Data science will play an increasingly important role in solving the widespread public health challenges of the 21st century–‒that is the key takeaway from Columbia University’s Data Science and Public Health Summit on January 17.
The summit, which was produced by the Mailman School of Public Health in partnership with the Data Science Institute (DSI), attracted statisticians, mathematicians, biostatisticians, physicists, sociologists, and political scientists from across the nation. They gathered to discuss how to use advanced data science techniques, including machine learning, AI, and deep learning, to mitigate population health challenges such as obesity, diabetes, and disease outbreaks. Speakers also stressed the importance of using data in ethical ways that safeguard everyone’s privacy.
Linda Fried, dean of the Mailman School, noted that machine learning and AI have sometimes introduced “worrisome biases that, rather than advancing humanity, have accentuated the health disparities we spend so much time trying to combat.”
“[We] must provide the ethical guardrails to the field and protect all populations through just approaches,” Fried added.
Jeff Goldsmith, an associate professor of biostatistics at the Mailman School and a DSI member, detailed how gender bias crept into Google Translate in 2015. Initially, he said, the translator was heralded for its efficiency in translating languages faster than anyone thought possible. It soon became apparent, though, that the translator was gender biased, with egregious translations such as “he is a soldier; she is a teacher; he is a doctor; she is a nurse.”
“When you train an algorithm in a way that seems to give good answers, but you really don’t understand it, you can reproduce biases, which is what happened here,” Goldsmith said. He also suggests that data scientists develop a shared ethics of practice to guide their research and teaching.
Jeannette M. Wing, DSI’s Avanessians director, agrees and says the institute is at the forefront of developing ethical data practices. “You can sum up [DSI’s] mission with three words—data for good,” she said.
Wing also highlighted four current collaborations with Mailman School faculty members, two of which are research-based and two will create new, data-driven, public health courses at Columbia. “Both DSI and Mailman teach students the importance of using data responsibly for the benefit of society,” she said.
— Robert Florida