Predicting inpatient flow and improving experiences with machine learning algorithms Surrey and Borders Partnership NHS Foundation Trust

The pressures on demand and capacity in NHS hospitals is well documented. Being able to forecast patient demand patterns and match capacity to that demand could help ease this pressure.

This project from a team at Surrey and Borders Partnership NHS Foundation Trust will use machine learning to develop a digital dashboard to predict patient flow and length of stay (LoS) at the trust.

The project team will use the trust’s historical records to provide accurate predictions of LoS of inpatients. This could improve care by highlighting discharge barriers or identifying complex cases; and improve resource management and planning.

The digital dashboard will provide clinicians, front-line staff and decision makers with the opportunity to see what is happening in real time. Predictive analytic visualisations will provide data on the expected LoS a patient might need, given their presentation and clinical history; where the patient will be discharged to; and the likelihood of them being stepped-up or stepped-down in level of care.

Accurately predicting this should greatly improve the patient experience, and reduce readmission rates and the risk of over-staying. Having a predictable bed flow should reduce costs, increase patient flow, reduce LoS and allow the trust to minimise ‘out of area’ placements (when a person with acute mental health needs who requires inpatient care is admitted to a unit that is not part of the usual local network of services).

The team will embed the dashboard within the service by educating and empowering end-users, with the aim of increasing engagement with the dashboard and analytics in general.

Contact

For more information about this project, please contact Mayoor Dhokia, Data Science Lead, Surrey and Borders Partnership NHS Foundation Trust.