Work
The power of data to keep people well at home, and avoid the hospital
Routine data collected by care workers can help predict, and in many cases avoid, hospital admissions. We worked with Cera Care to explore how machine learning fed with this data could help reduce pressure on hospitals while keeping people well in their homes.
80%
of hospitalisations predicted
1-7 days in advance
2.6x
more accurate hospitalisation
estimates versus clinicians
50%
up to half of these potentially
preventable with low-cost
interventions
Background
Cera Care is Europe’s largest provider of digital-first domiciliary care, delivering 50,000 visits to people in their homes each day. Every time carers visit a service user they fill in a report answering questions about their wellbeing using the Cera Care app.
At Faculty, we explored whether data captured for observation purposes could be used in more innovative ways to help predict which service users might soon end up in the hospital, enabling proactive interventions to help prevent avoidable hospitalisations.
Solution
We created a machine-learning algorithm that produces a prioritised list of patients for proactive intervention based on their hospitalisation risk. Cera Care’s registered nurses can then work down this list from highest risk to lowest and arrange low-cost, preventative interventions. These include GP telephone consultations, district nurse visits, or pharmacist medication reviews, helping to avoid admission for many service users. Predicting hospitalisation and recommending a mitigating intervention helps to reduce the unnecessary use of ambulance and emergency services. As well as reduce the severity of patient conditions and the level of nursing care that might be needed after discharge.
Impact
The algorithm, which can predict 80% of hospital admissions up to a week in advance, helps cut down on hospitalisations while also reducing unnecessary strain on a stretched NHS. This technology is enabling Cera Care to quantify, to great effect, the single best health observation measure at predicting hospitalisation – human intuition.
“Working with Faculty has shown the power of artificial intelligence and data in supporting the NHS and community, keeping patients safe at home and reducing hospitalisations. Using data in innovative ways and empowering our network through machine learning and additional technologies will be critical to the future of healthcare.”
Nathan Windle, Director of Data, Cera
More of our work:
COVID-19: Helping the NHS to save lives by forecasting patient demand
Faculty drives millions in profit for a global retailer with machine learning
How Hywel Dda University Health Board optimised capacity and cut costs with Frontier
Automating media monitoring for a leading global charity
Developing an Intelligent Gas Grid to reduce carbon emissions
