IoT Enabled Dairy Herd Health Monitoring: Machine Learning Based Analysis of Productivity, Estrus Detection and Transport Stress

Abstract

The research paper reflects an IoT-based smart neck belt that has a combination of physiological sensors to measure such parameters like Body temperature, Heart rate, Respiration, Rumination and Activity levels to monitor the health of dairy herds and provides data of the environmental conditions through the Ambient temperature, Humidity, Air Quality Index (AQI) and Noise levels. The effects of the environmental parameters on the health of dairy herd were determined under the Controlled (Batch-1), natural condition (Batch-2) and transport conditions. Its results indicate that there is a significant change in the dairy herd vitals with a decrease in rumination and feed conversion efficiency, and subsequent increase in body temperature, heart rate and stress index for conditions of Batch-2 and during inconvenient transportation. Comparative results of Batch-1 with Batch-2 showed that the welfare of the animals improved significantly, such as the reduction of stress levels by 20%, average milk yield improved by 65%, estrus detection accuracy increased to 69%, faster recovery and higher conception rates. In order to enhance the predictive capabilities, several machine learning (ML) algorithms were tested in relation to stress detection, disease prediction and estrus detection. The analyzed data indicated that more complex models like Hist Gradient Boosting and 1D Convolutional Neural Networks indicated more accurate outcomes compared to the traditional ML models. These results confirm that the combination of IoT-based sensing with ML based analytics can transform conventional dairy herd management into more sustainable, welfare based and productivity focused.