United States of America · Pittsburgh · See the United States of America profile
Since 2016 Allegheny County has used a predictive risk model to support child-maltreatment call screening. Evaluations found it narrowed racial gaps in investigation rates, but independent studies also flag bias — a transparent, contested case.
The Allegheny Family Screening Tool (AFST) helps call-screening staff in Allegheny County's child-protective services by integrating hundreds of administrative data elements into a 'Family Screening Score' predicting future child-welfare involvement. It has been in use since 2016.
Official evaluations found the tool decreased investigations of low-risk referrals, increased them for high-risk referrals, and reduced the racial gap in investigation rates without raising worker burden. Independent research has separately flagged racial bias in the tool's predicted decisions, which workers partly corrected by overriding the tool.
AFST is notable for being unusually transparent and externally evaluated, while remaining genuinely contested — a realistic picture of predictive risk models in high-stakes social services.
Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.
Where this practice's information was retrieved from, and when.
Singapore
Singapore's National Environment Agency runs a peer-reviewed machine-learning system that forecasts dengue cases up to three months ahead. In the …
Algeria
Algeria's Ministry of Higher Education used an AI matching platform nationwide for the first time in 2025, placing 97.34% of …
Argentina
In 2018 Salta province (Argentina) and Microsoft built a predictive-analytics algorithm claiming to flag named girls "predestined" for teen pregnancy …
Switzerland
SPSP integrates AI-powered genomic analysis to detect pathogen variants across Switzerland. Mandated by the Federal Office of Public Health since …
Open full copilot Grounded in cited practices — always check the sources.