A data-driven approach for building energy efficiency of healthcare facilities in England and Wales using regulatory building energy datasets

Healthcare buildings are among the most energy-intensive and carbon-emitting building types. In the UK, healthcare buildings account for approximately 15% of the total carbon footprint of the National Health Service (NHS). Reducing energy consumption and carbon emissions in existing healthcare buildings through energy-efficient refurbishment is therefore a key strategy for achieving the NHS Net Zero targets. Data-driven approaches have been widely applied for the rapid and accurate assessment of building energy performance. However, their application at the building stock level remains relatively limited. This study investigates a data-driven approach for energy performance prediction and assessment, with a focus on primary healthcare facilities in England and Wales. A surrogate - residual machine learning model was developed based on regulatory building energy datasets and evaluated against measured energy consumption data. The results demonstrate robust predictive performance of annual gas, electricity and total energy consumption across the primary healthcare building stock. In addition, the model can be used for evaluating the energy saving potential of refurbishment measures for the entire stock. The proposed data-driven approach provides a computationally efficient solution that enables rapid energy performance evaluation across hundreds or thousands of buildings and multiple refurbishment scenarios by updating the values of input features to represent the intended measures. As such, it offers a scalable decision-support tool for large-scale healthcare building refurbishment planning and the implementation of strategies aligned with the NHS Net Zero targets.