Premium
IC‐P‐062: Effects of Mapt Over Brain Grey Matter Atrophy in The Aibl Cohort
Author(s) -
Dore Vincent,
Porter Tenielle,
Bourgeat Pierrick,
Fripp Jurgen,
Burnham Samantha,
Macaulay S Lance,
Masters Colin L.,
Ames David,
Martins Ralph N.,
Salvado Olivier,
Rowe Christopher C.,
Villemagne Victor L.
Publication year - 2016
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.2016.06.092
Subject(s) - precuneus , atrophy , neurodegeneration , posterior cingulate , medicine , psychology , alzheimer's disease , temporal lobe , dementia , pathology , neuroscience , oncology , cortex (anatomy) , disease , cognition , epilepsy
sifier was provided the final SAS regression model’s candidate features for each diagnostic group alongside age, gender and APOE4 genotype. The classifier first ranked all features and iteratively removed those with lower weights until it found the optimal prediction model for presence or absence of brain amyloidosis while using the leave-one-out cross validation approach. We computed receiver operating characteristic (ROC) curves, predictive accuracy, and area under the curve (AUC) values to assess the classifier’s prediction performance. Results: The variants selected by step-wise linear regression models for each diagnostic category can be seen in Table. The best predictive accuracy achieved by the NC amyloidosis classifier was 79% (AUC1⁄40.83) using only APOE4, age, and gender. The best predictive accuracy achieved by the MCI amyloidosis classifier was 82% (AUC1⁄40.81) using APOE4, FERMT2, ABCA7, SORL1, and EPHA1. The best predictive accuracy achieved by the AD amyloidosis classifier was 90% (AUC1⁄40.77) using APOE4, gender, DSG2, MEF2C, EPHA1, age, and BIN1. Conclusions:Automated multimodal classifiers using AD risk genes show a promise for predicting brain amyloidosis. Further improvement of classifier accuracy may be achieved by the addition of other cognitive or biomarker measures.