A Markov chain model for animal estrous cycling data

Area of Science:

• Reproductive biology and toxicology
• Statistical modeling in animal science

Background:

• Estrous cycling data, represented by sequences of daily stages (e.g., DPEMD), are crucial for understanding animal reproductive health.
• Changes in estrous cycle patterns, influenced by stage lengths, can indicate adverse events or treatment effects.
• Direct observation of estrous stage lengths is often not feasible, necessitating methods to infer this information from available data.

Purpose of the Study:

• To develop and apply a statistical model for analyzing estrous cycling data, focusing on inferring stage lengths.
• To investigate the relationship between treatment effects and variations in estrous cycle stage lengths.
• To provide a framework for detecting adverse events through the analysis of estrous cycle patterns.

Main Methods:

• A Markov chain model was proposed to approximate the estrous cycling process, enabling the derivation of transition probabilities.
• Interval-censored stage lengths were extracted from the data for modeling, excluding the first and last stages.
• Weibull distributions were assumed for stage lengths, incorporating treatment effects and animal-specific random effects.
• Regression models were fitted to the censored stage lengths, utilizing a Bayesian approach for inference on dose effects.
• Markov Chain Monte Carlo (MCMC) methods were employed for analysis using WinBUGS software.

Main Results:

• The proposed Markov chain model successfully approximated the estrous cycling process and allowed for the estimation of stage lengths.
• The Bayesian analysis, incorporating Weibull distributions and regression on censored data, effectively estimated treatment and random effects.
• The methodology was demonstrated on a real-world dataset from a National Toxicology Program study, showing its practical applicability.

Conclusions:

• The developed Markov chain model, combined with Bayesian inference and Weibull distributions, offers a robust method for analyzing estrous cycle data.
• This approach enables the inference of unobserved stage lengths and the assessment of treatment effects on reproductive patterns.
• The study highlights the utility of this statistical framework for identifying potential adverse events in toxicological studies involving animal reproduction.