Social action innovations Bank

AFTS, An algorithm for risk of children neglect or abuse
Department of Human Services (DHS) Allegheny County, Pennsylvania

A Big Data System that provides social services professionals with an objective assessment of the situations of risk of child helplessness and helps them in the detection of cases and in the decision process of activation of the protocols for social intervention.

The Allegheny Family Screening Tool (AFST) is a tool based on specially designed algorithms that analyzes a large number of data from different sources of information. When social services receive information about a possible situation of helplessness or mistreatment, the algorithm calculates the risk index of the case by analyzing more than a hundred parameters such as criminal history, drug use, mental illness or History of child abuse of parents, guardians or people who live in the child’s home

The algorithm provides great accuracy in case detection and as a recent article in the New York Times concluded, “the Allegheny experience suggests that its screening tool is less bad at weighing biases than human screeners have been”.

Photo: DHS Allegheny County

Type of activity:

Instrumental improvement


Allegheny County, Pennsylvania, USA

Partners / Funders:

University of Southern California, Auckland, University of Technology, Nueva Zelanda


In 2015, the Allegheny County, after several very serious incidents that affected children in families that had been considered at low risk, initiated a project with two groups of highly recognized researchers to improve their system of selection and follow-up of cases. The project immediately focused on the process of deciding on the follow-up or discarding of calls to the hotline, which to that date had been carried out by people who, although they had access to the department’s database, could not perform an immediate exhaustive analysis which could trigger an action if necessary.

Implementation level:

The algorithm has been used by the DHS since August 2016 and it is continuously evaluated and reviewed in order to improve its accuracy.

Pin It on Pinterest

Share This