Feed Restriction Impairs Immunity and Lymphoid Organ Development in Broiler Chicks: Enabling Accurate Body Weight Prediction Using Machine Learning

Abstract

This research assessed the effect of feed limitation on immune response, growth rate, and body weight prediction in poultry. For 56 days, 80 one-day-old chicks were reared and were distributed into four groups: Groups A–C got 6, 12, and 24 h of daily feed constraint (days 21–35), respectively, while Group D was the control group which were fed ad libitum. Feed deprivation resulted in  considerably high impaired hematological factors (p<0.01), such as, minimized Newcastle Disease virus (NDV) antibody titers, higher heterophil-to-lymphocyte ratios, reduced overall leukocyte count in overly stressed groups as compared to the control group. In restricted groups, the lymphoid organs (spleen, bursa of Fabricius and thymus) displayed considerably poor weights, and the mortality rate was increased (P<0.05), exclusively after the NDV trial. From physiological readings, augmented dataset was derived that was used in developing machine learning models to predict body weight. Ensemble models (Gradient Boosting, XGBoost, , AdaBoost) and  Ridge Regression accomplished exceptional predictive performance (R² >0.98), while Deep Neural Networks and Lasso underachieved due to over-regularization and overfitting, respectively. Limited feed greatly effects the growth nd immunity of poultry; ensemble machine learning models are effective in  predicting the body weight changes caused by feed deprivation stress, offering practical tools for monitoring poultry production.