Premium
P1–147: A quadratic, P‐spline, mixed model to predict cognitive decline in Alzheimer's disease
Author(s) -
Mouiha Abderazzak,
Duchesne Simon
Publication year - 2013
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2013.05.369
Subject(s) - smoothing , neuropsychology , medicine , biomarker , alzheimer's disease neuroimaging initiative , demographics , cog , oncology , dementia , disease , cognition , mathematics , statistics , artificial intelligence , psychiatry , demography , biology , computer science , biochemistry , sociology
Cog test as a surrogate marker of time, related to disease progression. We constructed the model used for analyzing the data in two levels (within-subjects and between-subjects model). Results: Figure 1 shows mean and individual profiles for ADAS-Cog vs. CSF A b, CSF total tau, FDG PET and hippocampal volumes, for controls and MCI having progressed to probable AD, per sex. In all cases, there were significant yintercept differences for control and MCI subjects. For all biomarkers, there was a significant y-intercept difference between males and females in the MCI population, but no difference in the mean regression slope, whereas this situation was only statistically significant in controls for pTau. Conclusions: Linear mixed effect models demonstrated baseline differences in all biomarkers possibly extending some years before inclusion in the study, yet no statistically significant difference in progression over the 3-year timeframe. The analysis also showed that sex should be taken into consideration when attempting to derive biomarker trajectories.