Online Program Home
  My Program

Abstract Details

Activity Number: 291 - SPEED: Biopharmaceutical Statistics
Type: Contributed
Date/Time: Tuesday, August 1, 2017 : 8:30 AM to 10:20 AM
Sponsor: Section on Statistical Learning and Data Science
Abstract #323973 View Presentation
Title: Robust Feature Selection and Cell Line Classification with Electric Cell-Substrate Impedance Sensing Data
Author(s): Megan Gelsinger* and David S Matteson and Laurie Tupper
Companies: Cornell University and Cornell University and Williams College
Keywords: time series ; biophysics ; functional data ; statistical learning
Abstract:

Electric Cell-substrate Impedance Sensing is a powerful technology for differentiating cell lines. By culturing cells upon gold electrodes and passing through various AC currents, scientists are able to draw impedance measures whose behavior is unique to each cell line. What has not yet been introduced is a statistical "database" of the features which best identify each cell line. To this end, we explore cell line classification through the use of multivariate functional time series data obtained through an ECIS device. Namely, we investigate features of the time courses which act as the best identifiers of specific cell lines using classification trees, SVD, and other statistical learning techniques. With a heavy emphasis placed on the need for robustness to variation across experimental runs, we present our classification dictionary, and outline next steps for broader application of ECIS in biophysical experimentation.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association