AgroAnimal Health Prediction Framework Using Environmental and Physiological Monitoring Data

Abstract

Livestock productivity and welfare are becoming more vulnerable to physiological stress and disease onset due to the inability of visual inspection to detect them until they result in significant losses to the economy and welfare of the livestock. This research develops the AgroAnimal Health Prediction Framework (AAHPF), which is an environment that incorporates environmental sensor streams, including ambient temperature, relative humidity, and air quality measurements, along with physiological monitoring streams, including body temperature, heart rate, respiration rate, and rumination, for health risk prediction of the livestock animals. The proposed AAHPF comprises four functional layers: sensing and acquisition layer, pre-processing and feature engineering layer, predictive analytics layer based on ensemble learning models and sequence-aware learning models, and decision support layer which issues multi-tier alerts to farm owners. Unlike previous approaches, AAHPF models the relationship between environmental and physiological signals, in particular, how the effect of heat and humidity loading on the animals results in increased respiration and reduced rumination. The proposed architecture also incorporates a feature schema which is agnostic to the species, making the system applicable to dairy cattle, poultry, and small ruminants without requiring much modification. Comparison to baseline single-source approaches shows that the combination of environmental and physiological modeling results in increased early warning sensitivity and decreased false positives when compared to either the physiology or environment alone. Architectural decisions are considered alongside data management aspects, including the continuous animal monitoring data required for the implementation of such a system on smaller farms. The proposed framework seeks to help the veterinarians and farmers move towards a preventative approach to animal care.