A semi-mechanistic mathematical framework for simulating multi-hormone dynamics in reproductive endocrinology

Area of Science:

• Reproductive endocrinology
• Computational biology
• Mathematical modeling

Background:

• Ovarian hormone dynamics are crucial for reproductive health but complex to analyze.
• Existing methods face challenges in simulating cyclic hormone variations and individual differences.
• A need exists for robust tools to model and understand these complex physiological processes.

Purpose of the Study:

• To develop a computational framework for generating physiologically constrained, multi-hormone synthetic time series.
• To capture intra- and inter-individual variability across different reproductive phenotypes.
• To create a tool for analysis, simulation, and education in reproductive physiology.

Main Methods:

• A semi-mechanistic mathematical model was created to generate synthetic profiles for estradiol, FSH, LH, AMH, testosterone, and GnRH.
• Parametric equations incorporated known physiological feedback loops and stochastic components.
• Eumenorrheic and PCOS-like phenotypes were simulated by adjusting model parameters.

Main Results:

• Synthetic profiles accurately reflected distinct hormonal patterns for eumenorrheic (classical peaks) and PCOS-like (elevated LH/testosterone, blunted estradiol) phenotypes.
• Principal Component Analysis (PCA) effectively separated phenotypes, explaining 82% of variance.
• Logistic regression achieved 100% accuracy in discriminating between simulated phenotypes.

Conclusions:

• The simulation framework generates physiologically accurate hormone dynamics.
• It successfully discriminates between ovulatory and anovulatory (PCOS-like) cycles.
• Applications include AI training, phenotype discovery, and enhancing medical education in reproductive endocrinology.