness is that certain individuality approaches a kind of of target signature function and estimates, and makes that T is the pulse profile data set, comprises the pulse profile data in m the moment, y
jRepresent the j actual measurement numerical value of a pulse condition measurement index constantly,
Be the value of calculating gained, make M=100, then fitness
Note, in other embodiments can certainly be for the pulse condition fundamental function extracts task custom-made design fitness computing method automatically, as long as can measure the close degree of this functional expression and pulse profile data;
Simultaneously, individual decoded functional expression with fitness minimum in the epicycle circulation in 124 steps will be placed among the body storehouse, like this, when treating the next round evolutionary circulation, the individuality that is decoded as this functional expression will tolerate failure in 114 step self-tolerance steps, and will be withered away in the individual extinction step of 118 steps and no longer participate in follow-up operation; Owing to known that before this close degree of this individuality decoding gained functional expression and pulse profile data is not high, this individuality can not be final fundamental function certainly, so just wither away it when early stage, can reduce the waste of subsequent operation, improve system performance, from this point, the characteristics that the memory historical information is arranged from the body storehouse among the present invention.
The 126th, the pulse profile data storehouse is a kind of data storage, and what wherein deposit is through 102,104, the pulse profile data that 105,106 step process are crossed.
The 128th, the Rule of judgment whether program stops, if the fitness of the optimized individual in the mature groups (individuality that promptly can accurately represent the pulse condition feature) reaches the requirement of program setting, make s ∈ [0,100], if fitness 〉=s would carry out function that 132 steps export this individuality decoding back gained as the pulse condition fundamental function that characterizes the pulse condition feature to computer screen and terminal device, and terminator, otherwise carry out 130 steps; Note, because fitness≤100 make Automatic Program stop so can look for a s ∈ [0,100] to satisfy fitness 〉=s, rather than infinite loop are gone down.
The 130th, clonal vaviation step (variation sees also Fig. 9, Figure 10 explanation), title has been used for reference the term in the biological immune science of heredity, the clone is that a kind of replicate run can enlarge population size, and variation can make the individuality in original colony obtain upgrading, to search for new functional expression.When population size can't satisfy system requirements, if difference is v, v copy of the individual clone of picked at random adds among the colony.If the variation probability is p ∈ [0,1], the calling system random function gets a random value p ', if p ' 〉=p then, carry out the single-point variation, and carry out the cross and variation operation at random with the same manner, notice that these 2 kinds of mutation operations can not enlarge or dwindle population size; In other is implemented, can also use document " Ferreira; C..Gene Expression Programming:A New Adaptive Algorithm for SolvingProblems[J] .Complex Systems; 2001; 13 (2): 87-129 " etc. in mutation operator, can certainly design corresponding operator for the pulse condition fundamental function extracts the task custom-made automatically.
Fig. 2 is that the Chinese medicine pulse data are obtained synoptic diagram.The present invention carries out pulse-tracing collection by pulse diagnosis sensor and pulse-tracing collection circuit, by USB interface pulse condition is input in the computer processor, gather different pressures pulse signal down by programmed control, the arteries and veins diagram data of sign pulse condition is deposited among the pulse profile data storehouse in the computer system.
Fig. 3 is the signal of pulse condition oscillogram.As shown in Figure 3, the pulse condition information of feeling the pulse gained is through the electropulsograph device processes, can and constantly note the desired value of measuring, and one group of oscillogram that characterizes pulse condition tracing out, be arteries and veins figure, it can reflect some feature of pulse condition to a certain extent, can analyze different pulse condition features in objective quantification ground by analyzing different waveforms.The embodiment of the invention can extract the fundamental function that characterizes the pulse condition feature automatically from this type of shape information, quantize traditional Chinese medical science sphygmogram characteristic, and objectifying with the auxiliary effectively realization traditional Chinese medical science quantizes diagnosis.
Fig. 4 is the process flow diagram of definite pulse profile data starting point of the embodiment of the invention.Program 400 is to determine the process flow diagram of pending pulse profile data starting point among the figure.402 read pending pulse profile data to calculator memory from the pulse profile data storehouse; 404 pairs of data of reading begin to find in the cycle (heart beat cycle) that a pulsation is beated that moment point of pulse condition information numerical value maximum (or minimum) to be designated as t from original starting point; 406 steps have obtained demarcating the pulse profile data of ground zero to this segment data deletion from original starting point to the t moment point, and the starting point of pulse profile data will become t constantly like this.
Fig. 5 is the process flow diagram of definite pulse profile data terminal point of the embodiment of the invention.Program 500 is to determine the process flow diagram of pending pulse profile data terminal point among the figure.502 steps read pending pulse profile data to calculator memory from the pulse profile data storehouse; 504 from starting point, behind the n of a bypass system appointment pulsation period (heart beat cycle), last is designated as t ' constantly; 506 steps extract starting point to this segment data of t ' moment point as new pulse profile data, handle for next step.
Fig. 6 is the process flow diagram of the standardization pulse profile data of the embodiment of the invention.Program 600 is process flow diagrams of standardization pulse profile data among Fig. 6.
602 steps read pending pulse profile data to calculator memory from the pulse profile data storehouse.
604 find the maximal value in the pulse profile data to be designated as a (but requiring this step to be selection operation according to subsequent step in other embodiments) in pulse profile data.
606 find the minimum value in the pulse profile data to be designated as b (but requiring this step to be selection operation according to subsequent step in other embodiments) in pulse profile data;
608 steps begin circulation, and the value of each pulse profile data is designated as x.
610 steps are [1 with the x standard, 1] value between (being not limited in value standard with pulse profile data as the case may be and turning to value between [1,1], can also be other value in interval), and update data set, normalization method can be the linear scaling pantography: use formula
Standardize; In other is implemented, can certainly be the non-linear space transform method as: (1) makes that the numerical value after x=arcsin (x) standardization is interval to be [pi/2, pi/2], (2) make the numerical value interval after x=arcsin (x)/(pi/2) standardizes be [1,1], (3) make the numerical value interval after x=sin (x) standardizes be [1,1], can certainly be in other is implemented with the x standard to other numerical value interval.
612 steps, jump to 608 and continue circulation if data still have been untreated for stopping decision operation, otherwise this section EOP (end of program).
Fig. 7 is the synoptic diagram of individual configurations in the fundamental function extraction procedure.Structure 700 is structural representations individual in the example of the present invention among Fig. 7, and individuality is made of n word string vector in this synoptic diagram.N word string vector decodable code is a n functional expression, and this n functional expression can be according to operational symbol "+" couple together big functional expression of structure; Certainly also can use in other embodiments other functor exclusive disjunction operational character (as: POW ,+,-, * ,/etc.) couple together, big functional expression of structure is decoded as a plurality of functional expressions by rule even.
The value of concrete in other embodiments n, the length of word string vector in the individuality, functor (as: sin, cos, ln, exp, sqrt etc.), arithmetic operation symbol (as:+,-, * ,/etc.), the choosing and determine and can set flexibly of constant, variable terminal symbol (be variable name, abbreviate terminal symbol as), connector etc. according to the concrete needs of program.
Fig. 8 is that the word string vector decoding is the synoptic diagram of mathematic(al) representation in the individual configurations.Among the figure example 800 be in the example of the present invention in the individuality word string vector decoding be mathematic(al) representation tree, the mathematic(al) representation tree-encoding is the example of word string vector.
With "? " be digital placeholder, then wherein " F? " representative function accords with exclusive disjunction operational character (abbreviation operational character), " C? " represent the constant symbol, " T? " represent terminal symbol.Functor or operational character have " F1 ", " F2 ", " F3 " in the example, wherein need 2 parameter arguments during " F1 " computing, need 1 parameter argument during " F2 " computing, need 3 parameter arguments during " F3 " computing.The constant symbol has " C1 ", and terminal symbol has " T1 ".
Present embodiment use document " Ferreira; C..Gene Expression Programming:ANew Adaptive Algorithm for Solving Problems[J] .Complex Systems; 2001; 13 (2): 87-129 " in the program of K expression formula decoding, is the mathematic(al) representation tree on the right in the mode of example 800 with the word string vector decoding on the left side, is the word string vector on the left side with the mathematic(al) representation tree-encoding on the right.
Each word string vector in the individuality can be decoded as 1 mathematic(al) representation, a plurality of mathematic(al) representation connectors "+" can constitute 1 mathematic(al) representation after connecting.
The synoptic diagram that changes of word string vector wherein when Fig. 9 is the variation of individual single-point.Example 900 has been demonstrated the effect of individual word string vector single-point variation among the figure, and the variation of 1 individual word string vector is 1 new individual word string vector, and the single-point mutation operator in it and the traditional genetic algorithm is similar.
The synoptic diagram that changes of word string vector wherein when Figure 10 is individual cross and variation.Example 1000 has been demonstrated the effect of 2 individual word string vector cross and variation among the figure, and the variation of 2 individual word string vectors is 2 new individual word string vectors, and the cross and variation operator class in it and the traditional genetic algorithm seemingly.
It may be noted that, the operator of variation is not limited to the situation of Fig. 9, example shown in Figure 10, can set flexibly according to the specific requirement of program during realization, the method in other genetic algorithm or the immune algorithm can be used or improve to the method for operating of corresponding operator, can certainly design new operator and improve program effect, as long as can make original colony change new colony in a certain way into according to system requirements.
Claims (6)
1, a kind of Chinese medicine pulse fundamental function extraction method, comprise that the pulse condition collecting device that the method for obtaining pulse profile data is made up of pulse diagnosis sensor and pulse-tracing collection circuit obtains pulse signal, transfer to computing machine, perhaps employing itself can directly be stored as pulse profile data the electropulsograph of the discernible data of computing machine, can directly pass through USB interface, com interface, or network interface reads in the pulse profile data storehouse with pulse profile data, it is characterized in that realizing by following steps:
Obtaining the pulse profile data step comprises: gather pulse profile data from electropulsograph; Deposit in the computer system pulse profile data storehouse;
Determine that pulse profile data starting point step comprises: from the pulse profile data storehouse, read pending pulse profile data; The data of reading are begun to find maximum or minimum that moment point of pulse condition information numerical value from original starting point in the cycle that a pulsation is beated; This segment data deletion, obtained demarcating the pulse profile data of ground zero from original starting point to this moment point;
Determine that pulse profile data terminal point step comprises; From the pulse profile data storehouse, read pending pulse profile data; From starting point, after one or more pulsation periods of bypass system appointment, write down this moment point; Extract starting point to this segment data of this moment point as new pulse profile data, handle for next step;
Standardization pulse profile data step comprises; From the pulse profile data storehouse, read pending pulse profile data; Is value between one [1,1] for the value of each pulse profile data with its standard, and update data set; Heredity inspires the pulse condition fundamental function extraction step of evolution algorithm based on biological immune.
2, Chinese medicine pulse fundamental function extraction method according to claim 1 is characterized in that described pulse condition fundamental function extraction step based on biological immune heredity inspiration evolution algorithm may further comprise the steps:
Generate colony randomly; Carry out individual self-tolerance process,, otherwise change next step over to if based on individual tolerance gets nowhere then withers away and promptly delete this individuality from the body storehouse; The individuality of self-tolerance success is joined ripe colony; Check population size,, otherwise jump to the clonal vaviation step if meet the requirements then turn to next step;
Each individuality in the colony of maturation is decoded as functional expression, each functional expression is calculated corresponding fitness on the pulse profile data storehouse; If the fitness of the optimized individual in the mature groups (individuality that promptly can accurately represent the pulse condition feature) reaches the requirement of setting, the function of then exporting this individuality decoding back gained is as the pulse condition fundamental function that characterizes the pulse condition feature, and terminator, otherwise carry out the clonal vaviation step; The clonal vaviation step, the mature groups to previous step in rapid uses immunity/genetic operator to carry out the clonal vaviation operation and produces new colony; Going to individual self-tolerance step cycle carries out.
3, Chinese medicine pulse fundamental function extraction method according to claim 2 is characterized in that described colony is a group of individuals, and individual basic structure is made of the vector of word string vector, and the word string vector is made of by rule the designated character collection.
4, Chinese medicine pulse fundamental function extraction method according to claim 2, it is characterized in that described is a kind of data storage from the body storehouse, can deposit group of individuals or functional expression as required.
5, Chinese medicine pulse fundamental function extraction method according to claim 2, it is characterized in that described individual self-tolerance mode of operation is: comprise suddenly for previous step generating the new individuality that colony and clonal vaviation step are generated at random, if should the decoded functional expression of individuality identical with certain individual decoded functional expression in the body storehouse then tolerance is unsuccessful, otherwise tolerate successfully.
6, Chinese medicine pulse fundamental function extraction method according to claim 1, it is characterized in that described from the body storehouse in the process that algorithm circulation is carried out, be dynamically to adjust and changing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2008101531750A CN101408912B (en) | 2008-11-21 | 2008-11-21 | Method for automatically extracting characteristic function of traditional Chinese medicine pulse manifestation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2008101531750A CN101408912B (en) | 2008-11-21 | 2008-11-21 | Method for automatically extracting characteristic function of traditional Chinese medicine pulse manifestation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101408912A true CN101408912A (en) | 2009-04-15 |
| CN101408912B CN101408912B (en) | 2010-06-09 |
Family
ID=40571924
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2008101531750A Expired - Fee Related CN101408912B (en) | 2008-11-21 | 2008-11-21 | Method for automatically extracting characteristic function of traditional Chinese medicine pulse manifestation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101408912B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2011013007A2 (en) * | 2009-07-29 | 2011-02-03 | Purapharm International (Hk) Limited | Ontological information retrieval system |
| CN103932686A (en) * | 2014-04-22 | 2014-07-23 | 北京印刷学院 | Method and device for extracting pulse condition signal |
| CN104615894A (en) * | 2015-02-13 | 2015-05-13 | 上海中医药大学 | Traditional Chinese medicine diagnosis method and system based on k-nearest neighbor labeled specific weight characteristics |
| CN104820775A (en) * | 2015-04-17 | 2015-08-05 | 南京大学 | Discovery method of core drug of traditional Chinese medicine prescription |
| CN109124595A (en) * | 2018-07-20 | 2019-01-04 | 南开大学 | Intelligent sphygmus diagnostic method based on dynamic 3 D pulse wave image |
| CN113673409A (en) * | 2021-08-16 | 2021-11-19 | 点靓纳谱(上海)生物医药科技有限公司 | Automatic convergence spectrum correction method and device for spectrum analysis |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100493445C (en) * | 2005-11-03 | 2009-06-03 | 浙江大学 | Automatic testing method for traditional Chinese medical pulse manifestation characteristics parameter |
-
2008
- 2008-11-21 CN CN2008101531750A patent/CN101408912B/en not_active Expired - Fee Related
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2011013007A2 (en) * | 2009-07-29 | 2011-02-03 | Purapharm International (Hk) Limited | Ontological information retrieval system |
| WO2011013007A3 (en) * | 2009-07-29 | 2011-04-07 | Purapharm International (Hk) Limited | Ontological information retrieval system |
| US10089391B2 (en) | 2009-07-29 | 2018-10-02 | Herbminers Informatics Limited | Ontological information retrieval system |
| CN103932686A (en) * | 2014-04-22 | 2014-07-23 | 北京印刷学院 | Method and device for extracting pulse condition signal |
| CN104615894A (en) * | 2015-02-13 | 2015-05-13 | 上海中医药大学 | Traditional Chinese medicine diagnosis method and system based on k-nearest neighbor labeled specific weight characteristics |
| CN104615894B (en) * | 2015-02-13 | 2018-09-28 | 上海中医药大学 | A kind of Chinese Medicine Diagnoses System based on k neighbour's label certain weights features |
| CN104820775A (en) * | 2015-04-17 | 2015-08-05 | 南京大学 | Discovery method of core drug of traditional Chinese medicine prescription |
| CN109124595A (en) * | 2018-07-20 | 2019-01-04 | 南开大学 | Intelligent sphygmus diagnostic method based on dynamic 3 D pulse wave image |
| CN113673409A (en) * | 2021-08-16 | 2021-11-19 | 点靓纳谱(上海)生物医药科技有限公司 | Automatic convergence spectrum correction method and device for spectrum analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101408912B (en) | 2010-06-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101408912B (en) | Method for automatically extracting characteristic function of traditional Chinese medicine pulse manifestation | |
| TW202137072A (en) | Method for constructing and operating an artificial recurrent neural network | |
| Alvarez et al. | DARWIN: An evolutionary program for nonlinear modeling of chaotic time series | |
| CN112288021B (en) | Medical wastewater monitoring data quality control method, device and system | |
| CN110363090A (en) | Intelligent heart disease detection method, device and computer readable storage medium | |
| CN114612655B (en) | Vein recognition algorithm transplanting method and device | |
| Hagiwara et al. | BEANS: The benchmark of animal sounds | |
| CN113423005B (en) | Intelligent music generation method and system based on improved neural network | |
| Gan | Signal processing and image processing for acoustical imaging | |
| CN113421546B (en) | Speech synthesis method based on cross-test multi-mode and related equipment | |
| Dellaferrera et al. | A bin encoding training of a spiking neural network based voice activity detection | |
| CN117216522A (en) | Bearing fault type diagnosis method, device and equipment based on machine learning | |
| Kumar et al. | Modified soft rough set based ECG signal classification for cardiac arrhythmias | |
| CN112205990A (en) | Wrist angle prediction method and device based on sEMG under different loads | |
| CN115565669B (en) | Cancer survival analysis method based on GAN and multitask learning | |
| Urdy | Theoretical modelling of the molluscan shell: what has been learned from the comparison among molluscan taxa? | |
| CN116884648A (en) | Voice interaction optimization method, device, equipment and medium based on traditional Chinese medicine inquiry | |
| Haider et al. | Feature extraction for identification of extension and flexion movement of wrist using EMG signals | |
| CN115444419A (en) | Domain-adaptive intelligent emotion recognition method and device based on electroencephalogram signals | |
| CN114936153A (en) | Turing test method of artificial intelligence software | |
| Kneller | Measurement, control and other processes: to the problem of knowledge systematization | |
| Tyburek et al. | Comparison of the efficiency of time and frequency descriptors based on different classification conceptions | |
| CN114617555B (en) | Psychological assessment system, medium and equipment based on physiological characteristic stability detection | |
| CN117454902B (en) | Report analysis method applied to investigation processing system and marine environment investigation system | |
| CN113421643B (en) | AI model reliability judging method, device, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20100609 Termination date: 20111121 |