A method for combining bioelectrical impedance measurement and machine learning to predict the abdominal fat weight of broilers

Abstract

The present invention provides a method for combining bioelectrical impedance measurement and machine learning to predict the abdominal fat weight of broiler chickens, including: measuring bioelectrical impedance and chicken body size data to obtain a data set; constructing a multiple linear regression machine learning model, calculating the root mean square error and goodness of fit, and selecting the optimal measurement points; constructing a ridge regression machine learning model, calculating the root mean square error and goodness of fit, and selecting the optimal measurement points; constructing a lasso regression machine learning model, calculating the root mean square error and goodness of fit, and selecting the optimal measurement points; constructing a logistic regression machine learning model, obtaining the prediction results and calculating the model accuracy; comparing the root mean square errors and goodness of fit of the multiple linear regression machine learning model, the ridge regression machine learning model, and the lasso regression machine learning model, and selecting the learning model with a small root mean square error and a large goodness of fit as the final regression model; obtaining the abdominal fat weight of the chicken to be predicted. The present invention can accurately predict the abdominal fat content of live chickens and reduce the measurement cost.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and more particularly, to a method for predicting the abdominal fat weight of broilers by combining bioelectrical impedance measurement and machine learning. Background Art

[0002] Excessive abdominal fat deposition in chickens not only reduces feed conversion rate, but also affects the egg production performance of breeding hens, the fertilization rate and hatching rate of breeding eggs, increases the incidence of fatty liver, and increases the mortality rate during the laying period. Excessive body fat deposition in chickens has been widely concerned by scholars at home and abroad. Exploring the mechanism of abdominal fat deposition and lipid metabolism in chickens is a hot issue in poultry science. In recent years, extensive research carried out by scientific research workers in aspects such as genetics, nutrition, immunity, and environmental control has shown that through genetic means, breeding varieties with less abdominal fat deposition is an effective means to solve this problem.

[0003] Currently, genetic markers for genetic selection include feed efficiency, abdominal fat weight ratio, body fat traits, etc. From the existing research results, direct selection for the abdominal fat weight ratio trait has the best selection effect, but this method requires slaughtering and measuring chickens, which is labor-intensive, costly, and excellent individuals cannot be selected and retained. Pym and Thompson designed a caliper for in vivo measurement of abdominal thickness, and the correlation between the measured abdominal fat percentage and abdominal thickness was. Mirosh and Becker's research showed the correlation between the abdominal thickness of male chicks and the abdominal fat content, and it was considered that using abdominal thickness to predict body fat content is a simple, rapid, and low-investment method. From the above results, although this method is simple and easy to implement, its accuracy is not high, and the operator needs to be trained in advance, making it difficult to adopt in large-scale commercial breeding.

[0004] Therefore, it is necessary to develop a method that can accurately measure the abdominal fat of broilers without slaughtering for the determination of abdominal fat content. Summary of the Invention

[0005] Based on this, in order to more accurately predict the abdominal fat content of live chickens, the present invention provides a method for predicting the abdominal fat weight of broilers by combining bioelectrical impedance measurement and machine learning, and its specific technical solution is as follows:

[0006] A method for predicting the abdominal fat weight of broilers by combining bioelectrical impedance measurement and machine learning, which includes the following steps:

[0007] Perform bioelectrical impedance and chicken body measurement data, read the chicken body weight, body slant length, waist circumference, and bioelectrical impedance data, and slaughter and measure the abdominal fat weight of the chicken to obtain a data set, wherein the measurement points of bioelectrical impedance include multiple;

[0008] Build a multiple linear regression machine learning model, import the dataset into the multiple linear regression machine learning model, calculate the root mean square error and goodness of fit, compare the goodness of fit of multiple measurement points, and select the optimal measurement point;

[0009] Build a ridge regression machine learning model, import the dataset into the ridge regression machine learning model, calculate the root mean square error and goodness of fit, compare the goodness of fit of multiple measurement points, and select the optimal measurement point;

[0010] Build a lasso regression machine learning model, import the dataset into the lasso regression machine learning model, calculate the root mean square error and goodness of fit, compare the goodness of fit of multiple measurement points, and select the optimal measurement point;

[0011] Build a logistic regression machine learning model, set the data in the dataset that is higher than half of the average abdominal fat weight to 0 and lower than half of the average abdominal fat weight to 1 to obtain a new dataset, import the new dataset into the logistic regression machine learning model, obtain the prediction results and calculate the model accuracy;

[0012] Compare the root mean square errors of the multiple linear regression machine learning model, the ridge regression machine learning model, and the lasso regression machine learning model, compare the goodness of fit of the multiple linear regression machine learning model, the ridge regression machine learning model, and the lasso regression machine learning model, select the learning model with a small root mean square error and a large goodness of fit as the final regression model, and obtain the optimal measurement point corresponding to the final regression model;

[0013] Import the weight, body slant length, waist circumference, and bioelectrical impedance data of the chicken to be predicted into the final regression model to obtain the abdominal fat weight of the chicken to be predicted.

[0014] In the present invention, by building a multiple linear regression machine learning model, a ridge regression machine learning model, and a lasso regression machine learning model, measuring bioelectrical impedance and chicken body size data to obtain a dataset, importing the dataset into the multiple linear regression machine learning model, the ridge regression machine learning model, and the lasso regression machine learning model to obtain the corresponding root mean square error and goodness of fit, selecting the learning model with a small root mean square error and a large goodness of fit as the final regression model, and obtaining the optimal measurement point corresponding to the final regression model, it combines bioelectrical impedance and chicken body size to jointly predict the abdominal fat weight of live chickens, can accurately predict the abdominal fat content of live chickens, and reduces the measurement cost.

[0015] Furthermore, the measurement points of multiple bioelectrical impedances include a first measurement point, a second measurement point, and a third measurement point. The first measurement point is at the muscle between the ulna and radius of the left wing, the second measurement point is at the abdomen below the right side of the keel, and the third measurement point is at the abdomen at the lower edge of the keel.

[0016] Further, a multiple linear regression machine learning model is constructed. Importing the data set into the multiple linear regression machine learning model, the specific methods for calculating the root mean square error and goodness of fit include the following steps: splitting the data set into a training data set and a test data set, modeling according to the training data set, deleting the predictor variables in the test data set, making predictions with the remaining independent variables, calculating the regression coefficients, calculating the mean of the dependent variable in the modeled data, counting the number of variables and the number of observations, and calculating the root mean square error and goodness of fit.

[0017] Further, a ridge regression machine learning model is constructed. Importing the data set into the ridge regression machine learning model, the specific methods for calculating the root mean square error and goodness of fit include the following steps: splitting the data set into a training data set and a test data set, constructing an anonymous function, constructing an empty list to store the partial regression coefficients of the ridge regression machine learning model, iterating through different anonymous functions, plotting the relationship between the anonymous function and the regression coefficients, setting the parameters of cross-validation, performing 10-fold cross-validation for each anonymous function value, fitting the model, returning the best anonymous function, obtaining the regression coefficients, and calculating the root mean square error and goodness of fit.

[0018] Further, a lasso regression machine learning model is constructed. Importing the data set into the lasso regression machine learning model, the specific methods for calculating the root mean square error and goodness of fit include the following steps: splitting the data set into a training data set and a test data set, constructing an anonymous function, constructing an empty list to store the partial regression coefficients of the lasso regression machine learning model, iterating through different anonymous functions, plotting the relationship between the anonymous function and the regression coefficients, setting the parameters of cross-validation, performing 10-fold cross-validation for each anonymous function value, fitting the model, returning the best anonymous function, obtaining the regression coefficients, and calculating the root mean square error and goodness of fit.

[0019] Further, the output regression prediction equation of the final regression model is y = a * x1 + b * x2 + c * x3 + d * x4;

[0020] where y is the abdominal fat weight, x1 is the body weight, x2 is the body slant length, x3 is the waist circumference, x4 is the electrical resistance, and a, b, c, and d are constants.

[0021] A computer-readable storage medium stores a computer program which, when executed by a processor, implements the method for combining bioelectrical impedance measurement and machine learning to predict the abdominal fat weight of broilers. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The present invention can be further understood from the following description in conjunction with the accompanying drawings. The components in the drawings are not necessarily drawn to scale, but the emphasis is placed on showing the principles of the embodiments. In different views, the same reference numerals designate corresponding parts.

[0023] Figure 1 This is a schematic diagram of the overall structure of a method for combining bioelectrical impedance measurement and machine learning to predict the abdominal fat weight of broilers in an embodiment of the present invention;

[0024] Figure 2 This is a schematic diagram of the bioelectrical impedance measurement points in an embodiment of the present invention. Specific embodiments

[0025] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with its embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and do not limit the protection scope of the present invention.

[0026] It should be noted that when an element is referred to as being "fixed to" another element, it can be directly on the other element or there may also be an intermediate element. When an element is considered to be "connected" to another element, it can be directly connected to the other element or there may be an intermediate element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are only for the purpose of illustration and do not represent the only implementation.

[0027] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0028] The "first" and "second" in the present invention do not represent specific quantities and orders, but are only used for name distinction.

[0029] As Figure 1 shown, a method for combining bioelectrical impedance measurement and machine learning to predict the abdominal fat weight of broilers in an embodiment of the present invention includes the following steps:

[0030] S1. Perform bioelectrical impedance and chicken body size data measurement, read the chicken body weight, body slant length, waist circumference and bioelectrical impedance data, and slaughter and measure the abdominal fat weight of the chicken to obtain a data set. Among them, there are multiple measurement points for bioelectrical impedance.

[0031] Specifically, as Figure 2 shown, the multiple measurement points for bioelectrical impedance include a first measurement point, a second measurement point and a third measurement point. The first measurement point is at the muscle between the ulna and radius of the left wing, that is, Figure 2 AB in; the second measurement point is in the abdomen below the right side of the keel, that is, Figure 2the CD in; the third measurement point is on the abdomen at the lower edge of the keel, that is Figure 2 the EF in.

[0032] The chicken body size data is measured as follows: body weight: weigh the live chicken before slaughter (kg); body slant length: measure the distance between the shoulder joint and the ischial tuberosity along the body surface using a tape measure (cm); waist circumference: measure the length of a circle around the body vertically upward from the back roots of the two legs using a tape measure (cm); abdominal fat weight: weigh the fat around the abdomen and gizzard after peeling (g).

[0033] S2. Build a multiple linear regression machine learning model, import the data set into the multiple linear regression machine learning model, calculate the root mean square error and the goodness of fit, compare the goodness of fit of multiple measurement points, and select the optimal measurement point.

[0034] S3. Build a ridge regression machine learning model, import the data set into the ridge regression machine learning model, calculate the root mean square error and the goodness of fit, compare the goodness of fit of multiple measurement points, and select the optimal measurement point.

[0035] S4. Build a lasso regression machine learning model, import the data set into the lasso regression machine learning model, calculate the root mean square error and the goodness of fit, compare the goodness of fit of multiple measurement points, and select the optimal measurement point.

[0036] S5. Build a logistic regression machine learning model, set the data in the data set that is higher than half of the average abdominal fat weight to 0 and the data that is lower than half of the average abdominal fat weight to 1 to obtain a new data set, import the new data set into the logistic regression machine learning model, obtain the prediction result and calculate the model accuracy.

[0037] Accuracy is one of the criteria for evaluating the quality of a regression model. The higher the accuracy, the better the regression model. By calculating the model accuracy, the prediction effect of the regression machine learning model can be judged.

[0038] Logistic regression is a regression equation for predicting the level of abdominal fat content. The new data set is brought in to calculate the level of predicted abdominal fat weight, which is used to classify the abdominal fat weight and determine the fatness of the chicken.

[0039] S******6. Compare the root mean square errors of the multiple linear regression machine learning model, the ridge regression machine learning model, and the lasso regression machine learning model, compare the goodness of fit of the multiple linear regression machine learning model, the ridge regression machine learning model, and the lasso regression machine learning model, select the learning model with a small root mean square error and a large goodness of fit as the final regression model, and obtain the optimal measurement point corresponding to the final regression model.

[0040] That is to say, the method for selecting a learning model with a small root mean square error and a large goodness of fit as the final regression model is as follows: when the root mean square error of one of the multiple linear regression machine learning model, the ridge regression machine learning model, and the lasso regression machine learning model is the smallest and the goodness of fit is the largest, select the learning model with the smallest root mean square error and the largest goodness of fit as the final regression model; otherwise, select a learning model with a larger goodness of fit from the two learning models with smaller root mean square errors, or select a learning model with a larger root mean square error from the two learning models with a larger goodness of fit as the final regression model.

[0041] S7. Import the weight, body slant length, waist circumference, and bioelectrical impedance data of the chicken to be predicted into the final regression model to obtain the abdominal fat weight of the chicken to be predicted.

[0042] In the present invention, by constructing a multiple linear regression machine learning model, a ridge regression machine learning model, and a lasso regression machine learning model, the bioelectrical impedance and chicken body size data are measured to obtain a data set. The data set is imported into the multiple linear regression machine learning model, the ridge regression machine learning model, and the lasso regression machine learning model to obtain the corresponding root mean square error and goodness of fit. A learning model with a small root mean square error and a large goodness of fit is selected as the final regression model, and the optimal measurement points corresponding to the final regression model are obtained. By combining bioelectrical impedance and chicken body size, the abdominal fat weight of live chickens is predicted, which can accurately predict the abdominal fat content of live chickens and reduce the measurement cost.

[0043] In one embodiment, to construct a multiple linear regression machine learning model and import the data set into the multiple linear regression machine learning model, the specific method for calculating the root mean square error and goodness of fit includes the following steps: split the data set into a training data set and a test data set, model according to the training data set, delete the predictive variables in the test data set, use the remaining independent variables for prediction, calculate the regression coefficients, calculate the mean of the dependent variable in the modeling data, count the number of variables and the number of observations, and calculate the root mean square error and goodness of fit.

[0044] Specifically, import the data set into the multiple linear regression machine learning model, calculate the relevant parameters, calculate the regression deviation sum of squares and the error sum of squares, and then calculate the root mean square error.

[0045] By calculating the joint hypothesis test statistic value, return the joint hypothesis test statistic value in the model and calculate the goodness of fit.

[0046] In one embodiment, a ridge regression machine learning model is constructed. The specific method for importing a data set into the ridge regression machine learning model and calculating the root mean square error and goodness of fit includes the following steps: splitting the data set into a training data set and a test data set, constructing an anonymous function, constructing an empty list to store the partial regression coefficients of the ridge regression machine learning model, iteratively looping through different anonymous functions, plotting the relationship between the anonymous function and the regression coefficients, setting the parameters for cross-validation, performing 10-fold cross-validation for each anonymous function value, fitting the model, returning the best anonymous function, obtaining the regression coefficients, and calculating the root mean square error and goodness of fit.

[0047] In one embodiment, a lasso regression machine learning model is constructed. The specific method for importing a data set into the lasso regression machine learning model and calculating the root mean square error and goodness of fit includes the following steps: splitting the data set into a training data set and a test data set, constructing an anonymous function, constructing an empty list to store the partial regression coefficients of the lasso regression machine learning model, iteratively looping through different anonymous functions, plotting the relationship between the anonymous function and the regression coefficients, setting the parameters for cross-validation, performing 10-fold cross-validation for each anonymous function value, fitting the model, returning the best anonymous function, obtaining the regression coefficients, and calculating the root mean square error and goodness of fit.

[0048] The output regression prediction equation of the final regression model is y = a * x1 + b * x2 + c * x3 + d * x4;

[0049] Where y is the abdominal fat weight in g, x1 is the body weight in kg, x2 is the body slant length in cm, x3 is the waist circumference in cm, x4 is the electrical resistance, and a, b, c, and d are constants.

[0050] In summary, the method for combining bioelectrical impedance measurement and machine learning to predict the abdominal fat weight of broiler chickens combines bioelectrical impedance and chicken body dimensions to jointly predict the abdominal fat weight of live chickens, can accurately predict the abdominal fat content of live chickens, does not require slaughtering live chickens for measurement, and reduces the measurement cost.

[0051] In addition, by predicting the abdominal fat content of live chickens without slaughtering live chickens for measurement, the present invention also reduces the loss during the selection process when excellent individuals are slaughtered to measure performance, and improves the breeding efficiency. The machine learning model used in the present invention can provide certain reference for the future development of new regression models and play a reference role in large-scale farming, scientific breeding, etc.

[0052] A computer-readable storage medium stores a computer program, which implements the method for combining bioelectrical impedance measurement and machine learning to predict the abdominal fat weight of broiler chickens when the computer program is executed by a processor.

[0053] The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0054] The above-described embodiments only express several implementation manners of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the appended claims.

Claims

1. A method for combining bioelectrical impedance measurement and machine learning to predict the abdominal fat weight of broilers, characterized in that, It includes the following steps: Perform bioelectrical impedance and chicken body size data measurement, read the chicken body weight, body slant length, waist circumference, and bioelectrical impedance data, and slaughter the chicken to measure the abdominal fat weight to obtain a data set. Among them, there are multiple measurement points for bioelectrical impedance; Construct a multiple linear regression machine learning model, import the data set into the multiple linear regression machine learning model, calculate the root mean square error and goodness of fit, compare the goodness of fit of multiple measurement points, and select the optimal measurement point; Construct a ridge regression machine learning model, import the data set into the ridge regression machine learning model, calculate the root mean square error and goodness of fit, compare the goodness of fit of multiple measurement points, and select the optimal measurement point; Construct a lasso regression machine learning model, import the data set into the lasso regression machine learning model, calculate the root mean square error and goodness of fit, compare the goodness of fit of multiple measurement points, and select the optimal measurement point; Construct a logistic regression machine learning model, set the data in the data set that is higher than half of the average abdominal fat weight to 0 and lower than half of the average abdominal fat weight to 1 to obtain a new data set, import the new data set into the logistic regression machine learning model, obtain the prediction result and calculate the model accuracy; Compare the root mean square errors of the multiple linear regression machine learning model, the ridge regression machine learning model, and the lasso regression machine learning model, compare the goodness of fit of the multiple linear regression machine learning model, the ridge regression machine learning model, and the lasso regression machine learning model, select the learning model with a small root mean square error and a large goodness of fit as the final regression model, and obtain the optimal measurement point corresponding to the final regression model; Import the body weight, body slant length, waist circumference, and bioelectrical impedance data of the chicken to be predicted into the final regression model to obtain the abdominal fat weight of the chicken to be predicted.

2. The method for combining bioelectrical impedance measurement and machine learning to predict the abdominal fat weight of broilers according to claim 1, wherein, The multiple bioelectrical impedance measurement points include a first measurement point, a second measurement point, and a third measurement point. The first measurement point is at the muscle between the ulna and radius of the left wing, the second measurement point is at the abdomen below the right side of the keel, and the third measurement point is at the abdomen of the lower edge of the keel.

3. A method for combining bioelectrical impedance measurement and machine learning to predict the abdominal fat weight of broilers according to claim 2, wherein Construct a multiple linear regression machine learning model, import the data set into the multiple linear regression machine learning model. The specific methods for calculating the root mean square error and goodness of fit include the following steps: Split the data set into a training data set and a test data set, model according to the training data set, delete the predictor variables in the test data set, use the remaining independent variables for prediction, calculate the regression coefficients, calculate the mean of the dependent variable in the modeling data, count the number of variables and the number of observations, and calculate the root mean square error and goodness of fit.

4. The method for combining bioelectrical impedance measurement and machine learning to predict the abdominal fat weight of broilers according to claim 3, wherein Construct a ridge regression machine learning model, import the data set into the ridge regression machine learning model. The specific methods for calculating the root mean square error and goodness of fit include the following steps: Split the data set into a training data set and a test data set, construct an anonymous function, construct an empty list to store the partial regression coefficients of the ridge regression machine learning model, loop and iterate different anonymous functions, draw the relationship between the anonymous function and the regression coefficients, set the parameters of cross-validation, for each anonymous function value, perform 10-fold cross-validation, perform model fitting, return the best anonymous function, obtain the regression coefficients, and calculate the root mean square error and goodness of fit.

5. A method for combining bioelectrical impedance measurement and machine learning to predict the abdominal fat weight of broilers according to claim 4, characterized in that The specific methods for constructing a lasso regression machine learning model, importing a dataset into the lasso regression machine learning model, and calculating the root mean square error and goodness of fit include the following steps: splitting the dataset into a training dataset and a test dataset, constructing an anonymous function, constructing an empty list to store the partial regression coefficients of the lasso regression machine learning model, iteratively looping through different anonymous functions, plotting the relationship between the anonymous function and the regression coefficients, setting the parameters for cross-validation, performing 10-fold cross-validation for each anonymous function value, fitting the model, returning the best anonymous function, obtaining the regression coefficients, and calculating the root mean square error and goodness of fit.

6. A method for combining bioelectrical impedance measurement and machine learning to predict the abdominal fat weight of broilers according to claim 5, characterized in that, The output regression prediction equation of the final regression model is y = a * x1 + b * x2 + c * x3 + d * x4; where y is the abdominal fat weight, x1 is the body weight, x2 is the body slant length, x3 is the waist circumference, x4 is the electrical resistance, and a, b, c, and d are constants.

7. A computer-readable storage medium, characterized in that, A computer program is stored, and when the computer program is executed by a processor, it implements the method for predicting the abdominal fat weight of broilers by combining bioelectrical impedance measurement and machine learning as described in any one of claims 1-6.

Images