Predicting Alzheimer’s Risk

In Alzheimer’s patients, disease progression begins long before major symptoms occur, sometimes decades earlier. An urgent need exists for quantitative biomarkers and genetic tests that can predict the risks for and progression of Alzheimer’s Disease (AD) at the individual level and during this preclinical stage.

Mert R. Sabuncu, Electrical and Computer Engineering/Meinig School of Biomedical Engineering, is developing cutting-edge machine learning algorithms—able to mine high-dimensional, multi-modal, and longitudinal data—to derive models that yield clinical predictions for those at risk or in the early stages of dementia. The prognostic models use ubiquitous and affordable data types: structural brain MRI scans, genome-wide sequence data derived from blood or saliva, and demographic variables (age, education, and sex). These variables are strongly associated with clinical decline to dementia, but to date, no model exists that can harvest all the predictive information embedded in these high-dimensional data.

Machine learning algorithms are increasingly used to compute clinical predictions from high-dimensional biomedical data such as clinical scans. Most prior methods were developed for applications where the data was limited to certain factors, and multiple modalities, such as genotype and images, and longitudinal data were not fully exploited. This new application’s core innovation is to develop rigorous, flexible, and practical machine learning methods that can fully exploit the available data to compute prognostic clinical predictions.

The algorithms and software will yield invaluable tools for stratifying preclinical AD subjects in drug trials, optimizing future therapies, and lowering the risk of AD’s most devastating effects. NIH Award Number: 1R01AG053949-01A1

Cornell Researchers

Funding Received

$2 Million spanning 5 years

Other Research Sponsored by National Institutes of Health, National Institute on Aging