Indiana Department of Child Services Predictive Analytics Demonstration Project: Placement Matching - The Indiana Department of Child Services (DCS) proposes a three-year predictive analytics demonstration project to modernize placement decision-making and strengthen agency data systems. The initiative will help improve child welfare outcomes and serve as a model for national replication of predictive analytics approaches. Indiana’s supports about 13,000 Child in Need of Services (CHINS) monthly, many of whom may require stable, well-matched placements. Current placement practices vary across regions, rely heavily on staff judgment, and are limited by inconsistent and unstructured data. These gaps slow decision-making and contribute to placement disruptions. To address this, DCS will build and test a predictive placement matching model that uses historical administrative data and newly structured information to identify strong child–caregiver matches. The model will analyze child characteristics, placement history, caregiver attributes, and compatibility factors using machine learning techniques, with safeguards for fairness, accuracy, and bias mitigation. DCS will also create a bi-directional web interface that displays match scores, explains contributing factors, and supports both staff and caregiver decision making while preserving professional judgment. The project includes major data quality improvements, such as converting narrative information into structured data, auditing data elements, and enhancing pipelines and privacy controls. Temporary data stewards will support this effort. To build long-term capacity, analytic staff will receive training in predictive methods, data visualization, and model evaluation. Part-time research assistants will expand the team’s bandwidth through data cleaning, literature reviews, simple analyses, and visualization tasks, while helping create a future workforce pipeline. The project will explore questions related to variables influencing placement stability, caregiver–child compatibility, drivers of disruptions, and integration of predictive tools into practice. Expected outcomes include faster placements, greater stability, fewer disruptions, reduced use of congregate care, and better caregiver retention—advancing state and federal priorities for improved permanency outcomes.
