FIGURE 5.

Multiblock data fusion applied to study the metabolic alterations in human plasma samples from patient suffering from chronic diseases. (A) Data sets used in this study: metabolomics measurements (LC–MS and NMR), the metadata set containing life-style information, and the label information corresponding to each sample. (B) Acute coronary syndrome. (a) Average ROC curves showing the forecasting performance of individual data sets as well as fusion methods for women and men. (b) Boxplots summarize the performance of different approaches across 100 training/test sets. (C) Average ROC curves illustrating the forecasting performance of (a) breast cancer and (b) colon cancer. Figure reproduced with permission from (Acar et al., 2017).