Community Data Roundtable (CDR) is proud to announce its first peer-reviewed publication. Appearing online on April 24, 2021 the article is a scientifically structured presentation of some of CDR’s most cutting-edge data analytic techniques.
The article is appearing in Children and Youth Services Review, a monthly peer-reviewed journal with international reach. “We wanted to publish in a journal that was already publishing cutting edge work on social service data analytics” said executive director, Dan Warner, Ph.D.
“Community Data Roundtable aspires to use a scientific and data driven approach in all that we do,” continued Dr. Warner. “We are excited to publish our work alongside the work of the most respected researchers in the child serving field.”
The article describes a technique for analyzing Child and Adolescent Needs and Strengths (CANS) data, using advanced ‘machine learning’ (aka ‘artificial intelligence’) techniques. The CANS is a form used for treatment planning, assessment, and outcomes tracking in child services, especially mental health services.
“We’re basically having a computer look at a large sample of CANS data at one time point, and having it predict what it will look like at another point,” explains Ryan Torrie, a Silicon Valley product manager by day, and CDR volunteer who did most of the high-end data crunching for the project. “Because the CANS has so many data points, a computer is necessary to find meaningful relationships between data points.”
Many important areas in child services are impacted by the study’s findings. First, the publication further supports the use of CANS for outcomes tracking. “Some people assume that a treatment planning and assessment document can’t also be used to track outcomes. Our data are not the first to refute that supposition, but this is certainly one more example of how good the CANS is in measuring—and predicting—meaningful change.”
This leads to the second important contribution from the article: it demonstrates that with appropriate analytic techniques, CANS profiles for success or failure can be identified. “This means we can have a brand-new child enter the system and receive assessment, and right there in real time know their prospect for success in our system. This helps make better decisions for that child, that are individualized for their unique needs and prospects.”
The article’s last important contribution is specific to Pennsylvania, which is where the data came from. “CANS data is always local,” Dr. Warner explains, “The profiles and results we found from Pennsylvania’s in-home Behavioral Health Rehabilitation Services speak uniquely to that program. We are excited to share some interesting findings from the analysis, especially those relevant to children on the autism spectrum.”
“Community-based autism care, like applied behavioral analysis, is getting a lot of attention in Pennsylvania,” explains Jesse Troy Ph.D., the author of the article, and CDR’s chief data scientist. “Pennsylvania has one of the largest-community based autism intervention programs in the country, and very little is known about whom it works for, and under what conditions. This project helps shed some light on who the program is working for.”
The full article can be accessed here: A machine learning approach for identifying predictors of success in a Medicaid-funded, community-based behavioral health program using the Child and Adolescent Needs and Strengths (CANS)