Predictive analytics

As the saying goes, there are 10 types of people in this world: those who understand binary, and those who don't (10 in binary is equivalent to two in base). Pressures are growing on frontline services, so commissioners are increasingly looking to data and intelligence to reach more children.

There's some technical terms we need to get out of the way before discussing the potential impact of this on children and families' outcomes.

• The term predictive analytics is used for assessing large quantities of information to see if there are trends. Those indicators are used to identify families who might be at risk.
• Big data means the huge amount of local intelligence that can now be collated and searched by computers. We can now include partner data and new information such as social media and school-level surveys.
• Risk and protective factors are identified by predictive analytics. For example, risk factors for a child in care becoming homeless include having a relative in prison. Protective factors would include good educational outcomes.
• Most people will have heard of artificial intelligence, or machine learning algorithms. This means the apps used to trawl through the data don't need to be told exactly what to look for, they can learn about connections between multiple indicators.

Predictive analytics represents the golden bullet of service design, offering the chance to know exactly what works, who to target, and how to help them earlier, in the following ways:

• New services - With the right data we can see what combination of factors (for example environmental, family, income, attainment, history) have the biggest impact on outcomes and the likelihood of needs escalating. Using these risk and protective factors we can change services to focus on the most impactful - a commissioning redesign.
• Who to help - Predictive analytics can provide us with a list of children and families who are likely to need help, enabling professionals to target their time in the most effective way.
• Intelligence - Up to now we have relied on methods such as the joint strategic needs analysis to inform local strategies. But this data is two or three years out of date and more often than not shows an escalating need in a geographic area that we cannot afford. Predictive analytics will, over time, give us more immediate, local and targeted management information.
• Time - The most important aspect might be the time it gives us to respond to need. Most complex need is hidden until it escalates and triggers an intervention. If we could predict need, then we might have years to respond with more cost-effective interventions, managing demand to expensive services.

This can feel a bit academic without hard examples. Around half of local authorities are now testing predictive analytics, but we have to look across the country or even internationally for proven beneficial outcomes.

• The Behavioural Insight Team recently tested whether computer analysis of social care case notes might identify children that will be re-referred after the case is closed. The team was able to spot half of these cases with a low level of errors. Here the implications are tantalising, both for better safeguarding and more efficient interventions.
• In Auckland, using health and care data, researchers were able to identify children most at risk of maltreatment by five years old. The accuracy was 50 per cent for the top decile most at risk. What is remarkable is that the data describes children under two, opening the potential to target support much earlier.
• In a shire authority early analysis of youth offending needs shows that it is possible to identify young people who are more likely to offend, so the time spent with young people based on their likely risk can be adjusted.
• Durham police is using artificial intelligence to predict risk of crime. The algorithms are trained on data from 2008 to 2013, with 88 per cent accuracy in predicting high-risk cases, and 98 per cent accurate for low-risk cases. This is used to inform bail applications.
• New York Fire Service has identified 60 risk factors to show which buildings are most at risk of fire and to target inspections. There are similar projects in the UK to identify failing schools or GP surgeries.
• To demonstrate the longer-term potential, Google has detected breast cancer from pathology images using artificial intelligence trained on high-resolution scans. The best professionals have an accuracy of 73 per cent, Google has an accuracy of 92 per cent.

It is early days for predictive analytics. Will it have a tangible impact on service delivery and on families' outcomes? What new markets might emerge for intelligence or early intervention? What are the ethical implications of predicting a crime or a child's needs? And will future commissioners speak in binary?

• Richard Selwyn is a member of the Association of Directors of Children's Services resources and sustainability policy committee @rjselwyn

FURTHER READING

Using Data Science in Policy, Behavioural Insights Team, December 2017

The Benefits of Predictive Analytics in Councils, University of Essex, August 2017

Detecting Cancer Metastases on Gigapixel Pathology Images, Cornell University, March 2017

Big Data in the Big Apple, Policy Exchange, June 2015

Can Administrative Data be Used to Identify Children at Risk of Adverse Outcomes? University of Auckland, September 2012