Thursday, March 8, 2018
4:00 pm

Young Investigator Lecture Series-Medical Engineering

Machine Learning-based Microchip Platform for the Diagnosis of Disease
Jina Ko, PhD Candidate, Bioengineering, University of Pennsylvania

We have developed a diagnostic platform that combines magnetic nanopore sorting of bloodbased biomarkers (e.g. exosomes) and machine learning detection of the RNA signatures contained in the exosomes' cargo. My work solved two fundamental problems in exosome diagnostics: 1. We applied precision nanofluidic exosome sorting to crude clinical samples by inventing a design wherein millions of nanofluidic devices are incorporated onto a microchip platform and operated in parallel, increasing throughput by a million fold and eliminating susceptibility to clogging. 2. We overcame the variability of any individual biomarker between individual patients, by measuring panels of RNA and applying machine learning to identify signatures that persist across this variability. To demonstrate the power of this approach, we have focused on diagnosing two extremely challenging diseases, pancreatic cancer and traumatic brain injury (TBI). We accurately classified pancreatic cancer patients and TBI patients from healthy controls and more challenging classifications of specific disease states using mouse models (e.g. pre-cancerous lesions versus tumor, severity, history, and time post brain injury).

Biography: Jina Ko is a PhD candidate in Bioengineering at the University of Pennsylvania. She graduated in 2013 with a B.S. in Bioengineering and a B.A. in French Studies at Rice University. She is working in the Issadore lab that combines microelectronics, microfluidics, and nanomaterials to develop miniaturized platform for disease diagnosis. Her research has focused on the development of machine learning-based microchip diagnostics that can detect blood-based biomarkers to diagnose
two extremely challenging diseases, pancreatic cancer and traumatic brain injury (TBI). This work was accomplished using a multidisciplinary approach that combines microfluidics, molecular biology, bioinformatics, and machine learning. She accurately classified pancreatic cancer patients and TBI patients from healthy controls and more challenging classifications of specific disease states using mouse models (e.g. pre-cancerous lesions versus tumor, characterization of severity, history, and time post brain injury).

Contact Christine Garske ccgarske@caltech.edu
Add this event to my calendar