Using electronic health records to look for new Alzheimer’s treatments through drug repurposing

Photo of Yonghui Wu_Square
Yonghui Wu, Ph.D.

A team led by HOBI researcher Yonghui Wu, Ph.D., received a 2-year, $85,000 Research Opportunity Seed Fund award from UF to develop less costly, more effective ways to speed up drug development for Alzheimer’s disease through drug repurposing. Key researchers on this team include co-PIs Jiang Bian, Ph.D., and Yi Guo, Ph.D., from HOBI; Glenn Smith, Ph.D., ABPP, in the department of clinical and health psychology; and Michael Jaffee, M.D., FAAN, FANA, in the department of neurology.

Drug repurposing, which involves using big data analytics and electronic health records to identify new uses for existing drugs, is one potential solution to speed up drug development for Alzheimer’s disease and Alzheimer’s disease-related dementia.

“The explosive growth of biomedical data has made it possible to discover new uses of existing drugs through computational approaches,” the researchers pointed out in their proposal. They cited  a recent study which found that drug repurposing had accounted for 20% of new drug products. Drug development has become increasingly expensive and time-consuming.  The development of a new drug typically costs from $648 million to $2.5 billion dollars and takes an average of 9-12 years.

Alzheimer’s disease is the the sixth leading cause of death in America, affecting about 5.7 million Americans. There is currently no cure, nor are there any effective treatments.

“The quality of life of these patients is gradually diminished, and the cost of caring for them is onerous and expensive for both health systems and family caregivers,” Wu said.

Previous studies have shown that patient data in electronic health records (EHR) can be used as an efficient, low-cost resource to detect drug repurposing signals in EHR as well as validate drug repurposing signals detected from other sources.

Prior studies by Wu have demonstrated the feasibility of using EHR for drug repurposing to identify potential cancer treatments. Those studies also identified the challenge of using incomplete patient information, such as family history and social and behavioral determinants of health, which are only available in clinical narratives.

In this project, the team seeks to detect potential new Alzheimer treatments by analyzing data on existing drugs in electronic health records from more than 60% of Floridians across all 67 counties housed in the OneFlorida Data Trust.

The researchers will provide a clinical natural language processing (NLP) package to extract Alzheimer’s-related family history and information about social and behavioral determinants of health from clinical notes. They will also provide computable phenotyping algorithms to identify patients with Alzheimer’s disease and their intermediate disease stages. Finally, they will validate their central hypothesis – that drug repurposing signals for Alzheimer’s disease and Alzheimer’s disease-related dementia can be detected from EHRs using a computational drug repurposing algorithm.

The team’s proposal was one of seventeen selected out of 49 finalists submitted by various colleges across UF. The 49 proposals had already undergone internal review and selection at the department and college level.  The two-year grant period began on July 15.