Research Interests

Wordcloud generated using Abstract Wordcloud

My research focuses on understanding early brain changes in preclinical Alzheimer’s disease using neuroimaging methods, in particular, positron emission tomography (PET) scans of amyloid plaques and neurofibrillary tau tangles. I am interested in quantifying the contribution of different neuropathologies to the propagation of brain changes among cognitively normal individuals and ultimately, to cognitive decline. I design statistical models and methods for analyzing longitudinal neuroimaging and other biomarker data as well as cognitive measures to enable progress towards this goal.

A major challenge in studying preclinical Alzheimer’s disease is the extensive duration of this period and the limited longitudinal follow-up in existing observational studies. As a result, there is insufficient data at the individual level spanning the entire preclinical period leading up to clinical presentation of disease. Statistical modeling provides ways of leveraging short-term observations to obtain long-term trajectories.

The Disease Progression Score Model is a model that I have made significant contributions to that allows for the delineation of the natural history of biomarkers and cognitive measures implicated in Alzheimer’s. This model can be used examine the temporal ordering of biomarkers, to determine a given individual’s disease stage based on a collection of longitudinal biomarker measurements, and to predict future progression.

Amyloid and tau pathologies are the two major neuropathological hallmarks of Alzheimer’s. I designed a model to estimate the age at onset of amyloid accumulation. This work has been followed by improved methods applied to multiple cohorts to examine the interval between amyloid and cognitive impairment onset and modifiers of this interval. Estimated amyloid onset age can be used as an outcome in studies and potentially in clinical trials to assess the effects of risk factors and therapies in altering the disease timeline.

Such statistical methods will aid in the understanding of the earliest changes leading to Alzheimer’s disease, enable the identification of strategies and time windows for therapeutic intervention, and allow for performing prognostic predictions at the individual level and offer personalized prevention approaches.

I also design streamlined workflows for 3D/4D PET image analysis and address data harmonization challenges in multi-site collaborations.