Evaluation of multiple linear regression function and generalized linear model types in estimating natural menopausal...

Area of Science:

• Reproductive Health
• Biostatistics
• Epidemiology

Background:

• Menopause significantly impacts women's health, affecting approximately one-third of their lifespan.
• Accurate prediction of menopausal age and its influencing factors is crucial for enhancing women's life expectancy and well-being.

Purpose of the Study:

• To compare the predictive performance of generalized linear models (GLM) and ordinary least squares (OLS) for the age of natural menopause.
• To identify key demographic and clinical factors associated with the age of natural menopause in a large Iranian population.

Main Methods:

• A cross-sectional study utilizing data from the Shahedieh Cohort Study in Yazd, Iran.
• Inclusion of 1251 women who had experienced natural menopause.
• Application of multiple linear regression models using both OLS and GLM (Gaussian family with log link function) for age of menopause prediction.
• Model performance evaluation using Akaike information criterion, root-mean-square error (RMSE), and mean absolute error.

Main Results:

• The mean age of menopause was 49.1 ± 4.7 years (median 50 years).
• GLM with a Gaussian family and log link function demonstrated lower RMSE and mean absolute error compared to OLS.
• Factors significantly associated with menopausal age included education, history of salpingectomy, diabetes, cardiac ischemia, and depression.

Conclusions:

• Generalized linear models (GLM) with a Gaussian family and log link function offer a more accurate alternative for predicting the age of natural menopause.
• Identifying and understanding factors associated with menopausal age can inform targeted health interventions and improve women's health outcomes.