[O2–16–03]: MODELLING EYE‐TRACKING DATA TO DISCRIMINATE BETWEEN ALZHEIMER’S PATIENTS AND HEALTHY CONTROLS

Background Eye movement abnormalities are common and underrecognised in posterior cortical atrophy (PCA), and may help distinguish PCA from typical (amnestic) Alzheimer's disease (tAD)(Shakespeare et al., 2015). We model data obtained from eye-tracking in a way that incorporates how well tasks are performed along with the raw data. This model allows us to use machine learning techniques to predict diagnoses and analyse any individual's data in comparison to a healthy individual. Methods Eye tracking data were collected on fifty-seven participants (36 patients with young onset dementia [N= 26 typical tAD; N= 10 PCA] and 21 age-matched healthy controls) while they were performing a smooth pursuit task. We use a hidden Markov model (HMM)(Baum & Petrie, 1966) to model movements in gaze location in the smooth pursuit task. Task performance is incorporated into the model via a mechanism that …