I am interested in the early diagnosis and tracking of neurodegeneration in Alzheimer’s dementia through fusion of data from in-home, multimodal sensor arrays. My PhD project, part of the SPHERE/CUBOiD collaboration at Bristol University, will use computational models to explore how changes in natural speech reflect behavioural disturbances in mild cognitive impairment (MCI) and early Alzheimer’s disease. I will examine a) how variability in features of natural speech relates to disruptions in biomarkers of daily living activities (such as sleep and locomotion patterns) and b) how behavioural and speech features evolve together with disease progression. By fusing neuropsychological, speech and behavioural sensor data to create enriched feature sets, I aim to build optimised pipelines for early classification of MCI and Alzheimer’s disease from natural speech alone. The overarching aim of this project is to predict future progression to Alzheimer’s disease from passively-acquired natural speech.