Livestock and poultry growth phenotype early prediction method and livestock and poultry growth recovery capability evaluation method

Abstract

The invention provides a livestock and poultry growth phenotype early prediction method and a livestock and poultry growth recovery capability evaluation method. The early prediction method comprises the following steps: continuously collecting an early daily gain sequence and an early daily feed intake sequence of each animal in an early stage of a monitoring period; calculating the maximum correlation coefficient of the early-stage daily gain sequence and the early-stage daily feed intake sequence as an early-stage feed-growth coupling strength index; and predicting the growth phenotype category of each animal in the early stage according to the early feed-growth coupling strength index and the early average daily gain. According to the method, by analyzing the feed-growth coupling strength instead of only paying attention to the absolute growth rate, early warning is conducted on the early growth type of each animal, individuals with different recovery potentials can be recognized before growth trajectory differentiation (such as an early inhibition stage), and then early intervention can be achieved in advance.

Description

Early prediction methods for livestock and poultry growth phenotypes and methods for assessing livestock and poultry growth recovery capacity Technical Field

[0001] This invention relates to the field of livestock and poultry breeding technology, and more specifically, to a method for early prediction of livestock and poultry growth phenotypes and a method for assessing livestock and poultry growth recovery capacity. Background Technology

[0002] In modern livestock and poultry farming, the growth and fattening stage is a critical period that determines the economic benefits of farming. Taking pig farming as an example, the first few weeks after weaning (the nursery period, usually 0-28 days after weaning) is a critical window that determines the animal's lifelong growth performance. During this period, animals face multiple metabolic demands, including skeletal muscle growth, gastrointestinal maturation, and immune system development. However, various stress factors in commercial production environments (such as weaning, transportation, group mixing, feed conversion, and pathogen exposure) may cause some animals to exhibit early growth inhibition.

[0003] A striking biological phenomenon is the significant difference in recovery ability among individuals after experiencing early growth inhibition. Some animals recover quickly and exhibit compensatory growth (accelerating growth to make up for early losses), while others remain in a state of growth inhibition, unable to achieve effective growth recovery even with sufficient or increased feed intake. This heterogeneity in recovery ability has profound implications for farming efficiency, animal welfare, and economic sustainability: individuals with poor recovery ability consume large amounts of feed resources but contribute little to output, while individuals with strong recovery ability can optimize resource utilization and reduce production costs.

[0004] In recent years, advancements in precision animal husbandry technologies have made high-frequency data acquisition possible. The application of automated weighing systems, RFID-based feed intake monitoring equipment, and edge computing capabilities has enabled the acquisition of vast amounts of daily body weight and feed intake data for animals in both commercial and research environments. This high-resolution time-series data provides unprecedented opportunities for in-depth research into the temporal dynamics of growth recovery.

[0005] Currently, livestock and poultry growth performance assessment mainly relies on the following technical solutions:

[0006] (1) Traditional endpoint indicator evaluation method.

[0007] Existing technologies primarily use end-point indicators for evaluation throughout the entire production cycle, including: total weight gain, average daily weight gain over the entire period, and feed conversion ratio. These indicators are calculated by subtracting the beginning weight from the ending weight and dividing the cumulative feed intake by the cumulative weight gain, and are used for production management and genetic selection decisions.

[0008] (2) Stage growth curve fitting method.

[0009] Growth curve models such as Gompertz, Logistic, or Richards are used to fit body weight time series and extract features such as growth rate parameters and inflection point time to describe the growth patterns of individuals or groups.

[0010] (3) Unsupervised clustering classification method.

[0011] Some studies have used unsupervised learning methods such as K-means clustering and Gaussian mixture models to classify animals into different groups based on growth performance indicators, in an attempt to identify different growth patterns or phenotypes.

[0012] The disadvantages of the above-mentioned existing technical solutions are as follows:

[0013] (1) Disadvantages of the endpoint indicator evaluation method:

[0014] Causal reasoning: Endpoint indicators (such as average daily weight gain and feed conversion ratio) compress the data of the entire monitoring period into a single value. This approach masks the temporal dynamic changes during the growth process, fails to capture the recovery pattern after early growth inhibition, cannot distinguish between compensatory growth that is initially suppressed and then recovers and growth failure that is continuously inefficient, and cannot identify problematic individuals in the early stages of production, thus missing the best intervention opportunity and resulting in waste of feed resources and economic losses.

[0015] (2) Disadvantages of the staged growth curve fitting method:

[0016] Causal reasoning: The growth curve model assumes that growth follows a pre-defined mathematical function (such as an S-curve). In actual production, the stress faced by animals is diverse and dynamic, and the growth trajectory often deviates from the ideal curve shape. Forced fitting reduces the biological interpretability of the model parameters. More importantly, this method only focuses on the time series of body weight and ignores the temporal coupling relationship between feed intake and growth. It cannot identify the metabolic disorder state of 'normal feed intake but inefficient growth', and therefore cannot achieve early warning.

[0017] (3) Disadvantages of unsupervised clustering classification:

[0018] Causal reasoning: Unsupervised clustering methods are sensitive to initialization parameters and the number of clusters, and different runs may produce different classification results. The cluster boundaries lack clear biological definitions, making it difficult to reproduce and generalize across different batches and scenarios. Furthermore, clustering methods do not consider the dynamic coupling relationship between feed intake and weight gain, and cannot reveal the intrinsic mechanism differences between successful and failed recovery.

[0019] The common drawback of the above three existing technical solutions is:

[0020] Current technologies have not addressed the temporal coupling dynamics between feed intake and growth. However, research indicates that during the early stages of growth inhibition, individuals exhibiting similar growth rates may show significant differences in the strength of their feed-growth coupling (e.g., a 2.3-fold difference in correlation coefficients, from 0.74 to 0.32). Individuals with high coupling maintain intact metabolic regulation and possess subsequent recovery potential, while those with low coupling exhibit metabolic disorders and struggle to achieve effective recovery. This difference in coupling can be detected before growth trajectory differentiation, providing crucial information for early warning, yet current technologies completely ignore this dimension. Summary of the Invention

[0021] This invention addresses the technical problems existing in the prior art by providing an early prediction method for livestock and poultry growth phenotypes and a method for assessing livestock and poultry growth recovery capacity.

[0022] According to a first aspect of the present invention, a method for early prediction of livestock and poultry growth phenotypes is provided, comprising:

[0023] Daily body weight and daily feed intake data for each animal are continuously collected during the early stage of the monitoring period, wherein the early stage refers to the pre-preset proportion of the monitoring period.

[0024] The daily body weight data and the daily feed intake data are preprocessed and standardized to form the early daily weight gain sequence and early daily feed intake sequence for each animal;

[0025] The maximum correlation coefficient between the early daily weight gain sequence and the early daily feed intake sequence was calculated as an indicator of the early feed-growth coupling strength.

[0026] Based on the described early daily weight gain sequence for each animal, obtain the early average daily weight gain for each animal in the early stage;

[0027] Based on the early feed-growth coupling strength index and the early average daily weight gain, predict the growth phenotype category of each animal in the early stage.

[0028] Based on the above technical solution, the present invention can also be improved as follows.

[0029] Optionally, the daily body weight and daily feed intake data of each animal during the early stage of the continuous collection and monitoring period include:

[0030] The weight information of each animal is collected by an animal weight data acquisition device, which includes at least one of a weighbridge weighing system, a machine vision weight estimation system, a lidar volume measurement system, or a wearable sensor.

[0031] The intake information of each animal is collected by an animal feed intake acquisition device, which includes at least one of a radio frequency identification feeding station, a feed tower weighing system, or an image recognition feeding behavior analysis system.

[0032] Optionally, the daily food intake data is replaced by daily water intake data, wherein:

[0033] When precise daily feed intake data cannot be directly obtained, daily water intake sequences are collected as proxy variables for daily feed intake sequences.

[0034] The maximum correlation coefficient between the early daily weight gain sequence and the early daily water intake sequence was calculated as an indicator of the early feed-growth coupling strength.

[0035] Optionally, calculating the maximum correlation coefficient between the early daily weight gain sequence and the early daily feed intake sequence includes:

[0036] Z-score normalization was performed on the early daily weight gain sequence and the early daily feed intake sequence.

[0037] Within the preset range of sliding time delay values, select multiple time delay values. ;

[0038] Calculate at each time delay value The correlation coefficients of the early daily weight gain sequence and the early daily feed intake sequence are as follows ;

[0039] The maximum absolute value of all calculated correlation coefficients is taken as the maximum correlation coefficient. .

[0040] Optionally, based on the early feed-growth coupling strength index and the early average daily weight gain, the growth phenotype category of each animal in the early stage is predicted, including:

[0041] If an animal's early average daily weight gain is lower than the preset growth standard, but its maximum correlation coefficient is lower... If the value is greater than or equal to the first preset threshold, the animal is predicted to be a potential compensatory growth individual and it is recommended to retain it.

[0042] If the animal's early average daily weight gain is lower than the preset growth standard, and its maximum correlation coefficient is lower than the preset growth standard, then the animal's average daily weight gain is lower than the preset growth standard. If the value is less than or equal to the second preset threshold, the animal is predicted to be a potentially inefficient individual for weight gain, and it is recommended to cull it.

[0043] Otherwise, the animal is an individual to be observed;

[0044] Wherein, the first preset threshold is greater than the second preset threshold.

[0045] According to a second aspect of the present invention, a method for assessing the growth recovery capacity of livestock and poultry is provided, comprising:

[0046] Daily weight and daily feed intake data for each animal are continuously collected during the monitoring period to form a daily weight gain sequence and daily feed intake sequence for each animal.

[0047] The daily weight gain sequence and the daily feed intake data are preprocessed and standardized.

[0048] Based on the preprocessed daily weight gain sequence and daily feed intake sequence of each animal, the stage-specific growth characteristics, overall efficiency characteristics, and compensatory growth characteristics of each animal are extracted.

[0049] Based on the stage-specific growth characteristics, overall efficiency characteristics, and compensatory growth characteristics of each animal, the growth phenotype of each animal at the end of the monitoring period is determined.

[0050] Optionally, based on the preprocessed daily weight gain sequence and daily feed intake sequence of each animal, the stage-specific growth characteristics, overall efficiency characteristics, and compensatory growth characteristics of each animal are extracted, including:

[0051] The monitoring period was divided into early stage, middle stage and late stage. The average daily weight gain of each animal in the early stage and the average daily weight gain of each animal in the late stage were calculated respectively to obtain the early average daily weight gain (Early_ADG) and late average daily weight gain (Late_ADG) of each animal. The early average daily weight gain (Early_ADG) and late average daily weight gain (Late_ADG) constitute the stage growth characteristics.

[0052] Calculate the whole-period feed conversion ratio (Whole_FCR) for the monitoring period. The whole-period feed conversion ratio is calculated by dividing the cumulative daily feed intake by the cumulative daily weight gain for the entire monitoring period. The whole-period feed conversion ratio (Whole_FCR) is an overall efficiency characteristic.

[0053] Calculate the compensation gap, which is the difference between the standardized late-stage average daily weight gain and the early-stage average daily weight gain. The compensation gap is a compensatory growth characteristic.

[0054] Optionally, determining the growth phenotype of each animal at the end of the monitoring period based on the stage-specific growth characteristics, the overall efficiency characteristics, and the compensatory growth characteristics of each animal includes:

[0055] If the standardized Late_ADG, standardized Early_ADG, and standardized Whole_FCR satisfy the preset first rule combination, the animal is determined to be the optimal grower.

[0056] If the standardized Late_ADG, standardized Early_ADG, standardized Whole_FCR, and Gap satisfy the preset combination of the second rule, the animal is determined to be a compensatory grower.

[0057] If the standardized Late_ADG or standardized Whole_FCR satisfies the preset third rule combination, the animal is determined to be an inefficient weight gainer.

[0058] Other animals were determined to be in stable development.

[0059] Optionally, the first rule combination is: Late_ADG ≥ third preset threshold, Whole_FCR ≤ fourth preset threshold and Early_ADG > fifth preset threshold;

[0060] The second rule combination is: Late_ADG ≥ the sixth preset threshold, Early_ADG ≤ 0, Whole_FCR ≤ the eighth preset threshold and Compensation Gap ≥ the ninth preset threshold;

[0061] The third rule combination is: Late_ADG (late-stage average daily weight gain) ≤ the tenth preset threshold or Whole_FCR (whole-to-meat ratio) ≥ the eleventh preset threshold.

[0062] Optional, also includes:

[0063] The daily average growth trajectory of animal groups with different growth phenotypes was statistically tested on a daily basis.

[0064] Identify the starting point when a statistically significant difference is maintained for more than a preset number of days;

[0065] The preset time period before and after the starting time point is defined as the trajectory bifurcation window, which serves as the recommended intervention time.

[0066] This invention provides an early prediction method for livestock and poultry growth phenotypes and an assessment method for livestock and poultry growth recovery capacity. It involves continuously collecting early daily weight gain and early daily feed intake sequences for each animal during the early stages of a monitoring period; calculating the maximum correlation coefficient between the early daily weight gain and early daily feed intake sequences as an indicator of early feed-growth coupling strength; and predicting the growth phenotype category of each animal in the early stages based on the early feed-growth coupling strength indicator and the early average daily weight gain. This invention provides early warning of each animal's early growth type by analyzing feed-growth coupling strength rather than focusing solely on absolute growth rate. It can identify individuals with different recovery potentials before growth trajectory differentiation (such as in the early inhibition stage), thus enabling early intervention. Attached Figure Description

[0067] Figure 1 is a flowchart of an early prediction method for livestock and poultry growth phenotypes according to an embodiment of the present invention;

[0068] Figure 2 is a flowchart of cross-correlation analysis;

[0069] Figure 3 shows the flowcharts for the early prediction stage and the complete cycle classification stage.

[0070] Figure 4 is a flowchart of an embodiment of the present invention for early prediction of animal growth phenotype and prediction of complete life cycle;

[0071] Figure 5 is a flowchart of a method for assessing the growth recovery capacity of livestock and poultry according to an embodiment of the present invention;

[0072] Figure 6 is a schematic diagram of the structure of an early prediction system for livestock and poultry growth phenotypes provided in an embodiment of the present invention;

[0073] Figure 7 is a schematic diagram of a livestock and poultry growth recovery capacity assessment system according to an embodiment of the present invention. Detailed Implementation

[0074] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. In addition, the technical features of the various embodiments or individual embodiments provided by the present invention can be arbitrarily combined with each other to form feasible technical solutions. Such combinations are not constrained by the order of steps and / or structural composition patterns, but must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0075] Figure 1 illustrates an early prediction method for livestock and poultry growth phenotypes according to an embodiment of the present invention, comprising the following steps:

[0076] Step 1: Continuously collect daily weight data and daily feed intake data for each animal during the early stage of the monitoring period, wherein the early stage refers to the pre-preset proportion of the monitoring period.

[0077] Understandably, in the early stages of the monitoring period, such as the first third of the monitoring period, an automated data acquisition system is used to continuously record the daily weight and daily feed intake of each animal during the first third of the monitoring period.

[0078] The weight data may come from, but is not limited to: weighbridge systems (static or dynamic channel type), wearable sensors (such as collar-type or ear tag-type weight sensors), machine vision weight estimation systems (estimate weight through 2D / 3D image analysis), LiDAR volume measurement combined with weight conversion models, and any other device or method that can directly or indirectly obtain changes in animal weight.

[0079] Feed intake data may come from, but is not limited to: RFID-based feeding stations, feed tower weighing systems (group or individual level), image recognition feeding behavior analysis systems (estimate feed intake through video analysis), water consumption change estimation (indirect estimation using the correlation between feed and water), acoustic sensors (analyzing feeding behavior through chewing sounds), and any other device or method that can directly or indirectly reflect the animal's feed intake level.

[0080] In cases where accurate daily feed intake data cannot be directly obtained, daily water intake sequences are collected as proxy variables for the daily feed intake sequence.

[0081] Monitoring cycle: The cycle is set according to the animal species and growth stage, with a typical cycle of 30-60 days to cover the early adaptation period, transition period and stable growth period.

[0082] Step 2: Preprocess and standardize the daily body weight data and the daily feed intake data to form the early daily weight gain sequence and early daily feed intake sequence for each animal.

[0083] Understandably, the daily weight and feed intake data of each monitored animal in the early stages are preprocessed and standardized to form an early daily weight gain sequence for each animal. Based on the daily feed intake data of each animal in the early stages, an early daily feed intake sequence for each animal is formed.

[0084] Data preprocessing mainly includes quality control and standardization of the collected raw data, specifically including:

[0085] Outlier detection: Data that exceed the biologically reasonable range of daily weight gain (e.g., daily weight gain ADG < -500g / day or daily weight gain ADG > 3000g / day) are marked as outliers.

[0086] Outlier handling: A linear interpolation method is used to fill out outliers based on adjacent valid measurements.

[0087] Data integrity check: Exclude data for animal individuals with more than 10% missing data.

[0088] Z-score standardization: Standardizes all data, z i = (x i -μ) / σ makes different data comparable.

[0089] Step 3: Calculate the maximum correlation coefficient between the early daily weight gain sequence and the early daily feed intake sequence as an indicator of the early feed-growth coupling strength.

[0090] Understandably, referring to Figure 2, this embodiment of the invention introduces feed-growth coupling analysis to quantify the temporal coordination between daily feed intake and daily weight gain. Since daily feed intake (kg level) and daily weight gain (g level) are completely different in magnitude and physical unit, and individual baseline weight varies greatly, Z-score standardization is a prerequisite for performing temporal coupling calculations.

[0091] Z-score normalization was performed on the early daily weight gain and early daily feed intake sequences for each animal. Cross-correlation analysis was then performed on the early daily weight gain and early daily feed intake sequences for each animal.

[0092] Cross-correlation analysis is used to quantify the degree of linear correlation between two time series and to account for possible time delays. In this invention, the "extra feed eaten by animals today" does not immediately translate into "today's weight gain," but rather requires processes such as digestion, absorption, and metabolism, typically with a delay of 1-3 days.

[0093] Specifically, a range of time delay values ​​is set, and within this range, multiple time delay values ​​are selected according to a certain rule, such as selecting multiple time delay values ​​according to a fixed step size.

[0094] For each animal, the Pearson correlation coefficient between the standardized early daily feed intake sequence and the standardized early daily weight gain sequence was calculated at each time lag (τ = -7 to +7 days), and the cross correlation function (CCF) was constructed.

[0095] Here, the standardized early daily feed intake sequence is denoted as DFI = {dfi1, dfi2, ..., dfin}, and the standardized early daily weight gain sequence is denoted as ADG = {adg1, adg2, ..., adgn}.

[0096] For each time lag τ, calculate the Pearson correlation coefficient r(τ) between the early daily weight gain sequence and the early daily feed intake sequence;

[0097] ;

[0098] in, This represents the daily food intake on day t. τ represents the average value of the daily food intake sequence. Indicates the first The daily increase in weight This represents the average of the daily weight gain sequence. Let Variance be the variance of the daily food intake sequence. Let τ be the average of the daily weight gain sequence, τ = -7, -6, ..., 0, ..., +6, +7 days. τ > 0 indicates that feed intake leads weight gain, τ = 0 indicates that feed intake and weight gain are compared synchronously, and τ < 0 indicates that weight gain leads feed intake.

[0099] The maximum correlation coefficient, r_max = max{r(τ) |τ∈[-7, +7]}, reflects the strongest correlation between feed intake and weight gain. The value of r_max ranges from -1 to +1, with typical values ​​of 0.7-0.9 for healthy animals and 0.2-0.4 for animals with metabolic disorders. The maximum correlation coefficient between the calculated early daily weight gain sequence and the early daily feed intake sequence is used as an indicator of the early feed-growth coupling strength for subsequent early predictive analysis of animal growth phenotypes.

[0100] It should be noted that although this embodiment uses the Pearson correlation coefficient as an example for illustration, those skilled in the art will understand that other statistical indicators that can quantify the synchronicity of two time series, such as the Spearman rank correlation coefficient and the Kendall rank correlation coefficient, can be used to replace the Pearson coefficient to calculate the coupling strength described in this invention, and these substitutions all fall within the protection scope of this invention.

[0101] As an optional embodiment, the coupling analysis can be replaced by the following analysis method:

[0102] (1) Dynamic Time Warping (DTW) method: Calculate the DTW distance between the DFI and ADG time series to evaluate the nonlinear coupling relationship.

[0103] (2) Mutual information analysis: Calculate the mutual information between DFI and ADG to capture nonlinear dependencies.

[0104] (3) Granger causality test: Analyze the causal relationship and directionality between DFI and ADG;

[0105] (4) Wavelet coherence analysis: Analyze the coupling characteristics of DFI and ADG at different frequency scales.

[0106] (5) Phase space reconstruction method: The phase space is reconstructed by delayed embedding and the attractor characteristics are analyzed.

[0107] (6) Transfer entropy analysis: quantify the amount of information transferred between DFI and ADG and between ADG and DFI.

[0108] Step 4: Based on the early daily weight gain sequence of each animal, obtain the early average daily weight gain of each animal in the early stage.

[0109] Understandably, based on the early daily weight gain sequence of each animal, the early average daily weight gain (Early_ADG) is calculated for each animal: the average daily weight gain during the monitoring period from day 0 to day 14 (or the first 1 / 3 of the period).

[0110] Step 5: Based on the early feed-growth coupling strength index and the early average daily weight gain, predict the growth phenotype category of each animal in the early stage.

[0111] Understandably, the study found that in the early stages (first third of the cycle), the ADG of compensatory growers and inefficient gainers were almost the same, but the r_max differed significantly: r_max ≈ 0.74 in the early stage for compensatory growers and r_max ≈ 0.32 in the early stage for inefficient gainers. This difference makes r_max a key feature for early prediction.

[0112] Referring to Figures 3 and 4, in the early prediction phase, the growth type of each animal is predicted based on its early average daily weight gain (Early_ADG) and maximum correlation coefficient (r_max).

[0113] When the monitoring period has not reached the first preset time point (e.g., day 15, i.e., the first 1 / 3 of the cycle), the system enters the early warning mode, using only the collected early data for prediction. The classification results and suggestions for the early warning stage are shown in Table 1 below: Table 1 Classification results and suggestions for the early warning stage

[0114] Based on the conditional judgment results, it is recommended that if Early_ADG < preset growth threshold and r_max ≥ first preset threshold (e.g., 0.7), potential compensators should be retained for observation and nutritional optimization; if Early_ADG < preset growth threshold and r_max ≤ second preset threshold (e.g., 0.3), potential inefficient individuals should be closely monitored and eliminated. Other situations require further observation and monitoring. surface

[0115] It should be noted that the method of this invention differs from traditional complete-cycle assessment of animal growth phenotypes, providing an early prediction method for livestock and poultry growth phenotypes. Its core lies in: acquiring daily body weight and daily feed intake data only for the first preset proportion of the monitoring period (e.g., the first 1 / 3 of the period); calculating the feed-growth coupling strength index (e.g., the maximum correlation coefficient) within this early period; if this coupling strength index is higher than a preset threshold, even if the animal's current absolute growth rate is low, it is still predicted as an individual with high recovery potential. This method can achieve advanced identification of potential individuals before their growth trajectories actually differentiate.

[0116] Referring to Figure 5, this embodiment of the invention also provides a method for assessing the growth recovery capacity of livestock and poultry, including the following steps:

[0117] Step 1': Continuously collect daily weight data and daily feed intake data for each animal during the monitoring period to form a daily weight gain sequence and daily feed intake sequence for each animal.

[0118] Understandably, the difference between this step and step 1 above is that this step continuously collects the daily weight data and daily feed intake data of each animal throughout the entire monitoring period, and performs preprocessing and standardization to form the daily weight gain sequence and daily feed intake sequence of each animal throughout the entire monitoring period.

[0119] Step 2': Preprocess and standardize the daily weight gain sequence and the daily feed intake data.

[0120] Step 3': Based on the preprocessed daily weight gain sequence and daily feed intake sequence of each animal, extract the stage-specific growth characteristics, overall efficiency characteristics, and compensatory growth characteristics of each animal.

[0121] Understandably, based on the preprocessed daily weight gain and daily feed intake sequences, the stage-specific growth characteristics, overall efficiency characteristics, and compensatory growth characteristics of each animal are extracted. Specifically, these include:

[0122] (1) Stage-specific growth characteristics:

[0123] Early Average Daily Weight Gain (Early_ADG): The average daily weight gain during the monitoring period from day 0 to 14 (or the first 1 / 3 of the period);

[0124] Late Average Daily Weight Gain (Late_ADG): The average daily weight gain during the last two weeks (or the last third of the monitoring period).

[0125] (2) Overall efficiency characteristics:

[0126] Whole FCR: Cumulative feed intake divided by cumulative weight gain over the entire monitoring period.

[0127] (3) Compensatory growth characteristics:

[0128] Compensation Gap: The difference between the standard score of late ADG and the standard score of early ADG, i.e., Gap = z(Late_ADG) - z(Early_ADG), where z(.) represents standardization.

[0129] Step 4': Based on the stage-specific growth characteristics, overall efficiency characteristics, and compensatory growth characteristics of each animal, determine the growth phenotype of each animal at the end of the monitoring period.

[0130] As shown in Figures 3 and 4, during the complete cycle classification phase, the growth phenotype of each animal throughout the entire monitoring period is determined based on its Early Average Daily Gain (Early_ADG), Late Average Daily Gain (Late_ADG), Whole FCR, and Compensation Gap. The growth phenotype of each animal throughout the entire monitoring period includes four categories: optimal grower, compensatory grower, inefficient gainer, and stable grower.

[0131] It should be noted that the phenotypic classification rules used in this invention are not set out of thin air, but are formed through the following data-driven process, mainly including the following steps:

[0132] Step 1: Unsupervised clustering to discover natural clusters:

[0133] Unsupervised clustering analysis was performed on the multidimensional feature data (Early_ADG, Late_ADG, Whole_FCR, Gap, r_max, etc.) of the experimental population. This invention employs the consensus clustering method, which evaluates the stability of the clustering results and determines the optimal number of clusters through multiple resampling and clustering.

[0134] Experimental results show that the clustering results are most stable when K=4 (the consensus matrix heatmap clearly shows four blocks), indicating that the experimental population is naturally divided into four growth patterns.

[0135] Step 2: Biological interpretation of clustering results:

[0136] Analysis of the eigenvalues ​​of the four clusters revealed that they have clear biological meanings.

[0137] Cluster A: High early ADG, high late ADG, low FCR → "Optimal grower"

[0138] Cluster B: Low early ADG, high late ADG, large gap, high r_max → "Compensatory growers"

[0139] Cluster C: Low ADG or high FCR in late stages, low r_max → "Inefficient weight gainers"

[0140] Cluster D: All indicators are centered → "Stable developers"

[0141] Step 3: Extract classification rules from cluster boundaries:

[0142] Analyze the distribution boundaries of each cluster across all feature dimensions to extract actionable classification thresholds:

[0143] The boundary values ​​between the optimal grower and other classes are: z(Late_ADG) ≈ 0.5, z(FCR) ≈ -0.5

[0144] The characteristic combination of compensatory growers is: z(Early_ADG) < 0, Gap > 0.8, and r_max > 0.6.

[0145] The characteristic boundary for inefficient weight gainers is: z(Late_ADG) < -0.5 or z(FCR) > 0.8.

[0146] These thresholds are cross-validated to ensure classification accuracy and consistency on independent datasets.

[0147] Step 4: Rule Validation and Optimization

[0148] The extracted rules were applied to the independent validation set, and the classification consistency rate was > 90%.

[0149] Constructing phenotypic classification rules using consensus clustering has the following advantages:

[0150] Interpretability: Each rule has a clear biological meaning.

[0151] Reproducibility: Rules can be applied directly to new data without the need for re-clustering.

[0152] Portability: The threshold can be adjusted according to different conditions.

[0153] Transparency: The classification decision-making process is completely transparent, facilitating review and supervision.

[0154] Cluster analysis was used to classify the growth phenotypes of animals throughout their entire growth cycle into optimal growers, compensatory growers, inefficient gainers, and stable growers.

[0155] Based on the characteristic information extracted from each animal during the complete monitoring period, the growth phenotype of each animal is identified. Specifically, if the standardized Late_ADG, standardized Early_ADG, and standardized Whole_FCR satisfy a preset first rule combination, the animal is determined to be an optimal grower; if the standardized Late_ADG, standardized Early_ADG, standardized Whole_FCR, and Gap satisfy a preset second rule combination, the animal is determined to be a compensatory grower; if the standardized Late_ADG or standardized Whole_FCR satisfies a preset third rule combination, the animal is determined to be an inefficient gainer; other animals are determined to be stable growers.

[0156] The first rule combination is: Late_ADG ≥ third preset threshold, Whole_FCR ≤ fourth preset threshold and Early_ADG > fifth preset threshold.

[0157] The second rule combination is: Late_ADG ≥ the sixth preset threshold, Early_ADG ≤ 0, Whole_FCR ≤ the eighth preset threshold and Compensation Gap ≥ the ninth preset threshold.

[0158] The third rule combination is: Late_ADG (late-stage average daily weight gain) ≤ the tenth preset threshold or Whole_FCR (whole-to-meat ratio) ≥ the eleventh preset threshold.

[0159] The complete cycle classification results and recommendations are shown in Table 2 below.

[0160] Table 2. Complete Cycle Classification Results and Recommendations

[0161] Phenotypic Classification Criteria Biological Meaning Optimal Growers z(Late_ADG) ≥ Third Preset Threshold AND z(FCR) ≤ Fourth Preset Threshold AND z(Early_ADG) > Fifth Preset Threshold Highly efficient growth throughout, with optimal metabolic coordination Compensatory Growers z(Late_ADG) ≥ Sixth Preset Threshold AND z(Early_ADG) ≤ 0 AND z(FCR) ≤ Eighth Preset Threshold AND Gap ≥ Ninth Preset Threshold Successfully recovered after early inhibition, with metabolic elasticity Inefficient Weight Gainers z(Late_ADG) ≤ Tenth Preset Threshold OR z(FCR) ≥ Eleventh Preset Threshold Continuously poor growth or low feed conversion efficiency Stable Development Individuals Do not meet the above three categories Stable moderate growth performance, representing the main body of the population surface

[0162] In this embodiment of the invention, the third preset threshold is 0.7, the fourth preset threshold is -0.05, the fifth preset threshold is -0.5, the sixth preset threshold is 0.3, the eighth preset threshold is 0.5, the ninth preset threshold is 0.5, the tenth preset threshold is -0.4, and the eleventh preset threshold is 0.4.

[0163] It should be noted that, as an optional embodiment, animal growth phenotypes can be classified based on machine learning classifiers, fuzzy classification methods, or adaptive thresholding methods. Machine learning classifiers can employ support vector machines (SVM), random forests, or neural networks to classify phenotypes, potentially improving the classification accuracy of boundary samples. Fuzzy classification methods utilize fuzzy set theory to calculate the membership degree of each animal to each phenotype, reflecting the continuity of the phenotype. Adaptive thresholding methods automatically adjust the classification threshold based on the distribution of population data, adapting to differences between different batches or breeds.

[0164] After identifying and predicting the growth phenotype of each animal during the complete monitoring period, this invention introduces trajectory bifurcation analysis to identify the time window in which the growth trajectories of different phenotypes begin to significantly differentiate, and determines this time window as the optimal intervention time.

[0165] In practice, the daily average growth trajectory of animal groups with different growth phenotypes is statistically tested on a daily basis; the starting point when the statistically significant difference is maintained for more than a preset number of days is identified; and the preset time period before and after the starting point is determined as the trajectory bifurcation window, which is used as the recommended intervention time.

[0166] In summary, the method of this invention consists of two logical stages:

[0167] Training / Validation Phase (Ground Truth Establishment): Using all data from the complete monitoring period (e.g., days 0-42), including indicators that require a full period to calculate such as late ADG, whole-cycle FCR, and compensation gap, retrospectively classify the animals' phenotypic characteristics. The purpose of this phase is to establish a Ground Truth, analyze the differences in feed-growth coupling characteristics of different phenotypes in the early stages, and establish the association between early characteristics and the final phenotype.

[0168] Application / Prediction Phase (Early Warning): In actual production applications, only the data from the first 1 / 3 of the monitoring cycle (or other preset proportions) are used to calculate the early feed-growth coupling strength (r_max). Combined with the current ADG level, the future growth phenotype of the individual (such as "potential compensator" or "potential inefficient one") is predicted, thereby making intervention decisions before the growth trajectory actually differentiates.

[0169] Key Logic: Indicators such as Late_ADG and Whole_FCR used in the training phase can only be calculated at the end of the monitoring period and cannot be used for early prediction. However, analysis of the training phase revealed a strong correlation between early r_max and the final phenotype: early r_max ≈ 0.736 for those with compensatory growth and early r_max ≈ 0.320 for those with inefficient weight gain. Therefore, in the application phase, prediction can be made based solely on early r_max.

[0170] Finally, it should be noted that the technical solution proposed in this invention is not limited to pig farming, but is also applicable to: growth performance evaluation of poultry farming (broilers, laying hens, ducks, geese, etc.); growth or lactation performance evaluation of beef cattle and dairy cattle; growth evaluation of other economic animals such as sheep and rabbits; and coupling analysis of feed amount and growth in aquaculture (fish, shrimp, etc.).

[0171] The following section uses typical experimental data to illustrate the early prediction method for livestock and poultry growth phenotypes and the assessment method for growth capacity recovery provided by this invention. Table 3 below contains data from an actual study of 309 growing pigs monitored for 43 days, which can serve as data support for the embodiments of this invention.

[0172] Table 3. Results of actual research on livestock and poultry monitoring

[0173] Phenotypic Count (%) Early ADG Late ADGFCR Early r_max Optimal Growth 72 (23.3%) 657 g / d 916 g / d 2.32 40.812 Stable Growth 139 (45.0%) 638 g / d 861 g / d 2.49 50.654 Compensatory Growth 35 (11.3%) 634 g / d 895 g / d 2.49 10.736 Inefficient Weight Gain 63 (20.4%) 642 g / d 757 g / d 2.76 70.320 surface

[0174] The data in Table 3 reveals the following:

[0175] (1) Predictive value of early coupling differences: Compensatory growers and inefficient gainers had almost the same early ADG (634 vs 642 g / d, p=0.31), but the feed-growth coupling strength was significantly different (r=0.736 vs 0.320), proving that coupling indicators are more predictive than absolute growth rate.

[0176] (2) Economic significance: The FCR of compensatory growers (2.491) is about 10% lower than that of inefficient gainers (2.767), which means that about 0.28 kg of feed can be saved for every 1 kg of meat produced.

[0177] Figure 6 illustrates an early prediction system for livestock and poultry growth phenotypes according to an embodiment of the present invention. The system includes:

[0178] The first data acquisition module 601 is used to continuously acquire the daily weight data and daily feed intake data of each animal in the early stage of the monitoring period, wherein the early stage refers to the pre-preset proportion period within the monitoring period.

[0179] The first data preprocessing module 602 is used to preprocess and standardize the daily weight data and the daily feed intake data to form an early daily weight gain sequence and an early daily feed intake sequence for each animal.

[0180] The coupling analysis module 603 is used to calculate the maximum correlation coefficient between the early daily weight gain sequence and the early daily feed intake sequence, as an indicator of the early feed-growth coupling strength.

[0181] The data acquisition module 604 acquires the early average daily weight gain of each animal in the early stage based on the early daily weight gain sequence of each animal.

[0182] The early prediction module 605 is used to predict the growth phenotype category of each animal in the early stage based on the early feed-growth coupling strength index and the early average daily weight gain.

[0183] It is understood that the early prediction system for livestock and poultry growth phenotypes provided by the present invention corresponds to the early prediction methods for livestock and poultry growth phenotypes provided in the foregoing embodiments. The relevant technical features of the early prediction system for livestock and poultry growth phenotypes can be referred to the relevant technical features of the early prediction methods for livestock and poultry growth phenotypes, and will not be repeated here.

[0184] Referring to Figure 7, a livestock and poultry growth recovery capacity assessment system provided in an embodiment of the present invention is shown. The system includes:

[0185] The second data acquisition module 701 is used to continuously collect the daily weight data and daily feed intake data of each animal during the monitoring period, forming the daily weight gain sequence and daily feed intake sequence of each animal.

[0186] The second data preprocessing module 702 is used to preprocess and standardize the daily weight gain sequence and the daily feed intake data;

[0187] The feature extraction module 703 is used to extract the stage growth characteristics, overall efficiency characteristics and compensatory growth characteristics of each animal based on the preprocessed daily weight gain sequence and daily feed intake sequence of each animal.

[0188] Phenotyping module 704 is used to determine the growth phenotype of each animal at the end of the monitoring period based on the stage growth characteristics, the overall efficiency characteristics and the compensatory growth characteristics of each animal.

[0189] It is understood that the livestock and poultry growth recovery capacity assessment system provided by the present invention corresponds to the livestock and poultry growth recovery capacity assessment method provided in the foregoing embodiments. The relevant technical features of the livestock and poultry growth recovery capacity assessment system can be referred to the relevant technical features of the livestock and poultry growth recovery capacity assessment method, and will not be repeated here.

[0190] The present invention provides an early prediction method for livestock and poultry growth phenotypes and a method for assessing livestock and poultry growth recovery capacity, which have the following advantages:

[0191] (1) Significantly enhanced early warning capability: This invention, by analyzing feed-growth coupling strength rather than focusing solely on absolute growth rate, can identify individuals with different recovery potentials before growth trajectory differentiation (such as in the early inhibition stage). Studies have shown that differences in early coupling strength can predict subsequent recovery outcomes, providing a crucial time window advantage for early intervention.

[0192] (2) The classification results are interpretable and reproducible: The rule-based classification method is adopted, and all classification criteria have clear biological meanings (such as 'compensation gap ≥ 0.5 standard deviation' represents substantial growth acceleration). The classification results are reproducible in different batches and different scenarios, which facilitates cross-field comparison and industry promotion.

[0193] (3) Significant economic benefits: Through early identification and timely intervention, the ineffective consumption of feed resources by inefficient individuals can be reduced (studies show that the feed conversion ratio of inefficient gainers is about 10% higher than that of compensatory growers), feed conversion efficiency can be improved, breeding costs can be reduced, and the economic benefits of enterprises can be enhanced.

[0194] (4) Animal welfare improvement: Early detection of individuals with abnormal growth and timely implementation of health management measures can help improve animal welfare and meet the requirements of sustainable development of modern animal husbandry.

[0195] (5) High system integration: The system provided by this invention integrates a complete functional chain from data acquisition, analysis and processing to decision support, and can be seamlessly connected with existing precision animal husbandry equipment, which is convenient for actual production application.

[0196] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0197] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0198] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more blocks of the flowchart illustrations and / or one or more blocks of the block diagrams.

[0199] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0200] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

[0201] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0202] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for early prediction of livestock and poultry growth phenotypes, characterized in that, include: Daily body weight and daily feed intake data for each animal are continuously collected during the early stage of the monitoring period, where the early stage refers to the pre-preset proportion of the monitoring period. The daily body weight and daily feed intake data are preprocessed and standardized to form an early daily weight gain sequence and an early daily feed intake sequence for each animal. The maximum correlation coefficient between the early daily weight gain sequence and the early daily feed intake sequence is calculated as an early feed-growth coupling strength index. Based on the early daily weight gain sequence of each animal, the early average daily weight gain of each animal in the early stage is obtained. Based on the early feed-growth coupling strength index and the early average daily weight gain, the growth phenotype category of each animal in the early stage is predicted.

2. The method for early prediction of livestock and poultry growth phenotypes according to claim 1, characterized in that, The continuous collection and monitoring of daily weight and daily feed intake data for each animal in the early stage of the monitoring period includes: collecting weight information for each animal through an animal weight data acquisition device, which includes at least one of a weighbridge system, a machine vision weight estimation system, a lidar volume measurement system, or a wearable sensor; and collecting feed intake information for each animal through an animal feed intake acquisition device, which includes at least one of a radio frequency identification feeding station, a feed tower weighing system, or an image recognition feeding behavior analysis system.

3. The method for early prediction of livestock and poultry growth phenotypes according to claim 1, characterized in that, The daily feed intake data is replaced by daily water intake data, wherein: when accurate daily feed intake data cannot be obtained directly, daily water intake sequence is collected as a proxy variable for the daily feed intake sequence; the maximum correlation coefficient between the early daily weight gain sequence and the early daily water intake sequence is calculated as an indicator of the early feed-growth coupling strength.

4. The method for early prediction of livestock and poultry growth phenotypes according to claim 1, characterized in that, The calculation of the maximum correlation coefficient between the early daily weight gain sequence and the early daily feed intake sequence includes: performing Z-score standardization on the early daily weight gain sequence and the early daily feed intake sequence; and selecting multiple time lag values ​​within a preset range of sliding time lag values. ; Calculate at each time delay value The correlation coefficients of the early daily weight gain sequence and the early daily feed intake sequence are as follows Extract the maximum absolute value of all calculated correlation coefficients as the maximum correlation coefficient. 。 5. The method for early prediction of livestock and poultry growth phenotypes according to claim 1, characterized in that, Based on the early feed-growth coupling strength index and the early average daily weight gain, predict the growth phenotype category of each animal in the early stage, including: if the animal's early average daily weight gain is lower than a preset growth standard threshold, but its maximum correlation coefficient is higher... If the animal's average daily weight gain in the early stages is greater than or equal to the first preset threshold, it is predicted to be a potential compensatory growth individual and it is recommended to retain it; if the animal's average daily weight gain in the early stages is lower than the preset growth standard threshold, and its maximum correlation coefficient is greater than or equal to the first preset threshold, it is predicted to be a potential compensatory growth individual and it is recommended to retain it. If the value is less than or equal to the second preset threshold, the animal is predicted to be a potentially inefficient weight gain individual and is recommended to be culled; otherwise, the animal is an individual to be observed; wherein, the first preset threshold is greater than the second preset threshold.

6. A method for assessing the growth recovery capacity of livestock and poultry, characterized in that, include: Daily body weight and daily feed intake data for each animal are continuously collected during the monitoring period to form daily weight gain and daily feed intake sequences for each animal. The daily weight gain and daily feed intake sequences are preprocessed and standardized. Based on the preprocessed daily weight gain and daily feed intake sequences for each animal, the stage-specific growth characteristics, overall efficiency characteristics, and compensatory growth characteristics of each animal are extracted. Based on the stage-specific growth characteristics, overall efficiency characteristics, and compensatory growth characteristics of each animal, the growth phenotype of each animal at the end of the monitoring period is determined.

7. The method for assessing the growth recovery capacity of livestock and poultry according to claim 6, characterized in that, Based on the preprocessed daily weight gain sequence and daily feed intake sequence for each animal, the stage-specific growth characteristics, overall efficiency characteristics, and compensatory growth characteristics of each animal are extracted, including: dividing the monitoring period into an early stage, a middle stage, and a late stage; calculating the average daily weight gain for each animal in the early stage and the average daily weight gain for each animal in the late stage, respectively, to obtain the Early Average Daily Weight Gain (Early_ADG) and Late Average Daily Weight Gain (Late_ADG) for each animal, wherein the Early Average Daily Weight Gain (Early_ADG) and the Late Average Daily Weight Gain (Late_ADG) constitute the stage-specific growth characteristics; calculating the Whole FCR for the entire monitoring period, wherein the Whole FCR is calculated by dividing the cumulative daily feed intake by the cumulative daily weight gain for the entire monitoring period, wherein the Whole FCR is the overall efficiency characteristic; and calculating the compensation gap (Gap), wherein the compensation gap is the difference between the standardized Late Average Daily Weight Gain and the Early Average Daily Weight Gain, wherein the compensation gap is the compensatory growth characteristic.

8. The method for assessing the growth recovery capacity of livestock and poultry according to claim 7, characterized in that, The process of determining the growth phenotype of each animal at the end of the monitoring period based on its stage-specific growth characteristics, overall efficiency characteristics, and compensatory growth characteristics includes: if the standardized Late_ADG, standardized Early_ADG, and standardized Whole_FCR satisfy a preset first rule combination, the animal is identified as the optimal grower; if the standardized Late_ADG, standardized Early_ADG, standardized Whole_FCR, and Gap satisfy a preset second rule combination, the animal is identified as the compensatory grower; if the standardized Late_ADG or standardized Whole_FCR satisfies a preset third rule combination, the animal is identified as the inefficient gainer; other animals are identified as those with stable development.

9. The method for assessing the growth recovery capacity of livestock and poultry according to claim 8, characterized in that, The first rule combination is: Late_ADG ≥ the third preset threshold, Whole_FCR ≤ the fourth preset threshold, and Early_ADG > the fifth preset threshold; the second rule combination is: Late_ADG ≥ the sixth preset threshold, Early_ADG ≤ 0, Whole_FCR ≤ the eighth preset threshold, and Compensation Gap ≥ the ninth preset threshold; the third rule combination is: Late_ADG ≤ the tenth preset threshold or Whole_FCR ≥ the eleventh preset threshold.

10. The method for assessing the growth recovery capacity of livestock and poultry according to claim 6, characterized in that, Also includes: The daily average growth trajectory of animal groups with different growth phenotypes was statistically tested on a daily basis; the starting point when the statistically significant difference was maintained for more than a preset number of days was identified. The preset time period before and after the starting time point is defined as the trajectory bifurcation window, which serves as the recommended intervention time.

Images