Predictive software has been assessing data on the lives of thousands of children, from their potential exploitation by gangs to their risk of not being ready for primary school.
The technology company IBM has been working with Brent council to try to predict which children were at risk of gang exploitation, while Essex county council has profiled all of the children living in one of the wards of Basildon to try to identify those that might be unable to read or write.
On Sunday, the Guardian revealed how local authorities have been using machine learning and predictive technologies to intervene before children were referred to social services.
However, the programmes being run by Brent and Essex illustrate how advocates of predictive analytics believe the technology can be adapted for a far broader array of purposes. Brent council identifies “gangs, CSE [child sexual exploitation], missing, education and youth offending” as priorities for its predictive system.
The data of 12,000 Brent children, who were already known to the council, was incorporated into its predictive model, according to a council document, which went on to claim that the system “illustrated evidence to suggest we can start identifying children at risk of criminal exploitation and offending well before the outcome occurred”.
Brent uses data from schools, social care and gang area intelligence in order to construct its system. A scrutiny committee memo claimed that information gathered from external sources such as the Home Office and the NHS was also used in the model,though the council said such data had ultimately not been used.
The model identified children presenting multiple risk factors drawn from research by the Children’s Commissioner into gang exploitation. These include school exclusions, poor mental health and drug and alcohol misuse.
Brent council began working with IBM in 2016, after a prototype of the system won an award at an international criminal intelligence conference. The prototype, which scored children living in Brent based on the presence of factors deemed likely to result in child sexual exploitation, was based on the borough’s entire resident child population of 70,000.
The scrutiny committee memo suggested that the system could be used to generate income for the council. A spokesperson said it was possible that Brent would sell the finished software to other councils once trials were completed, but stressed that no data would change hands.
Discussions about the programme have been held with departments within the Metropolitan police, including the counter-terrorism unit, according to the memo. Brent council said no data had been shared with partner agencies and that any safeguarding decisions were made by social workers.
“This is used to assist council staff with their risk judgments and decision-making processes. It does not replace the existing decision making and judgments but seeks to enhance the knowledge and response to known risk factors,” a spokesperson said.
Separately, Essex county council has received more than £3m from the Ministry of Housing, Communities and Local Government for a pilot project aimed at improving school readiness in Vange, a ward in Basildon with a population of about 10,000.
The programme, which has been developed by an external contractor, incorporated housing and benefits data, children’s social care data, youth offending data and police and crime data, as well as “school readiness indicators” from council data sets. Ethnicity was one of the indicators included.
All data has been pseudonymised, which means that anyone included in the model has their name replaced with a unique identifier such as a string of digits. A protocol specifying how information was to be shared said that those included would not be able to request their own information under the Data Protection Act on the grounds that pseudonymised data was no longer personal data.
Essex county council said the data on 7,000 children had been used to build the predictive model, and would be used to predict those who by the age of five might be at risk of deficiency in areas including speech, language, social interaction and being able to use a toilet.
A council spokesperson said existing data showed that a high number of five-year-olds in Basildon were not ready to go to school in 2013. “This is important, because Ofsted reports that few children who start school behind are able to catch up before they leave full-time education at age 18.”
The spokesperson said Vange had been identified as a high-risk community, where one in four children were not school-ready. It said a pilot programme identified 511 households at risk, with 280 of those households not already known to the council or police.
