1.jpeg (1)

YOSHI SODEOKA / The New York Times

February 3, 2020

Cornell Research Group Explores Potential of Machine Learning in Medicine

Print More

Medicine and artificial intelligence are ever-evolving fields at the forefront of scientific discovery. A new Cornell research group — Machine Learning in Medicine — aims to coalesce the two, with the goal of improving methods for disease detection and diagnosis.

This endeavor is a collaboration between faculty at Cornell Tech and Weill Cornell Medicine, bringing together “researchers with common interests and complementary expertise.” MLIM’s work is primarily an interdisciplinary dialogue, bridging campuses and research fields. 

“The idea was to link people with a machine learning background in Ithaca to [people working with] clinical data and hypotheses at Weill,” said Prof. Amy Kuceyeski, mathematics and radiology, one of the organizing members of the group.

While Kuceyeski’s background is in mathematics, she started learning methods for modeling biological systems as a postdoctoral researcher at Weill. Seeing this as an area for innovation, Kuceyeski helped establish MLIM in 2018.

“The brain is not so well understood, but it gives rise to everything around us,” Kuceyeski said. “That mystery pulled me in.”

This past year, the group studied the ability for computer-generated models to predict changes in the brains of subjects with and without autism — finding that the model was significantly worse in predicting changes in the brains of those with autism.

Currently, Kuceyeski is working with members of Prof. Mert Sabuncu’s, electrical and computer engineering, lab to predict brain activity using previous data sets of brain activity.

One recurring event that the MLIM research group holds is their virtual seminar series, typically twice a month. This series features medical professors and researchers in medicine from around the world to engage students from diverse academic backgrounds.

In this month’s installment, Prof. Konrad Kording, neuroscience, the University of Pennsylvania, presented his lecture “Is Most of Medical Machine Learning Wrong or Misleading?” In it, Konrad questioned some of the faulty methods in collecting and analyzing clinical data.

Looking ahead, MLIM plans to host seven more speakers in their virtual lecture series, including Prof. Danielle Bassett, bioengineering, University of Pennsylvania and Prof. Ben Glocker, engineering, Imperial College London.

The MLIM group also invites participation from students of both campuses. The virtual series presents a unique opportunity for students to interact with leading researchers from around the world from their own campus. In addition, MLIM will be hosting a symposium titled “Bridging the Divide: Machine Learning in Medicine,” tentatively set to take place in October in New York City.