Use of a non-linear spline regression to model time-varying fluctuations in mammary-secretion element concentrations of periparturient mares in Michigan, USA.

This research focused on modeling the changes in certain element concentrations in a mare’s mammary secretions around the time of birth. Using non-linear regression techniques on data from pregnant Arabian mares, researchers found notable changes in element concentrations over time and individual variations. This might be used to predict when the mares will give birth and could contribute to developing neonatal milk substitutes.

Study Objective and Design

• The research aimed to model the changing concentrations of nine specific elements in mammary secretions from periparturient (around the time of birth) mares over time. This could potentially help predict when a mare is about to give birth.
• The study was conducted during the 1992 foaling season and collected serial samples of mammary secretions from all 18 pregnant Arabian mares at the Michigan State University equine teaching and research center.

Methodology and Analysis

• The researchers used non-linear regression techniques to model the relationship between element concentration in mammary secretions and days from foaling. The model incorporated two distinct sigmoid (S-shaped) curves, connected via a spline function—an advanced mathematical tool to account for non-linear relationships.
• The research also took into account variables such as the mare’s individual characteristics and parity (the number of times a mare had previously given birth).

Findings

• Element concentrations in mammary secretions showed significant variation during the periparturient period in mares. There were trends over time and individual mares also varied significantly.
• Multi-parity mares showed lower concentrations of potassium (K) and zinc (Zn), but higher concentrations of sodium (Na).
• The study detected considerable serial and spatial correlation, regardless of modeling efforts. This meant that p-values obtained for parameter estimates were likely biased toward zero.

Implications and Future Opportunities

• The research suggests that monitoring changes in mammary-secretion element concentrations might be used to predict impending parturition in mares.
• The findings also imply that these element concentrations need consideration in developing neonatal milk-replacement therapies.
• The study successfully demonstrated the use of non-linear regression to model time-series data in animal health management, offering an approach that could be beneficial to other research with analogous analytical challenges.