BETHLEHM, Pa. (AP) — In a typical week, the state Department of Corrections receives a list of 1,000 inmates who need to be assigned to one of its 25 correctional facilities.
The process isn’t straightforward. There are nearly 100 factors that determine where an inmate ends up — from the medical care they require and their age to their family’s proximity to the prison and which programs they need to become eligible for parole.
It used to take seven corrections employees a week to figure out where they would go.
Now, with the push of a button, an algorithm designed by a group of Lehigh University engineering students and their professors assigns the same number of inmates in 10 minutes.
Most Read Nation & World Stories
- Hundreds of experts say coronavirus is airborne; consequences for our daily lives would be significant
- South Dakota governor, exposed to virus, joined Trump on jet
- Militias flock to Gettysburg to foil supposed flag burning
- How Fauci, 5 other health specialists deal with COVID-19 risks in their everyday lives
- Study uncovers most effective non-medical face mask for protecting against coronavirus
The new system not only is expected to save the prison system nearly $3 million a year but also has landed the Lehigh team in contention for the international Daniel H. Wagner Prize for Excellence in Operations Research.
“The Lehigh model looks at everything simultaneously and holistically. It makes the most appropriate recommendation for everybody based on the resources that are available at that time,” said William Nicklow, a major at the State Correctional Institution in Camp Hill, Cumberland County.
Nicklow ??oversaw the implementation of the Inmate Assignment Decision Support System over the past 10 months.
The program was used this summer to reassign 2,000 inmates when the State Correctional Institution in Pittsburgh was closed as part of Gov. Tom Wolf’s plan to stem the state’s deficit and to address a nearly 4 percent drop in inmate population in recent years.
That process was anticipated to take six months, but with the new program it took 4½ months, Nicklow said.
This new efficiency is expected to save the state $2.9 million a year by making sure inmates are in the correct facility from the start, thereby reducing the number of prison transfers, and even some inmates’ prison stays by ensuring they get the programs they need in a timely manner.
Since the state began using the model 10 months ago, the time for prisoners waiting to start programs necessary for parole has been reduced by 56 days, Nicklow said.
Lehigh began designing the program five years ago, when state officials put out a call for a better population management system. State Secretary of Corrections John Wetzel saw there were a high number of younger inmates at one facility, which correlated to an increase in fights and other infractions at the prison.
“Secretary Wetzel saw a need for us to make some better decisions on inmate assignment based on security, age demographics, medical needs, programmatic needs and educational needs,” Nicklow said.
Wetzel asked Bret Bucklen, the Corrections Department’s director of planning, research and statistics, to look into the problem. Bucklen put out the call to companies and universities.
As it turned out, Lehigh student Dan Li was doing an internship with the Department of Corrections and developed a decision tree identifying factors involved in inmate assigning. After Li earned her doctoral degree in 2013, the project was taken up by PhD candidate Mohammad Shahabsafa.
In addition to Li and Shahabsafa, the group that worked on the project includes Lou Plebani and George Wilson, both associate professors of industrial and systems engineering; Anshul Sharma, a graduate student; and Chatainya Gudapti, who is earning a master’s degree. The group was led by professor Tamas Terlaky, chairman of Lehigh’s department of industrial and systems engineering.
A graphic user interface allows access to information on inmates in the Department of Corrections database, enables users to review and approve the optimal assignment and provides several measures to evaluate assignment recommendations. The system uses the same optimization model used by the airline industry to assign pilots, passengers and other staff to flights, Terlaky said.
While such algorithms have been used in other industries, Terlaky said it’s unusual for the Department of Corrections because it’s not a field that typically employs engineers.
“At this point it’s really revolutionary and unique. No other state is doing anything like this,” said Terlaky, who served as team leader.
Terlaky and his students and colleagues are finalists for the Wagner prize, awarded each year by the Institute for Operations Research and Management Sciences, a professional association for analytics and operations research.
This year’s prize will be awarded Oct. 24. Contenders include groups from Georgia Tech, Brown University, and the Massachusetts Institute of Technology, Terlaky said.
Terlaky said his group hasn’t yet patented its program, but plans to look into it. It also plans to demonstrate the program for other correctional departments during the annual American Correctional Association conference in Orlando, Fla., in January.
Some of the students who worked on the project might consider creating their own company to sell the program, for which Nicklow said the state paid $300,000. The money was used as a grant to help the students develop the program, he said.
“It’s certainly been a huge return on investment for us,” Nicklow said.
Information from: The Morning Call, http://www.mcall.com