Premium
P1‐273: ASSESSING HETEROGENEITY OF ALZHEIMER'S DISEASE THROUGH THE EYE
Author(s) -
Csincsik Lajos,
Quinn Nicola,
MacGillivray Tom,
Shakespeare Timothy J.,
Crutch Sebastian J.,
Peto Tunde,
Lengyel Imre
Publication year - 2019
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.2019.06.828
Subject(s) - corticobasal degeneration , macular degeneration , posterior cortical atrophy , dementia , dementia with lewy bodies , pathological , retinal , atrophy , ophthalmology , medicine , retina , occipital lobe , progressive supranuclear palsy , pathology , disease , neuroscience , psychology
included. Sparse Partial Least Squares-Discriminant Analysis was performed in order to classify individuals as AD, MCI or CN. Four models were built, one for each of the four variables groups separately. We built two additional models with the four variables groups considered as (1) a unique modality, or (2) independent modalities. Classification performance of each of the 6 models was assessed using Area Under the Curve (AUC) computed for each diagnosis versus the other two, leading to 18 AUC values. Results: Three AUC were above 0.75. For AD vs MCI and CN diagnoses, the unique modality and imaging variables group models had AUCs of 0.75 [0.66-0.83] and 0.76 [0.68-0.84], respectively. For CN vs AD and MCI diagnoses, the model with the unique modality had an AUC of 0.76 [0.68-0.84]. For MCI versus AD and CN diagnoses, no model reached an AUC higher than 0.6. Conclusions: We built diagnostic models in order to improve classification of AD, MCI and CN individuals. Whereas no model reached a high AUC, we were however able to build two models with AUC of 0.76 to classify (i) AD vs MCI and CN and (ii) CN versus AD andMCI. Interestingly, identification ofMCI versus AD and CN diagnoses seems more complex to discriminate, potentially reflecting its intermediate and heterogeneous status between AD and CN.