BGSU partners with BCI to speed analysis of sexual assault kits
BGSU participants in the Sexual Assault Kit Analysis project are (left to right) D.J. Heckman, Dr. James Albert and Dr. Jon Sprague.
A special initiative currently underway in the Ohio Attorney General’s office to analyze thousands of previously untested sexual assault kits (SAKs) has helped make the Ohio Bureau of Criminal Investigation (BCI) a leader in processing SAKs for DNA evidence. BCI has made great progress in testing these kits as quickly and efficiently as possible, and now, a nearly $440,000 grant from the Laura and John Arnold Foundation to Bowling Green State University and the BCI will help identify ways to streamline the analytical process even more, BGSU President Mary Ellen Mazey and Ohio Attorney General Mike DeWine have announced.
Leading the joint project, titled “Use of Statistical Modeling to Optimize Sexual Assault Kit Analysis,” is Dr. Jon Sprague, director of the Ohio Attorney General’s Center for the Future of Forensic Science at BGSU. The center is a joint endeavor between the University and the BCI facility located on its campus, and promotes collaboration between the two.
Co-investigators on the project are BGSU’s Dr. James Albert, a professor of statistics, and the BCI’s Dr. Lewis Maddox, DNA technical leader. Also on the project are D.J. Heckman, a first-year master’s student in statistical analysis, and a BCI data analyst working at the agency’s Richfield, Ohio, location.
The grant will enable the cross-disciplinary team to use data mining of the results of the over 11,000 SAKs processed by the BCI to date to develop best-practice models for SAK analysis.
The kits may include such items as clothing, hair and samples of bodily fluids, for example. “Where should you start your processing?” Sprague said. “By ‘moneyballing’ the data we have collected, we will establish protocols that will improve processing time on each SAK. We’ll use statistical modeling to drive our decision making and if we can reduce the processing time by even as little as a few hours per kit, if you multiply that by the 10,000 kits, it will help tremendously.”
Maddox said that although the BCI analyzes everything in the kits, the information resulting from the analysis will help shape the initial sequence, and he expects there to be a cost savings associated with greater efficiency. He also anticipates crime-solving benefits from the DNA profiles that can be added to the national database maintained by the FBI.
Ohio is ahead of other states on the issue as a result of Ohio Attorney General Mike DeWine’s SAK Testing Initiative. When DeWine took office in 2011, he learned that law enforcement agencies across Ohio had thousands of kits that had never been tested because they had not been sent to a crime lab. In response, he requested that all unprocessed SAKs related to a crime be turned over to the BCI for testing at no cost to local law enforcement. In total, the SAK Testing Initiative has resulted in over 13,000 SAKs coming to the BCI for analysis.
The problem is nationwide. In 2014, the U.S. Department of Justice estimated that across the country some 400,000 kits had been collected but were never analyzed because they had not been delivered to forensic science laboratories for processing.
Ohio Senate Bill 316, which went into effect on March 23, 2015, required Ohio law enforcement agencies to submit any remaining previously untested sexual assault kits associated with a past crime to a crime laboratory within one year. The law also requires that all newly collected rape kits be submitted to a crime lab within 30 days after law enforcement determines a crime has been committed.
As a result of the SAK Testing Initiative, “We knew we were going to be collecting a lot of data,” Maddox said. “And now that we’ve collected these data, we saw an opportunity to partner with the center for analysis. One of the things at the top of our list is defining the criteria for what we want to analyze.”
Once that has been determined, the BCI data analyst will pull information using the enormous database of information, called the Laboratory Information Management System, to give to Heckman, who will conduct the analysis, overseen by Albert.
“We call this statistical learning or machine learning, where we have a lot of variables and we look for patterns and the relationships between the variables,” Albert said. “This is very exciting because there are a lot of opportunities to get science more involved.”
He and Heckman will visit the Richfield data analyst this summer to learn more about the database before Heckman begins his work.
“Once we’ve established a data protocol for the first pass on processing the SAKs, we’d like to eventually partner with another state organization to test our results to see if they work,” Maddox said.