New-Onset Alzheimer’s Disease and Normal Subjects 100% Separated Statistically by P300 and ICA

Abstract

Previously, we described how patients with new-onset Alzheimer’s disease were differentiated from healthy, normal subjects to 100% accuracy, based on the amplitudes of the nonrhythmic back-projected independent components of the P300 peak at the electroencephalogram electrodes and their latency in the response to an oddball, auditory evoked potential paradigm. A neural network and a voting strategy were used for classification. Here, we consider instead the statistical distribution functions of their latencies and amplitudes and suggest that the 2-sample Kolmogorov-Smirnov test based upon their latency distribution functions offers an alternative biomarker for AD, with their amplitude distribution at the frontal electrode fp2 as possibly another. The technique is general, relatively simple, and noninvasive and might be applied for presymptomatic detection, although further validation with more subjects, preferably in multicenter studies, is recommended. It may also be applicable to study the other P300 peaks and their associated interpretations.