Year: 2025
Type: article
Abstract: This study develops a methodology using R software for modelling alanine transferase (ALT) with total cholesterol (TC), triglycerides (TG), and alkaline phosphatase (ALP) as predictors. Combining generalized additive models (GAMs) and neural networks (NN), the framework identified ALP as the most significant contributor (80.28%), followed by TG (12.37%) and TC (7.34%). Performance metrics, including RMSE and MAE, demonstrated the neural network’s effectiveness in capturing non-linear relationships, achieving 98.59% accuracy. This practical approach offers valuable insights for biostatistical applications in health and clinical studies.
Source: JP Journal of Biostatistics
Authors: Wan Muhamad Amir W Ahmad, Seng Fei Hong, Faiza Awais, Mohamad Nasarudin Adnan, Farah Muna Mohamad Ghazali, Nor Azlida Aleng
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: hybrid
Funder: Universiti Sains Malaysia
OpenAlex ID: W4407656166