Paschalidis Shares Health Data Findings in DeLisi Lecture

Professor Yannis Paschalidis (ECE, BME, SE) discussed data-driven reasoning—which he calls “the backbone of engineering systems”—and predictive health analytics as he delivered the Charles DeLisi Distinguished Lecture May 6 to an online audience of about 100 members of the Boston University community.

The DeLisi Award and Lecture honors a senior faculty member engaged in outstanding scholarship. Paschalidis has used data to improve efficiency in transportation systems, supply chains, and other areas for a quarter of a century, and in recent years, he has applied his expertise to the health care system. In his DeLisi lecture, titled “Data Science and Optimization Adventures in Computational Biology and Medicine,” he shared highlights of his wide-ranging, high-impact research, including a new study on developing models to predict hospitalizations for COVID-19.

Paschalidis explained that a kind of data-driven reasoning occurs even at the cellular level. “We can think of the cell as a chemical factory,” he said. “Within the cell, many chemical reactions take place, and these reactions produce all the molecules needed to sustain life—so for the cell to multiply and thrive.”

As that tiny factory, “the cell has to make decisions about what items and the quantities of those items that need to be produced,” said Paschalidis, who with colleagues has managed to reverse-engineer that production process in order to predict metabolic activity. Insights gained thereby “may be interesting if we want to change this chemical factory for our own purposes,” Paschalidis said, “and today we have that ability.”

Building on his previous work on diabetes and other chronic diseases, Paschalidis in the past year has collaborated on several studies around the globe predicting outcomes for COVID-19 patients. One of them hits quite close to home: a study of all 7,102 patients who tested positive for the disease at Boston Medical Center, the teaching hospital for the BU School of Medicine, in 2020.

“What’s interesting about this data set is that it includes a lot of information about social determinants of health,” Paschalidis said. “BMC patients are screened for their need for housing, need for food, for transportation, education, their ability to pay for utilities, and other social needs.”

The study showed quantitatively that these variables are strong predictors of a COVID-positive patient’s likelihood of being hospitalized, admitted to intensive care, and placed on a mechanical ventilator. It also shows their chances of dying are higher. “The mortality model is extremely accurate,” said Paschalidis.

The DeLisi lecture has been held annually since 2011 (Paschalidis’ talk was postponed from spring 2020). Recipients are nominated by their colleagues for extraordinary records of well-cited scholarship. A panel reviews nominees, and Dean Kenneth Lutchen names the awardee.

Paschalidis, the director of the Center for Information and Systems Engineering, has authored more than 220 peer-reviewed studies in systems and control, optimization, machine learning, networks, computational biology, and computational medicine. He has also given more than 130 invited lectures, served as the primary investigator or co-PI on more than $43 million in grants, and advised 26 dissertations. He is an Institute of Electrical and Electronics Engineers (IEEE) Fellow, founding editor-in-chief of IEEE Transactions on Control of Network Systems, and a past National Science Foundation CAREER award winner.

The DeLisi Award is named for Dean Emeritus and Metcalf Professor of Science and Engineering Charles DeLisi (BME). A pioneer in theoretical and mathematical immunology, DeLisi led the Human Genome Project and founded and chaired BU’s Bioinformatics Program, the first such program in the nation. He currently directs the Biomolecular Systems Laboratory.

A video of the lecture is available on the BU College of Engineering YouTube channel.