[02068] Variable selection aided by correlation networks
Session Time & Room : 5D (Aug.25, 15:30-17:10) @D515
Type : Contributed Talk
Abstract : Variable selection is important because it can provide improved quality of results, faster times of computation and more explainable models. We present recent work in which we use data from over 100 000 cells to find a selection of morphokinetic variables guided by nonlinear correlation networks, able to capture behavioral landscapes of inflammation. Our mathematical modeling, based on logistic and decision tree models, allowed us to identify the most important variables for immune cell prediction.