Modelling the growth performance and thermal environment of broiler chicken houses via different machine learning algorithms assisted by a customized Internet of Things

The growing global demand for broiler meat has underscored the importance of precision livestock farming to increase productivity, animal welfare, and sustainability. This study integrated Internet of Things (IoT) sensor networks with the extreme gradient boosting (XGBoost) algorithm to model how environmental factors—particularly temperature and humidity—affect broiler chickens’ performance indicators, including feed intake, body weight, and the feed conversion ratio (FCR). A total of 160 unsexed MB-Lohmann broiler chickens were raised under controlled conditions via a plant-based diet free of antibiotic growth promoters. Real-time data were collected via custom-built IoT sensor nodes, which generated over 19,000 environmental readings across a 35-day trial.Weekly growth performance measurements were collected and analysed via both statistical and machine learning approaches. Feed intake and weight gain were strongly positively correlated (r = 0.89), whereas FCR was strongly negatively correlated with feed efficiency (r = –0.95). Humidity was moderately associated with reduced feed efficiency (r = –0.82), suggesting possible environmental stress. Among the machine learning models tested, the multilayer perceptron (MLP) demonstrated the most consistent and accurate performance in predicting growth outcomes. In summary, this work highlights the potential of combining IoT technology with advanced analytics to support real-time decision-making in broiler production. By enabling more responsive and informed farm management, such approaches could play a vital role in improving efficiency, reducing mortality, and supporting sustainable poultry systems.