Select Page

Can Police Use Data Science to Prevent Deadly Encounters?

By Rob Mitchum // July 27, 2015

This summer’s Data Science for Social Good (DSSG) fellowshipencompasses a dozen projects focused on critical issues ranging from education to health to international development. But one project in particular feels very “ripped from the headlines” relevant: a partnership with the Charlotte-Mecklenburg Police Department (CMPD) and the White Houseto evaluate and improve early warning systems for police officer behavior. As Larry Greenemeier of Scientific American reports, many of these existing systems use crude statistics and methods to try to predict the risk of future adverse incidents involving police officers and direct high-risk officers to additional counseling and training. But through the White House Police Data Initiative, DSSG will work with CMPD to introduce new data sources and machine learning algorithms into the system, so that more accurate and context-rich predictions can be made.

Charlotte–Mecklenburg is one of the first police departments to formally commit to the initiative by agreeing to work with University of Chicago data scientists…The university will first analyze the department’s early intervention system as part of a summer data science fellowship program, Data Science for Social Good investigating early warning indicators for adverse police interactions with the public. The school’s Center for Data Science and Public Policy will continue this research when the fellowship program ends. Within the next few months the university expects to deliver a preliminary report to Charlotte–Mecklenburg with recommendations for upgrading its system, in place since 2005.

A DSSG project from last summer also recently received media attention, as part of a Chicago Tribune series looking at the continued struggle with lead poisoning in Chicago’s poorest communities. As funding for lead detection and mitigation decreases, public health departments seek new ways of identifying lead hazards before children are exposed, so that limited resources can be applied more effectively. Reporter Michael Hawthorne featured DSSG’s work with the Chicago Department of Public Health — which used public and blood test data to predict children at high risk of poisoning and buildings at high risk of containing dangerous amounts of lead — as one such program with the potential to reduce lead poisoning and its severe consequences.

The new computer model developed at the University of Chicago combines nearly two decades of childhood lead tests and home inspections with publicly available data on housing and demographics in Chicago neighborhoods. By aggregating massive amounts of disparate information, scientists hope to identify homes or apartments where pregnant women or children are most likely at risk for lead exposure.

“If you go in and fix lead hazards, you are going to protect kids in the future,” said Rayid Ghani, director of Data Science for Social Good, a U. of C. fellowship funded by Google Executive Chairman Eric Schmidt and his wife, Wendy. “It would be far better if we could protect the kids living in those homes now.”

Risk scores generated by the data project could help people find safer places to live or encourage families living in hazardous conditions to get their children tested and homes inspected. The scores also could be used to pressure landlords to eliminate lead hazards, a group of U. of C. researchers and city officials wrote in a summary of the project.

For more on the DSSG lead poisoning prevention project, see the video below: