Eye Tracking and Machine Learning Non-invasive Biomarker for Alzheimer’s Disease and Frontotemporal Dementia Diagnosis

In this publication, it is presented the first diagnostic biomarker for Alzheimer’s disease (AD) and the behavioural variant of frontotemporal dementia (BvFTD) based on machine learning techniques applied to eye movement analysis. Nowadays, the diagnostic of these dementias is clinic, so it is based on the specialist doctor by their own expertise knowledge and the results of complementary medical tests. This study includes 39 control participants, 38 participants with moderate AD diagnosis and 24 participants with a diagnosis of BvFTD. Ocular movements were recorded with video-oculography (saccadic, fixation and smooth pursuit paradigms). Results: applied methodology to data mining and feature extraction shows clear evidence of the differences between each type of participants. Accuracies and AUC overcome 95%. Conclusion: Video-oculography is a non-invasive, low expensive, objective and fast technique for cognitive evaluation that could help to the clinical diagnosis.