Andrew Beam received his PhD from North Carolina State University under the supervision of Jon Doyle. Beam’s doctoral research focused on machine learning methodology for genome-wide association studies. Since coming to DBMI, his work has shifted to clinical applications of machine learning and genomics. The central theme of his research is to help create a smarter healthcare system through the use of computational techniques. Gene-expression data contains a wealth of information that is currently undervalued as a clinical tool. For example, in inflammatory bowel disease (IBD), a patient’s response to standard of care is typically complex and heterogeneous. Deciding which of many therapies a patient will receive is often an ad-hoc procedure that is not necessarily conditioned on the specifics of an individual patient. Using large gene-expression datasets on both patients and drugs, he is developing techniques that will allow patients to be matched to the drug with the highest probability of success. Using this approach, thousands of drugs may quickly be analyzed to see if they have off-label therapeutic potential for IBD patients, which is very important in refractory patients for whom standard of care has not been successful. Additionally, Beam is developing a machine learning approach that synthesizes large amounts of the biomedical literature using natural language processing to provide clinical decision-support in the form of differential diagnoses. This project has the potential to improve many aspects of healthcare by reducing medical errors, improving rare-disease diagnosis, and increasing overall provider efficiency.
BioData mining, February 6, 2015
BMC bioinformatics, November 21, 2014
Journal of pharmacogenomics & pharmacoproteomics, March 1, 2014
Reproductive toxicology (Elmsford, N.Y.), December 9, 2011
Dose-response : a publication of International Hormesis Society, June 25, 2010
Journal of toxicology and environmental health. Part B, Critical reviews, February 1, 2010