Trending Topics

Indiana DOC uses software to reduce prison assaults

IDOC implemented the predictive analytic tech following a series of severe assaults on staff


Since implementing the technology across IDOC’s 18 facilities, including Plainfield Correctional, staff assaults and inmate-on-inmate assaults have both dropped significantly.


By Julia Edinger
Government Technology

INDIANAPOLIS, Ind. —The Indiana Department of Correction (IDOC) has addressed an ongoing issue of physical assaults within the facilities by implementing predictive analytics software.

In order to safeguard staff of correctional facilities, IDOC leadership decided to implement predictive analytic technology by SAS following a series of severe assaults on staff members in the state’s correctional facilities, said Sarah Schelle, IDOC’s executive director of data science and analytics.

“That was really what led us down this path: trying to actively and proactively intervene in these types of scenarios,” she said.

Since implementing the technology across IDOC’s 18 facilities, staff assaults and inmate-on-inmate assaults have both dropped significantly: 50% and 20%, respectively.

Furthermore, since revising their model in October 2020, Schelle noted an additional 12% reduction in overall assaults.

She added that the new model incorporated much of the information from the previous one but includes a greater focus on housing assignment.

As justice-involved populations director at SAS, Mary Beth Carroll, explained, the use of predictive analytics allows a facility to look at the variables within a set of data that may directly impact a person’s behavior.

These variables could include whether inmates are receiving mental health services, taking medication or have had a recent change in employment. Other variables can be added as needed to include any piece of data in IDOC’s data system.

Schelle explained that the information can be viewed through a dashboard, which will categorize people based on whether they are very high, high, moderate or low risk. The risk assessment information is color coded to simplify differentiation, she said, noting that the top 10 percent are considered to be very high risk.

Staff can look at this data for the entire population of a facility, within specific buildings or even with individuals.

The information in state systems can be siloed, Schelle said, but using analytics allows the data to inform staff, who can use the tool to acquire information that can help them make decisions when somebody is deemed to be at a high risk or very high risk. That could mean anything from changing somebody’s housing to offering an ear.

“Sometimes people just need somebody to talk to and vent,” Carroll explained, stating that being able to identify those at risk and provide that communication outlet can be a helpful intervention.

Cost savings have been another advantage of using predictive analytics in corrections facilities, Carroll said. Specifically, she said health-care costs and workers’ compensation costs are reduced when inmates are not getting treatment for assaults.

Schelle explained that there have been an even greater reduction of staff assaults in what she referred to as “high-adopting facilities.” Various facilities within the IDOC system may use predictive analytics in different ways. High-adopting facilities have more staff available to utilize the tool and review the information provided with a multidisciplinary panel to understand patterns or make more informed decisions.

When looking at making predictive models to address different operational needs, Schelle said it is important to use them specifically to predict what they are intended to predict — for example, she explained, a model for predicting violence should not be used for predicting contraband.

The model is rescored every week, Schelle said, and it can be changed to better suit the organization’s needs.

(c)2021 Government Technology