Bootstrapped neural network models for analyzing the responses of broiler chicks to dietary protein and branched chain amino acids

Abstract

, J. 2014. Bootstrapped neural network models for analyzing the responses of broiler chicks to dietary protein and branched chain amino acids. Can. J. Anim. Sci. 94: 79Á85. Reliable prediction of avian responses to dietary nutrients is essential for planning, management, and optimization activities in poultry nutrition. In this study, two bootstrapped neural network (BNN) models, each containing 100 separated neural networks (SNN), were developed for predicting average daily gain (ADG) and feed efficiency (FE) of broiler chicks in response to intake of protein and branched chain amino acids (BCAA) in the starter period. Using a re-sampling method, 100 different batches of data were generated for both the ADG and FE sets. Starting with 270 data lines extracted from eight studies in the literature, SNN models were trained, tested, and validated with 136, 67, and 67 data lines, respectively. All 200 SNN models developed, along with their respective BNN ones, were subjected to optimization (to find the optimum dietary protein and BCAA levels that maximize ADG and FE). Statistical analysis indicated that based on R 2 , the BNN models were more accurate in 76 and 56 cases (out of 100) compared with the SNN models developed for ADG and FE, respectively. Optimization of the BNN models showed protein, isoleucine, leucine, and valine requirements for maximum ADG were 231.80, 9.05, 14.03 and 10.90 g kg(1 of diet, respectively. Also, maximum FE was obtained when the diet contained 232.30, 9.07, 14.50, and 11.04 g kg(1 of protein, isoleucine, leucine, and valine, respectively. The results of this study suggest that in meta-analytic modelling, bootstrap re-sampling algorithms should be used to better analyze available data and thereby take full advantage of them. This issue is of importance in the animal sciences as producing reliable data is both expensive and time-consuming.Key words: Branched chain amino acids, broiler responses, neural networks, bootstrapping Faridi, A., Golian, A., Heravi Mousavi, A. et France, J. 2014. Mode`les de re´seaux neuronaux a`ve´rifications de type « bootstrap » pour l'analyse des re´ponses aux prote´ines alimentaires et aux acides amine´s a`chaıˆnes ramifie´es chez les jeunes poulets a`griller. Can. J. Anim. Sci. 94: 79Á85. La pre´diction fiable des re´ponses aviaires aux e´le´ments nutritifs est essentielle pour la planification, la gestion et l'optimisation des activite´s nutritionnelles de la volaille. Dans cette e´tude, deux mode`les de re´seaux neuronaux a`ve´rifications de type « bootstrap » (BNN Á « bootstrapped neural network »), chacun contenant 100 re´seaux neuronaux se´pare´s (SNN Á « separated neural networks »), ont e´te´de´veloppe´s pour la pre´diction des gains moyens quotidiens (ADG Á « average daily gain ») et de l'efficience alimentaire (FE Á « feed efficiency ») chez les jeunes poulets a`griller en re´ponse a`l'ingestion de prote´ines et d'acides amine´s à chaıˆnes ramifie´es (BCAA Á « branched chain amino acids ») dans la pe´riode de de´but. En utilisant une me´thode de re´...

Broiler responses to digestible threonine at different ages: a neural networks approach

“…Results show that the highest optimal value of DThr was that needed to minimize BL in all three age phases, and the optimal DThr levels for minimizing FCR were higher than those for maximizing ADG. These results are consistent with those reported earlier for Thr (Lugo et al., ; Kidd, ; Ahmadi and Golian, ), lysine (Leclercq, ; Faridi et al., ) and valine (Leclercq, ; Faridi et al., ).…”

“…The ability of this algorithm to define optimum AA levels has been demonstrated previously (e.g. Ahmadi and Golian, ; Faridi et al., ). In a recent study, 16 profiles of lysine dose–response data were taken from the literature then analysed using the random search method (Faridi et al., ).…”

“…). Despite its importance, there has to our knowledge been only one previous study conducted using the bootstrapping method to develop NN models in poultry nutrition (Faridi et al., ). statistica neural networks software version 10.0 was used to resample the data, construct and train the NN models, and ensemble them (StatSoft ).…”

“…In recent years, neural networks (NNs) have been employed in poultry nutrition although they are still considered a fairly new modelling tool. Neural network modelling has been a subject of interest in predicting the response of broilers to dietary nutrients, especially AAs (Faridi et al., , ). Recently, Ahmadi and Golian () composited the literature data on broiler response to Thr and analysed these data using NN models.…”

High leucine levels affecting valine and isoleucine recommendations in low-protein diets for broiler chickens

“…As per the fitted models for BW gain and gain:feed, the optimization response can be achieved by using dietary SID Leu, Val, and Ile levels of 1.33–1.37%, 0.94–0.96%, and 0.84–0.87% during the starter period; 1.23%, 0.82–0.83%, and 0.75% during the grower period; and 1.15–1.16%, 0.77%, and 0.68–0.70% during the finisher phase, respectively. Using the predicted levels, the optimal average ratios for Leu:Val:Ile in low-protein diets were estimated at 100:69:62 from 1 to 14 D, 100:67:61 from 14 to 28, and 100:67:60 from 28 to 42 D. In the literature, the estimated Val and Ile to Leu ratios ranged from 67 to 78 and 58 to 69, respectively, as per the nutritional recommendations proposed by researchers who studied the combined BCAA requirements ( Farran and Thomas, 1990 ; Faridi et al., 2014 ; Sedghi et al., 2015 ).…”