|
Artificial intelligence in large animal respiratory medicine: Current applications, clinical challenges, and future perspectives
|
S. Sadeghian Chaleshtori   , A. Avarseji * |
| Department of Internal Medicine, Faculty of Veterinary Medicine, University of Tehran, Tehran,Iran & * |
|
|
Abstract: (5 Views) |
| Respiratory diseases represent significant causes of morbidity, decreased productivity, poor athletic performance, antimicrobial consumption, and economic losses in large livestock. Traditional diagnostics involve intermittent clinical examination, respiratory scoring, imaging, laboratory testing, and invasive sampling procedures that can overlook subclinical and transient manifestations of these diseases. AI, machine learning (ML), deep learning (DL), computer vision, bioacoustics, wearable monitoring, infrared thermography, and precision-livestock technologies create novel possibilities to assess health of lungs continuously and objectively. This review provides an overview of recent studies on the use of AI for detecting, monitoring, predicting and assessing clinically relevant respiratory diseases in large animals with a focus on cattle and horses. In cattle, the most advanced research on the application of AI deals with the detection and monitoring of bovine respiratory disease (BRD) using automated feeding systems, accelerometers, positioning technologies, cough monitoring, computer vision, thermal imaging, and multimodal ML. For horses, emerging applications include analysis of videos recorded by smartphones for diagnosing equine asthma, assisted upper-airway endoscopy with AI, analysis of breathing sounds via deep learning during exercise, and wearable and non-invasive respiratory monitoring. There is limited research related to buffalo, camel, sheep, and goat diseases using enabling technologies. In general, there is a shift from one-sensor classification towards multimodal, longitudinal, and individual respiratory phenotyping in all species. Barriers to progress in this area include small training datasets, varying definitions of the conditions under consideration, imbalanced classes, imperfect labels, lack of external validation, domain shift, missing sensor data, model calibration, explainability, high costs, and implementation into clinical practice. Therefore, AI should be considered as a technology of decision support and not as a replacement of expert opinion. Further efforts need to focus on multicenter datasets, consistent reference diagnoses, multimodal fusion, prospective external validation, explainability and uncertainty-aware models, federated and transfer learning, edge computing, and longitudinal individual predictions. |
|
| Keywords: artificial intelligence, respiratory disease, bovine respiratory disease, equine asthma, machine learning, computer vision, bioacoustics, wearable sensors, precision livestock farming, veterinary medicine |
|
|
Full-Text [PDF 961 kb]
(19 Downloads)
|
Type of Study: Research |
Subject:
General
Received: 2026/09/5
|
|
|
|
|
|
|
| Send email to the article author |
|
|
|
| Add your comments about this article |
|
|
|
|
|
|
