Beta‐amyloid regions related to amyloid positivity using a feature selection method.

Abstract Background Cerebral beta‐amyloid (Aβ) is a hallmark of AD. Few studies have revealed which of the regional Aβ markers are associated with abnormal levels of Aβ status. The aim of the current study was to identify multivariate regional Aβ markers predicting Aβ positivity (Aβ+). Method Seven hundred and sixty‐five participants from the ADNI‐2 cohort at baseline visit were included in the study. We used regional Aβ uptake as predictors in a predictive model. Flobetapir PET images were classified as Aβ‐positive if the standard uptake value ratio was over 1.1. The least absolute shrinkage and selection operator (LASSO) model with 1,000 bootstrap was implemented to identify regional Aβ traits to predict cortical amyloid burden from participants. Result Out of 116 predictors, 27 markers in cortical thickness significantly predicted Aβ+ among participants at baseline. After bootstrapping, 2 phenotypes had 95% confidence intervals that did not overlap zero: left precuneus (β = 0.684) and right rostral middle frontal region (β = 0.605). Conclusion These results demonstrated that several brain regions are related to abnormal levels of cerebral Aβ. These markers suggest robust sign that may capture AD pathology. To this end, this research was supported by the MSIT (Ministry of Science and ICT), Korea, under the ITRC (Information Technology Research Center) support program (IITP‐2022‐2017‐0‐01630) supervised by the IITP (Institute for Information & communications Technology Promotion).