A Narrative Review of Early Pregnancy Diagnosis Technologies for Livestock: From Conventional to Intelligent Systems.

Accurate and efficient early pregnancy diagnosis (EPD) in livestock is crucial for optimizing breeding management and enhancing productivity in modern animal husbandry. Over the past century, EPD technology has evolved from empirical methods to sophisticated techniques, encompassing biochemical marker detection, ultrasonic imaging, and further extending to emerging non-invasive approaches such as infrared thermography (IRT) and spectroscopic analysis. These advancements have not only improved diagnostic accuracy but also broadened the research scope to include small livestock and multiple species. This review critically examines the historical evolution, current methodologies, and applications of EPD technology, with a focus on analyzing the advantages and limitations of both traditional and emerging techniques. Additionally, it explores the potential of multimodal fusion strategies and artificial intelligence (AI) in EPD. At present, machine vision, wearable monitoring, and several AI applications remain prospective approaches rather than validated tools for routine EPD. The conclusion highlights that, despite significant progress, current technologies still face limitations in achieving in situ, non-contact, and high-throughput detection. Looking ahead, the integration of cutting-edge technologies, such as AI, small wearable sensors, and physiological time-series data analysis, holds promise for overcoming these bottlenecks, enabling more intelligent and efficient pregnancy diagnosis, and providing scientific support for modern animal husbandry.