CN106339754A - Method of using genetic algorithm improved neural network modeling in human body composition analysis - Google Patents
Method of using genetic algorithm improved neural network modeling in human body composition analysis Download PDFInfo
- Publication number
- CN106339754A CN106339754A CN201610712139.8A CN201610712139A CN106339754A CN 106339754 A CN106339754 A CN 106339754A CN 201610712139 A CN201610712139 A CN 201610712139A CN 106339754 A CN106339754 A CN 106339754A
- Authority
- CN
- China
- Prior art keywords
- human body
- neural network
- layer
- genetic algorithm
- weight
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/17—Function evaluation by approximation methods, e.g. inter- or extrapolation, smoothing, least mean square method
Abstract
A method of using genetic algorithm improved neural network modeling in human body composition analysis comprises the following steps: S1, collecting multiple human body parameters, including bioelectricity impedance, height and weight; S2, determining the initial weight and offset of each parameter; S3, building a neural network model; S4, inputting the initial weight and offset of each parameter to the neural network model, and then outputting optimal weight and offset; and S5, determining human body composition content values according to the optimal weight and offset. The method of using genetic algorithm improved neural network modeling in human body composition analysis has the advantage that the precision of human body composition measurement by a bioelectrical impedance method can be improved.
Description
Technical field
The present invention relates to human body adult's detection method, the improved neural net model establishing of especially a kind of genetic algorithm is used for people
The method of body composition analysis.
Background technology
It is a kind of indirect method of assessment body composition based on the body composition test of bio-electrical impedance principle, its ultimate principle
It is that faint ac signal is imported human body, obtains the electrical impedance of human body various pieces by measurement, analysis obtains corresponding portion
The constituent dividing.This method have property easy to use and non-invasive the features such as, quickly grow in recent years, be applicable to house
Front yard, Community Doctor, Health Club and hospital.After recording human-body biological impedance, calculate human body according to various empirical equations
Component ratio.Empirical equation typically passes through resistance value and the specialties such as dual energy x-ray absorption measuring method of biological impedance gained
Direct test measured result carries out regression analyses and obtains.Existing regression analysis typically take linear regression mode.Linearly
Regression model assumes that the relation between other specification and human body component such as bio-electrical impedance and height, body weight is linear, actual
It is a kind of linear modelling, simple and convenience is mainly pursued in this modeling, the deviation existing between model and practical situation can lead to
The goodness of fit is poor.
Content of the invention
In view of above-mentioned condition is it is necessary to provide a kind of improved neutral net of genetic algorithm improving survey calculation precision
Model the method for bioelectrical impedance analysis.
For solving above-mentioned technical problem, provide a kind of improved neural net model establishing of genetic algorithm for bioelectrical impedance analysis
Method, comprise the steps:
S1, the multinomial human-body biological electrical impedance of collection and height, body weight parameters;
S2, the initial weight determining parameters and side-play amount;
S3, structure neural network model;
S4, by after the weight of parameters and offset data input neural network model, output optimal weights and side-play amount;
S5, determine human body component content value according to optimal weights and side-play amount.
It is used in the method for bioelectrical impedance analysis in the improved neural net model establishing of the above-mentioned genetic algorithm of the present invention, described step
Neural network model in rapid s3 includes input layer, hidden layer and output layer, and described input layer includes multinomial human parameterss variable
Unit, described output layer includes human body component parameter value cell, people's not parametric variable is passed to described hidden by described input layer
Obtain connection weight containing layer, described hidden layer connects described input layer and described output layer respectively, will be connected by described hidden layer
Connect weights and pass to described output layer, described output layer passes through the optimum human body component parameter value of human body component parameters unit output.
It is used in the method for bioelectrical impedance analysis in the improved neural net model establishing of the above-mentioned genetic algorithm of the present invention, described step
Rapid s4 specifically includes following steps:
S41, using each layer weights and side-play amount as gene, generate gene group at random;
The goodness of fit to human body component parameter for each body acupuncture in s42, one by one calculating gene group;
In s43, selection gene group, the preferably outstanding genetic entities of the goodness of fit carry out cut-out restructuring, make criss-cross inheritance operation;
S44, the new gene group that criss-cross inheritance operation is produced, choose a part of individuality and carry out random variation operation;
S45, the new gene group with generating after criss-cross inheritance and mutation operation, calculate the goodness of fit of each individuality one by one;
S46, when through multiple step s42 to s45 iteration, after obtaining optimum human body component parameter, stop iteration, obtain optimal base
Because of individuality.
The present invention takes the mode of neural net model establishing, and is improved with genetic algorithm.The advantage of neural net model establishing
It is can fully to approach arbitrarily complicated non-linear relation, can learn not know with self adaptation or uncertain system.And it is hereditary
The learning rules of neutral net can be realized Automatic Optimal and can improve the optimization speed of neutral net weight coefficient by the introducing of algorithm
Degree.Compared with prior art,
The present invention, due to can fully approach arbitrarily complicated non-linear relation using based on neural net model establishing, can learn and self adaptation
The advantage of uncertain system, realizes the optimum modeling between parameter and human body component such as bio-electrical impedance, height, body weight;Profit
With the feature of genetic algorithm, acceleration is optimized to neural network model;Improve the goodness of fit thus improving bio-electrical impedance side
Method measures the precision of human body component.Human-body biological impedance parameter is calculated, the human body component parameter obtaining and dual energy x-ray
The specialty such as absorption measuring method directly tests measured result in terms of the statistical indicators such as dependency, significant difference substantially due to tradition
Linear regression empirical equation.
Brief description
Fig. 1 is the theory structure of the method that the improved neural net model establishing of genetic algorithm of the present invention is used for bioelectrical impedance analysis
Figure.
Specific embodiment
Below in conjunction with drawings and Examples, to genetic algorithm of the present invention, improved neural net model establishing is used for human body component
The method of analysis is described in further detail.
The method that a kind of improved neural net model establishing of genetic algorithm of the embodiment of the present invention is used for bioelectrical impedance analysis, bag
Include following steps: s1, gather multinomial human-body biological electrical impedance and height, body weight parameters, by testing to some samples, note
Record data, test content includes bioelectric impedance value and the parameter such as corresponding height, body weight of human body;S2, determine parameters
Initial weight and side-play amount, by gathered data is carried out with mathematical statistics, count weight and the offset value of each item data;
S3, structure neural network model, by determining Artificial Neural Network Structures, set functional value;S4, by the weight of parameters and
After offset data input neural network model, export optimal weights and side-play amount;S5, determined according to optimal weights and side-play amount
Human body component content value.
The method that the improved neural net model establishing of genetic algorithm of the present invention is used for bioelectrical impedance analysis, as shown in figure 1, step
Neural network model in s3 includes input layer, hidden layer and output layer, and described input layer includes multinomial human parameterss variable list
Unit, described input layer passes to described hidden layer by the first transmission function f parameter value wij, and described hidden layer includes multinomial
The parameters unit corresponding with described input layer, described hidden layer connects described input layer and described output layer respectively, described hidden
Containing layer, described output layer is passed to parameter value wjk by the second transmission function g, described output layer includes multinomial human body component parameter
Unit, exports optimal weights and side-play amount by human body component parameters unit.
As shown in figure 1, the improved neural net model establishing of genetic algorithm of the present invention is used in the method for bioelectrical impedance analysis, institute
State step s4 and specifically include following steps:
S41, using each layer weights and side-play amount as gene, generate gene group, each layer weights x1, x2, x3 ... as described in Figure 1 at random
Xn and side-play amount wij generate gene group at random;
The goodness of fit to human body component parameter for each body acupuncture in s42, one by one calculating gene group;
In s43, selection gene group, the preferably outstanding genetic entities of the goodness of fit carry out cut-out restructuring, make criss-cross inheritance operation;
S44, the new gene group yj that criss-cross inheritance operation is produced, choose a part of individuality and carry out random variation operation;
S45, the new gene group with generating after criss-cross inheritance and mutation operation, calculate the goodness of fit tk, zk of each individuality one by one;
S46, when through multiple step s42 to s45 iteration, after obtaining optimum human body component parameter, stop iteration, obtain optimal base
Because of individuality.
The present invention builds gene with each weight and side-play amount, generates a number of initial gene group at random, with pin
The goodness of fit alternatively criterion to test sample, chooses parent gene to the gene in initial gene group and carries out cutting restructuring
Heredity and mutation operation, so through the corresponding gene of best fit goodness excessively for genetic iteration, can be obtained, also just obtain
Optimal weights and side-play amount.
The above, be only presently preferred embodiments of the present invention, not the present invention is made with any pro forma restriction, though
So the present invention is disclosed above with preferred embodiment, but is not limited to the present invention, any is familiar with this professional technology people
Member, in the range of without departing from technical solution of the present invention, when the technology contents of available the disclosure above make a little change or modification
For the Equivalent embodiments of equivalent variations, as long as being without departing from technical solution of the present invention content, according to the technical spirit pair of the present invention
Any simple modification, equivalent variations and modification that above example is made, all still fall within the range of technical solution of the present invention.
Claims (3)
1. a kind of improved neural net model establishing of genetic algorithm be used for bioelectrical impedance analysis method it is characterised in that include as
Lower step:
S1, the multinomial human-body biological electrical impedance of collection and height, body weight parameters;
S2, the initial weight determining parameters and side-play amount;
S3, structure neural network model;
S4, by after the weight of parameters and offset data input neural network model, output optimal weights and side-play amount;
S5, determine human body component content value according to optimal weights and side-play amount.
2. the method that the improved neural net model establishing of genetic algorithm as claimed in claim 1 is used for bioelectrical impedance analysis, it is special
Levy and be: the neural network model in described step s3 includes input layer, hidden layer and output layer, described input layer includes multinomial
Human parameterss variable cell, described output layer includes human body component parameter value cell, and people is stopped parametric variable by described input
Layer is passed to described hidden layer and is obtained connection weight, and described hidden layer connects described input layer and described output layer respectively, by institute
State hidden layer and connection weight is passed to described output layer, described output layer passes through the optimum human body of human body component parameters unit output and becomes
Divide parameter value.
3. the method that the improved neural net model establishing of genetic algorithm as claimed in claim 2 is used for bioelectrical impedance analysis, it is special
Levy and be, described step s4 specifically includes following steps:
S41, using each layer weights and side-play amount as gene, generate gene group at random;
The goodness of fit to human body component parameter for each body acupuncture in s42, one by one calculating gene group;
In s43, selection gene group, the preferably outstanding genetic entities of the goodness of fit carry out cut-out restructuring, make criss-cross inheritance operation;
S44, the new gene group that criss-cross inheritance operation is produced, choose a part of individuality and carry out random variation operation;
S45, the new gene group with generating after criss-cross inheritance and mutation operation, calculate the goodness of fit of each individuality one by one;
S46, when through multiple step s42 to s45 iteration, after obtaining optimum human body component parameter, stop iteration, obtain optimal base
Because of individuality.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610712139.8A CN106339754A (en) | 2016-08-23 | 2016-08-23 | Method of using genetic algorithm improved neural network modeling in human body composition analysis |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610712139.8A CN106339754A (en) | 2016-08-23 | 2016-08-23 | Method of using genetic algorithm improved neural network modeling in human body composition analysis |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106339754A true CN106339754A (en) | 2017-01-18 |
Family
ID=57824671
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610712139.8A Pending CN106339754A (en) | 2016-08-23 | 2016-08-23 | Method of using genetic algorithm improved neural network modeling in human body composition analysis |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106339754A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109497967A (en) * | 2019-01-04 | 2019-03-22 | 济南汇医融工科技有限公司 | A kind of human body component, stress and artery sclerosis sync detection device |
CN109871953A (en) * | 2019-01-25 | 2019-06-11 | 浙江大学 | The heavy oil pyrolysis process wavelet neural network modeling method of fpRNA genetic algorithm |
WO2021027295A1 (en) * | 2019-08-12 | 2021-02-18 | 岭南师范学院 | Human body composition prediction method based on improved adaptive genetic algorithm |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1517949A (en) * | 2003-01-17 | 2004-08-04 | 清华大学 | Determination method of artificial nerve network meaule parameter for cvomprehensive evaluation of human physique |
US20110245710A1 (en) * | 2009-10-01 | 2011-10-06 | Seca Ag | Apparatus and method for bioelectrical impedance measurements |
CN103123669A (en) * | 2013-02-28 | 2013-05-29 | 大连大学 | Human body composition analysis method based on genetic algorithm |
CN103637800A (en) * | 2013-12-20 | 2014-03-19 | 大连大学 | Eight-section impedance model based body composition analysis method |
CN104434098A (en) * | 2014-11-27 | 2015-03-25 | 北京四海华辰科技有限公司 | Human body energy detection method and device |
CN205322327U (en) * | 2015-12-31 | 2016-06-22 | 华南理工大学 | Wearable foot ring based on human composition of bio -electrical impedance measurable quantity |
US20160235309A1 (en) * | 2011-07-08 | 2016-08-18 | Lifeq Global Limited | Personalized Nutritional and Wellness Assistant |
-
2016
- 2016-08-23 CN CN201610712139.8A patent/CN106339754A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1517949A (en) * | 2003-01-17 | 2004-08-04 | 清华大学 | Determination method of artificial nerve network meaule parameter for cvomprehensive evaluation of human physique |
US20110245710A1 (en) * | 2009-10-01 | 2011-10-06 | Seca Ag | Apparatus and method for bioelectrical impedance measurements |
US20160235309A1 (en) * | 2011-07-08 | 2016-08-18 | Lifeq Global Limited | Personalized Nutritional and Wellness Assistant |
CN103123669A (en) * | 2013-02-28 | 2013-05-29 | 大连大学 | Human body composition analysis method based on genetic algorithm |
CN103637800A (en) * | 2013-12-20 | 2014-03-19 | 大连大学 | Eight-section impedance model based body composition analysis method |
CN104434098A (en) * | 2014-11-27 | 2015-03-25 | 北京四海华辰科技有限公司 | Human body energy detection method and device |
CN205322327U (en) * | 2015-12-31 | 2016-06-22 | 华南理工大学 | Wearable foot ring based on human composition of bio -electrical impedance measurable quantity |
Non-Patent Citations (1)
Title |
---|
胡彬: "基于人工神经网络的人体成分测量仪原型系统研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109497967A (en) * | 2019-01-04 | 2019-03-22 | 济南汇医融工科技有限公司 | A kind of human body component, stress and artery sclerosis sync detection device |
CN109871953A (en) * | 2019-01-25 | 2019-06-11 | 浙江大学 | The heavy oil pyrolysis process wavelet neural network modeling method of fpRNA genetic algorithm |
CN109871953B (en) * | 2019-01-25 | 2020-09-15 | 浙江大学 | Wavelet neural network modeling method for heavy oil cracking process of fpRNA genetic algorithm |
WO2021027295A1 (en) * | 2019-08-12 | 2021-02-18 | 岭南师范学院 | Human body composition prediction method based on improved adaptive genetic algorithm |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106897570A (en) | A kind of COPD test system based on machine learning | |
CN106295205A (en) | Body fat percentage measuring method based on BP neutral net and application thereof | |
CN106667493A (en) | Human body balance assessment system and assessment method | |
KR20190019397A (en) | System and method for providing individual customized health management | |
CN106339754A (en) | Method of using genetic algorithm improved neural network modeling in human body composition analysis | |
CN102307524B (en) | System and method for characteristic parameter estimation of gastric impedance spectra in humans | |
CN106295805A (en) | Human body maximal oxygen uptake evaluation methodology based on BP neutral net and application thereof | |
CN104545912A (en) | Cardiac and pulmonary impedance measuring method and device | |
CN107152995B (en) | The method for quantitatively evaluating of test repeatability in a kind of vehicle impact testing | |
CN104462744A (en) | Data quality control method suitable for cardiovascular remote monitoring system | |
CN113133762B (en) | Noninvasive blood glucose prediction method and device | |
CN107890342A (en) | Perform person under inspection's measurement | |
Mylott et al. | Bioelectrical impedance analysis as a laboratory activity: At the interface of physics and the body | |
CN105354414A (en) | Hierarchical analysis-based human body health condition assessment method | |
CN108985278A (en) | A kind of construction method of the gait function assessment models based on svm | |
CN109009148A (en) | A kind of gait function appraisal procedure | |
CN110210727A (en) | A kind of index system goodness integrated evaluating method of Kernel-based methods behavioral data | |
Topalovic et al. | Modelling the dynamics of expiratory airflow to describe chronic obstructive pulmonary disease | |
CN116453656B (en) | Psychological health assessment early warning system and psychological health assessment early warning method | |
CN106295202B (en) | Juvenile healthy situation dynamic analysing method based on Hale indexes | |
CN116861252A (en) | Method for constructing fall evaluation model based on balance function abnormality | |
Sintonen | Comparing properties of the 15D and the EQ-5D in measuring health-related quality of life | |
KR102477592B1 (en) | Sarcopenia artificial intelligence diagnosis system using body fluid samples and method thereof | |
CN115089147A (en) | Blood pressure measuring device | |
Altunay et al. | A new approach to urinary system dynamics problems: Evaluation and classification of uroflowmeter signals using artificial neural networks |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | 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: 20170118 |
|
RJ01 | Rejection of invention patent application after publication |