Machine learning for a dementia diagnosis utilising eyes open and eyes closed resting state EEG

Due to the ageing population within the UK, it is now estimated that 850,000 people live with some form of dementia, leading to dementia becoming the leading cause of death within the UK. With Alzheimer's disease (AD) accounting for 60% of all such cases.

Improving our ability to detect and correctly classify dementia and its different sub-types is of great importance, especially when one considers that some dementia types such as dementia with Lewy Bodies (DLB) are consistently misdiagnosed as AD leading to a greater mortality rate due to DLB having a much greater sensitivity to neuroleptics.

It has recently been shown that resting state eyes closed electroencephalography (EEG) abnormalities present within dementia groups can be used to differentiate from healthy control groups with machine learning methods. In addition, it has been shown that such methods allow for classification between dementia groups; most significantly having the ability to differentiate between the AD and DLB dementia types.

We propose that the inclusion of resting state eyes open EEG will improve the classification, and that similar results can be achieved using a lower nodal density setup with a reduced sampling rate. We also present an as yet uncommented upon phenomena between healthy controls and dementia patients for eyes closed and eyes open resting state EEG.