We are currently conducting a preliminary study to examine the feasibility of employing complex network methods in the analysis of the CARDIA dataset and will present our preliminary findings. Our main hypothesis is that network analysis can reveal information on how different combinations of risk factors can be assessed at a larger scale.
Participants in this study [baseline (1985-86) and follow-ups (1987-88, 1990-91, 1992-93, 1995-96, 2000-01, 2005-06, 2010-11, and 2015-16)] will be connected to each other according to the correlation of the measured parameters (e.g. biomarkers, anthropometric measurements, psychosocial factors). The planned analysis are based on the entirety of the studied population (baseline and eight follow-ups).
Variation of the correlation threshold will lead to the construction of a network between individuals whose measurements behave similarly to each other. Modularity analysis will allow us to determine the degree of clustering in the network and whether cluster membership (network property) can be a predictor of cardiovascular risk and morbidity. Different networks will be constructed based on different subsets of parameters and combinations of those subsets in order to determine the relative influence of each subset in the form of the resulting network.
Should this preliminary analysis be successful, next steps will delve into the ascertainment of complex mechanisms that determine critical transitions or regulate stability, resilience, and predictability in chronically elusive cardiovascular disease burden.