So currently for EEG, there are many limitations to its widespread implementation, ranging from data acquisition and access, appropriate training of technicians in order to do this type of data acquisition. And in addition to that, once we have data, actually generating good quality outputs from that data, particularly related to denoising and so data science approaches are necessary in terms of being able to distill signal from noise and then from there utilize that for downstream classification...
So currently for EEG, there are many limitations to its widespread implementation, ranging from data acquisition and access, appropriate training of technicians in order to do this type of data acquisition. And in addition to that, once we have data, actually generating good quality outputs from that data, particularly related to denoising and so data science approaches are necessary in terms of being able to distill signal from noise and then from there utilize that for downstream classification. This study builds on our recently published work in the Journal of Alzheimer’s Disease where we published an EEG study demonstrating that we can apply AI-based pre-processing and post-processing analytics to be able to utilize publicly available EEG data to differentiate between patients with Alzheimer’s disease relative to controls. And this study now sought to build on that to see if we can improve on our pre-processing using deep learning approaches. Again, as mentioned, particularly with application of denoising in order to be able to remove artifacts that in many cases require manual removal or rejection. And from there, that level of data pre-processing enhancement can in turn improve our classification using traditional unsupervised and supervised machine learning models.
This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.
