FORTH – FORecasting Turbulence in Hospitals

Lay Description

UK health trusts declared 141 critical incidents between December-2022 and October-2023. These cause increased last-minute cancellations of operations and appointments, delayed discharges, overcrowding, staff workload pressure, and worse quality-of-care.

Variability and demand surges for acute hospital services contribute to pressures creating slower and more chaotic patient flow, a phenomenon we call turbulence, which impacts the hospital’s ability to function, leading to crisis periods and operational incidents. Forecasting turbulence with sufficient time for proactive resource management can mitigate its effects by increasing predictability and stability of patient flows.

We will develop mathematical models to ‘predict and alert’ approaching turbulence affecting resource capacity within acute care hospitals, that will inform proactive resource management strategies for improving efficiency and avoiding critical incidents.

To do that, we will first predict the next steps in a patient’s pathway given the previous events registered in their current hospital spell. Then, by combining the prediction for current and prospective patients, we can infer the hospital resource utilisation in the short-term future and thereby identify any approaching turbulence, thus enabling hospital management to implement proactive demand management policies to prevent and mitigate turbulence.

This is a collaborative project involving the University of Southampton (lead Dr Edilson Arruda), University Hospital Southampton-UHS (Dr Mark Wright), Salisbury NHS Foundation Trust-SGH (Dr Alexandra Hogan) and funded by ARC Wessex. We will disseminate results to key stakeholders at UHS and SGH and to the wider southeast and southwest regions including the Bath and Somerset, Swindon and Wiltshire ICB. To maximise dissemination regionally, we will organise two workshops. We will invite leaders, planners and practitioners from UK’s NHS trusts in the south. Wider dissemination includes journal publications and conference presentations.

Public Benefit Statement

Current models often focus on simulating at hospital level but do not properly account for the variation in patient pathways which limit their applicability. To bridge this gap, we will apply data analytics and mathematical modelling to understand the variation in patient pathways and predict, at the patient level, the future needs in terms of resources and services at each point in time.

This will enable us to develop models to predict short term surges in the demand for specific services by combining the predictions for current and prospective patients over a short-term horizon. Such predictions will then support the development of proactive management strategies to prevent and mitigate the impact of these surges in the quality of the service provided to patients.