This study focuses on the glycosylation of IgG (immunoglobulin G) and its close association with physiological states and diseases. The authors aimed to predict biochemical and physiological parameters using deep learning models trained on IgG glycome composition data. Glycosylation of IgG is influenced by both genetic and environmental factors, making the glycome a potential biomarker for specific health conditions and diseases. Using data collected from study participants, including IgG glycome compositions and corresponding biochemical and physiological parameters (e.g., age, sex, BMI, blood pressure, cholesterol levels, etc.), the authors developed a deep learning model capable of accurately predicting these parameters.
The results demonstrated that the proposed deep learning model achieved high accuracy in predicting biochemical and physiological parameters based on IgG glycome data. Notably, for key parameters such as age, BMI, and blood pressure, the model outperformed traditional approaches like linear regression and random forest models.