Fig. 5: Integration of APPRS into disease prediction models.
From: Plasma proteome mediates the associations between air pollution exposure and disease risk
A Schematic framework of the disease prediction model incorporating APPRS. In addition to basic demographic (age, sex, smoking status, alcohol consumption, race, and BMI) and clinical information (ApoB/ApoA, albumin, creatinine, HbA1c, and glucose), APPRS is included as a feature to enhance disease risk prediction associated with air pollution exposure. Partly Created in BioRender. Li, W. (2025) https://BioRender.com/i8vhsv3. AUROC performance comparison of models with and without APPRS as an input feature: Basic model vs. Basic + APPRS model (B), Clinical model vs. Clinical + APPRS model (C), and Basic + Clinical model vs. Basic + Clinical + APPRS model (D). E Contribution of different features in disease prediction models across various diseases. F Validation of APPRS contribution to disease prediction in an external cohort. Model performance was compared between the baseline model and models incorporating Arrhythmias APPRS or PAD APPRS. Due to data availability constraints, the baseline model was built only using age, sex, race, and BMI. G Evaluation of APPRS contributions in the external validation cohort. Source data are provided in the Source Data file.
