MODELING FACTORS CAUSING ALZHEIMER’S DISEASE USING LOGIT, PROBIT, AND GOMPIT LINK FUNCTIONS IN GENERALIZED LINEAR MODEL

Abstrak

This study addresses the ongoing challenge of clarifying the risk factors contributing to Alzheimer's disease, a neurodegenerative condition marked by progressive cognitive decline and memory dysfunction, with cases rising globally. To provide a more accurate and comprehensive understanding of the predictors associated with the disease, this research models the contributing factors using logit, probit, and gompit link functions within the Generalized Linear Model (GLM). Utilizing secondary data from 2024, which includes predictor variables such as age, family history, head injury, hypertension, memory complaints, and behavioral disturbances, this research models the relationship between these variables and Alzheimer's diagnosis. The analysis finds that the logit, probit, and gompit link functions yield significant results in identifying risk factors associated with Alzheimer's diagnosis, particularly memory complaints and behavioral disturbances. The gompit link is selected as the best model due to its highest deviance R-squared value of 30.01%, indicating better reliability in predicting Alzheimer's diagnosis than other models. This GLM approach provides insights to support early prevention and intervention efforts for Alzheimer's disease and contribute to achieving Sustainable Development Goals (SDGs) number 3 on good health and well-being.