Haimovich ('10) Featured in Engineering News
Adrian Haimovich (B.S. Applied Mathematics, ‘10) was featured in the article “Predicting Diseases” in the Spring 2010 Leaders Making an Impact Issue of Columbia Engineering News.
Seven major diseases - diabetes types I and II, bipolar disorder, high cholesterol, coronary artery disease, rheumatoid arthritis, and high blood pressure - may be reliably predicted based on analysis of genome- wide association studies (GWAS), while unlocking complex problems like the biological cause of cancer - the second-leading cause of all deaths.
As researchers delve further into network-based biology, investigators have found themselves increasingly reliant on not only clinical knowledge, but also statistics, computational sciences, and mathematics. Adrian Haimovich ’10, an applied mathematics major, has been interested in computational biology since the 9th grade.
By the time he finished high school, he had several years of summer lab experience as well as an academic publication. Upon arriving at Columbia Engineering, he sought out those professors whose work provided the foundation of his own research, including electrical engineering Prof. Dimitris Anastassiou. Anastassiou’s genomics research spans seven major diseases and is of a computationally challenging scale. His work applies tools from electrical engineering to problems in quantitative biology.
“While working on those large-scale genomic data, I became interested in applications of Prof. Anastassious’s ideas in information theory to other types of clinically relevant problems,” says Haimovich, specifically those that trace physiological responses, in the form of gene expression, to either medical conditions or experimental protocols. By junior year, Haimovich had begun to work with new datasets based on the clinical condition sepsis, which is characterized by severe systemic inflammation.
He looked to extend the work on sepsis under the supervision of Prof. Chris Wiggins of the APAM Department. Wiggins suggested using machine learning methods central to his own lab’s work to help Haimovich analyze his data. Working in the Wiggins group, Haimovich applied support vector machine (SVM) techniques using data from patients who were treated with either an endotoxin or placebo. The results from this work indicate genes that are strong classifiers for sepsis. The work continues as Haimovich looks to use SVMs to make clinical studies more efficient.
“Engineering mathematics is a powerful and elegant way to look at a biological problem,” says Haimovich, “and computational biology can be used to make great advances in patient care.”