Optimal growth of broiler chickens and effective health monitoring are the main challenges in the livestock industry. Deep learning technologies, such as Differential Recurrent Neural Networks (DRNN), Long Short-Term Memory (LSTM), and Improved Feature Fusion Single Shot MultiBox Detector (IFSSD), offer innovative solutions for predicting chicken growth and detecting health conditions in real-time. This research aims to review the application of deep learning in growth forecasting and health monitoring of broiler chickens through systematic analysis using the PICO and PRISMA methods. Of the 150 studies identified, 30 met inclusion criteria for further analysis. The results show that DRNN and LSTM achieve high accuracy in processing temporal data, with an average growth prediction error of 1.8%. IFSSD excels in detecting chicken health via digital images, achieving 99.7% accuracy. Although the results are promising, challenges such as the need for high computing resources and the quality of data required for model training are major obstacles. This research suggests the need for investment in technological infrastructure, farmer training, and development of more efficient models to support the application of deep learning in broiler chicken farming. This technology has great potential to significantly improve production efficiency and animal welfare.