MilkBot™ Applications

In our development work, it has become clear that there is a great deal of variability in the shape of individual lactation curves, and also in the short term variability of data around each curve. There is little published research to explain this variability (with the exception of the relationship between persistence and parity). We presume these differences between lactations are caused by at least two different sorts of causes.

First, there are any number of reasons why individuals or entire herds show a temporary rise or fall in production. Examples would include changes in feed quality, lameness, illness, environmental changes, social changes (such as group changes), ... all things a dairy manager tries to control. Every good dairy manager watches milk production closely, but this is complicated by the normal rise and fall in milk production of the lactation curve. MilkBot can be used to increase the precision with which we measure changes in daily production.

Second, MilkBot can characterize an entire lactation not only with respect to total production (as is done now), but also with respect to critical aspects of the shape of the lactation curve. Some of the differences between lactations are certainly genetic, while others are likely environmental or nutritional. In either case, MilkBot can be a powerful tool for identifying and quantifying differences in dairy productivity.

"ME305", an example

The term "ME305" (Mature-Equivalent 305-day milk) has been used to estimate production over a standard 305-day lactation, normalized for age or lactation number, and also (often) for season of calving. There are a number of different ways of calculating this statistic, and each calculation method implies an inherent relationship between production and other variables (age, lactation number, season), and also a specific shape for lactation curves. If these hidden relationships are estimated incorrectly, that will bias the ME305 calculation, as will changes in lactation curve shape. ME305 calculations are commonly biased in practice, due to differences between populations or over time, and this bias can lead to significant errors.

MilkBot makes it easy to customize, update, or extend this kind of calculation. First we calculate "LT305" (Lactation Total, not adjusted) for a population of lactations (the integral of the MilkBot function from 0 to 305, which is easily calculated). Then we stratify the population by each variable of interest (such as lactation number), and express the relationship algebraicly or as a table of correction factors. We can then define a valid population-specific ME305 function tuned to a particular environment or genetic group, and monitor the performance of that function as conditions change.

Herd or group differences in factors entering into a calculation of ME305 are of interest in their own right. For example, we would expect seasonal correction factors to reflect the skills of the herd manager in managing heat stress.

For economic calculations, a 305-day lactation is unrealisticly low in many herds. MilkBot makes it easy to calculate ME350, or other variants which may reflect actual conditions more accurately.

Evaluating the Impact of Some Intervention

It is common to want to demonstrate what effect (if any) a product or intervention has on milk production. Since there is a lot of natural variability in production, it may take a very large trial, or multiple trials to quantify this response with any precision. MilkBot can cut out a lot of the "noise" inherent to this kind of analysis, and so increase our ability to quantify relationships.

Exploratory Analysis

It is reasonable to expect that not all factors influencing production will influence the lactation curve uniformly. That is, some may have their strongest influence primarily early lactation while others might influence Persistence, or Scale. By quantifying lactation curve shape as well as magnitude, it becomes possible to discover weak relationships that might be hidden with a more general measure like ME305.

Current statistical tools can efficiently identify covariance between variables in a large data set. By providing new derived variables that reflect lactation curve shape in a natural way, we increase the probability of discovering relationships that may influence the lactation curve unevenly. The pseudo-physiological derivation of the MilkBot model may then provide some guidance in interpreting results or directing future study.

Overview gives a high level view of MilkBot^™'s rationale. From there, you may wish to look at the model behind the computer method, or a detailed description of MilkBot^™ parameters. We discuss some possible uses of MilkBot^™ technology, and there is an interesting side discussion of the MilkBot^™ fitting engine. See our contact information, for more information on getting access to MilkBot^™ technology.