What We Measure
To detect mastitis, we track each animal’s milk yield, fat, protein, lactose, milk temperature, and milk conductivity during every milking. This gives us an extensive picture of changes in the animal’s health over time. Looking for anomalies and sudden changes in these variables is a good window into whether the animal is experiencing stress — from mastitis or another issue.
Side note: We don’t just look at the animal alone. We also compare each animal’s results with the rest of the herd to see if changes are unusual for that animal compared to normal herd patterns. We also don’t just look at the last day — we look at the last 30 days to see if today is different from the trends we observe over time.
Stress Detection
Stress is the first signal we look for when detecting mastitis. It shows up mainly as (but is not limited to):
- Milk yield goes down
- Fat goes up
- Lactose goes down
When these three changes happen together, the animal is likely experiencing stress. We always track the combination of variables, not just one of them, because changes in a single variable might be due to other events, coincidence, or simply noise. An increase in stress can signal many things, not just mastitis — stress can be caused by another illness, injury, or other disruptions (just like with humans).
Mastitis Detection
The mastitis score comes mainly from milk conductivity. When conductivity rises suddenly, it often means the udder is inflamed, which is typical of mastitis. The score is also affected by changes in milk yield, fat, protein, and lactose. These additional factors help refine the signal and make it more reliable, since mastitis rarely shows up in just one variable. Other measures also play a role, but conductivity is one of the clearest signals for mastitis.
How Alerts Are Issued
We combine the stress score with the mastitis score. When both are elevated, the system issues a mastitis alert. In basic terms, this means the animal is stressed and its milk conductivity has risen suddenly. Alerts are shown in three levels: unsure, likely, or very likely, with each step meaning a higher chance that the animal is truly sick with mastitis.
Important to Know
- Alerts can sometimes signal another illness, not necessarily mastitis, or be caused by rare noise in the data. This means we might occasionally mistake mastitis for another condition or issue an alert for a healthy animal due to unusual or noisy readings. This is why alerts are categorized as unsure, likely, or very likely. It will never be 100% accurate — there will sometimes be false positives.
- Mastitis is easiest to detect when it first begins; over time, if left untreated, symptoms may become less obvious.
- Mastitis alerts are paused during the dry period, since changes in yield, fat, lactose, and conductivity then can look very similar to stress or illness — but are not.
