Event
BIOE Seminar: Bahman Afsari
Friday, September 14, 2018
9:00 a.m.-10:00 a.m.
A. James Clark Hall, Room 2132
Dr. Giuliano Scarcelli
scarc@umd.edu
Dr. Bahman Afsari
Assistant Professor, Oncology
Sydney Kimmel Cancer Center
Johns Hopkins University
Early Detection of Cancer from Non-Invasive Liquid Biopsy using Machine Learning
Early detection of cancer significantly increases the chance of successful treatment. One promising technique is regular screening using liquid biopsy, the non-invasive analysis of the body’s liquids, specially detecting of circulating-tumor-DNA (ctDNA) among other circulating DNAs in the blood samples. In this talk, we discuss the challenges in liquid biopsy analysis from statistical point of view. Specifically, we concentrate on a new blood test CancerSEEK, which for the first time investigates ctDNA simultaneously with circulating proteins and detects eight cancers with high sensitivity. Moreover, CancerSEEK can distinguish the tissue where cancer originates from with relatively high accuracy. We elaborate the statistical machinery behind CancerSEEK: Logistic Regression classifier for cancer detection and Random Forrest classifier for distinguishing tissues. Later, we describe a similar liquid biopsy technique to detect endometrial and ovarian cancer from a Pap test. Finally, we talk about a promising new technique for distinguishing tissue which may improve CancerSEEK accuracy.
About the Speaker
Bahman Afsari received his BS and MS with minor in Control Engineering from the University of Tehran. He received his MS in Statistics and his PhD in Electrical Engineering from the Johns Hopkins University under the supervision of Prof. D. Geman. His dissertation was titled "Modeling Cancer Phenotypes with Order Statistics of TranscriptData" where he studied classifiers and distributions on the ordering of gene expression for cancer prediction. During his post-doctoral fellowship at the Cancer Center at the Johns Hopkins University under Dr. E.J. Fertig, he published multiple papers and was the leading member of the winning team of the prestigious DREAM Challenge. Later, he joined Dr. C. Tomassetti Lab where, with collaboration with Prof. B. Vogelestien Lab, they worked on early detection of cancer. Currently, he is an instructor at Cancer Center of the Johns Hopkins University. His main areas of research are genomics, computational biology, machine learning algorithms, and theoretical statistics.
