This proposal describes a non-invasive breath-acetone sensing system for early detection of subclinical ketosis (hyperketonemia) in dairy cows, integrated as a physiological module inside a farm-scale Digital Twin and connected to existing farm management systems and milking infrastructure. The approach directly fits the Innocrowd challenge requirement for continuous, objective, predictive monitoring, automated alerts, actionable recommendations, and practical pilot deployment followed by scale.
The scientific basis is that acetone is a ketone body measurable in exhaled breath and rises with metabolic ketone dynamics around calving; breath acetone in postpartum cows has been observed in the low–tens of ppm range, with reported ranges such as ~2.3–20 ppm in on-farm measurements and prior work reporting non-ketotic vs ketotic separations in roughly 0–2 ppm vs 4–13 ppm bands (study-dependent), implying practical detectability for an embedded sensor if sampling and dilution are controlled.
