Answers are grounded in this paper and cite their sources.
Poultry diseases and welfare issues can lead to significant production losses, compromise food safety, and create a need for ongoing farm monitoring. This narrative review summarises 81 peer-reviewed research papers published between 2016 and 2026 focusing on the use of artificial intelligence (AI) for monitoring poultry diseases and health. The papers were grouped into the following categories, namely disease diagnosis, health and welfare monitoring, mortality and biosecurity surveillance, and post-harvest food safety, and are summarised in a study-level appendix by modality, technique and reported data details. Conventional machine learning, convolutional neural networks, transfer learning, YOLO-based object detection, transformer models, explainable AI, multimodal fusion, IoT, edge AI, federated learning, generative AI, and vision-language models are the focus of the reviewed literature. The most common methods are image and faecal-image analysis, complemented by audio, thermal, video, behavioural, and environmental sensing. The trend in recent work is to use lightweight models on local devices and to integrate multiple modalities. Some limitations to direct comparison of reported results include small sample sizes and the nature of the datasets in some studies, differences in class definitions, unbalanced classes, inconsistent validation protocols, and limited cross-farm testing. Cross-farm generalisation, standardised benchmarking, explainability, prospective veterinary validation, privacy-preserving collaboration features, and cost-effective edge deployment are highlighted as key priorities. Future work should involve creating multi-country and multi-breed benchmark datasets, developing commercially validated decision-support tools and poultry-specific foundation models, and implementing continual-learning systems and multimodal fusion frameworks. The review suggests that AI has the potential to enhance precision poultry production, but successful long-term implementation of AI in the field requires not only high accuracy but also robustness, interpretability, and affordability.
No takes yet. Share an insight, caveat, or question.
S et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: