CSF/blood biomarker algorithms for the diagnosis of AD, DLB, FTD

We will develop specific algorithms for defining the biochemical signature of patients. Multivariate statistical methods will be applied, able to take into consideration the correlation structure of the data and the synergies and interactions existing among the potential biomarkers. Furthermore, we will investigate the use of machine learning approaches, such as clustering and deep learning, to develop robust classifiers. We will develop a classification scheme for FTD and DLB, similar to the A/T/N scheme for AD29. Results will be validated in independent cohorts listed in T3.2.