O1‐11‐04: Temporal Clustering Reveals Heterogeneity Of Cognitive Decline In Dementia

• DOI: 10.1016/j.jalz.2016.06.355
• Publication Date: 2016-10-17T07:24:13Z
• Authors: Steven John Kiddle; Alice Parodi; Caroline Johnston; Chris Wallace; Richard JB. Dobson
• Keywords: Cognitive Decline; Longitudinal data

It is challenging to study long-term progression of Alzheimer’s disease due short-term follow-up individuals and heterogeneity. Progression can be heterogeneous in itself, but additionally begin studies at different ages, times relative onset. Here we propose the first method – Temporal Clustering that jointly tackle these sources heterogeneity produce models progression. Data used preparation this article was obtained from PAQUID (data extracted R package lcmm), Disease Neuroimaging Initiative (ADNI; www.adni.loni.usc.edu), Australian Imaging Biomarkers Lifestyle flagship ageing (AIBL; www.aibl.csiro.au) Coalition Against Major Diseases (www.c-path.org/programs/camd/). For CAMD focused on placebo arms two longest trials (1013 1014). From all took data with least MMSE scores. In total studied longitudinal Mini Mental State Examination (MMSE) for 2,833 individuals. We so could compare ADNI, AIBL decade-long a subset PAQUID. clustering joint alignment capable dealing sparsely observed time series irregular time-points. Alignment allows account variability onset respectively. demonstrate temporal interpretable cognitive decline, results derived are broadly similar those longer periods. Finally show who appear progress faster more likely younger. These may aggressive subtype decline. have introduced new methodology combines build has potential stratifying based their rate potentially useful understanding etiology optimal design clinical trials. Future work will involve developing multivariate version relationship between biomarkers.