Information Fusion of Cognition, and Brain MRI on Cardio-Metabolic Risk Factors in the Middle-Aged: the CARDIA Study

点击翻译按钮获取中文摘要

Background: There is a lack of evidence of the role of cardio-metabolic risk factors in brain health from large cohort studies which are comprised of metabolites; specifically, from a life-course perspective with increasing CVD risk profiles. Aim: The primary aim of this study is to investigate associations between cardio-metabolic risk factors and cognition with brain MRI characteristics in the Coronary Artery Risk Development in Young Adults (CARDIA) Study at Year 30, and to investigate if the urinary metabolites are mediators of the associations. Significant epidemiological exposures and targeted urinary metabolites will be examined, and the information will be used to develop a risk assessment tool for brain health (i.e., cognition).Design and population: This present study focuses on the CARDIA Study sub-cohort of the 606 participants in middle-aged (aged 48-60 years who completed cognitive tests) at Year 30, 2015-16. The high-resolution nuclear magnetic resonance (NMR) spectroscopy and the liquid-chromatography mass spectrometry (LC-MS) data for the 606 participants and the brain MRI data for the 280 participant are available. Methods: Bayesian models and advanced network classification models will be employed to investigate metabolic phenotypes and other exposures. Different types of significant cardio-metabolic risk factors, other clinical, socio-demographic, lifestyle exposures and metabolite markers on cognition and brain MRI measures will be integrated (i.e., fusion) to provide a cardio-metabolic risk-specific integrated map of brain health. The longitudinal associations will be examined to explore the possibility of developing a novel, comprehensive risk assessment tool of brain health.Discussion: This study will employ a novel biostatistical approach of a fusion analytic method for the different types or levels of data on the large-scale epidemiological study. The approach may be able to recognise complex multivariate epidemiologic, pathologic and phenotypic relationships across the disease networks that are yet unidentified.