NHS to transform operational decision-making with AI
NHS England & NHS Improvement are partnering with UK-based Faculty to help enhance NHS AI forecasting and predictive capabilities, so that data insights can underpin its operational decision-making. The new partnership puts learnings from the Covid-19 data response into practice across the NHS to improve services and care delivery for patients.
The Covid-19 pandemic has driven a high demand for healthcare services and is the largest health crisis that the NHS has ever faced. As part of the data response to the pandemic, the NHS has developed the Early Warning System (EWS).
This first-of-its-kind AI tool is based on Bayesian hierarchical modelling, using aggregate data (eg. Covid-19 positive case numbers, 111 calls and mobility data) to warn hospitals about potential spikes in cases so they can divert staff, beds and vital equipment needed.
There are currently around 1,000 users of the forecasts across the NHS, using the outputs for their own analysis and to help inform the prioritisation of safe care delivery for patients on a daily basis.
The response to the pandemic has demonstrated the importance and impact of using data and machine learning techniques to make predictions about the future and inform more effective decision-making. The techniques applied in the EWS allow forecasts to use specific information from that trust, as well as incorporating a broader set of contextual information that impacts the forecast, such as what’s happening in COVID-19 admissions in other local hospital trusts. This allows the EWS to incorporate much more relevant information into the forecasts for individual trusts than was previously possible.
The partnership with Faculty is designed to help the NHS capitalise on its success to date using AI and enable their team to build and deepen their capabilities. In particular, the NHS will address new areas where AI forecasting tools can improve service delivery and patient care, such as understanding and predicting A&E demand and winter pressures.
Specialising in applied AI, Faculty was appointed as a partner following a tender process and the company will support the NHS’ in-house data science teams to build machine learning tools that leverage Faculty’s pre-built technology applications – such as its AI safety and explainability tools – and combine this with the deep expertise that is required to integrate AI forecasting successfully into the organisation. The company will also support the NHS team to get the most value out of these advanced AI tools and train NHS analysts as part of the partnership.
Faculty has also supported NHSX as partner for the NHS AI Lab, where it most recently helped develop the National Covid-19 Chest Imaging Database (NCCID). This database of over 40,000 images is helping to improve the diagnosis of COVID-19 symptoms and enabling fast, accurate, early diagnoses that can be critical for treating patients and saving lives.
AI safety and model validation software will continue to feature in the AI tools developed to support operational decision makers. Faculty’s latest explainability techniques help users to understand and interpret how each input, such as past hospital admissions or local testing data, influences the outcome of the forecast.
For example, users can see how much recent historical admissions data for a particular trust is driving that trust’s forecast, versus how much local testing data is influencing it. This ensures even sophisticated machine learning models are interpretable by everyday users in the NHS, helping them to make more informed decisions and driving adoption and trust in the tool.
Covid-19 will likely continue to be a challenge for healthcare services, especially as winter approaches. This new partnership therefore also provides continued support with the running and usage of the current EWS tool alongside the NHS team.
Faculty will assist with machine learning operations (MLOps) ensuring these models remain safe and high-performing. This is hugely important as AI software, unlike conventional software, is trained on constantly evolving real-world data, meaning the models need to be maintained and retrained to continue to provide the NHS with the most accurate forecasts.
The newly developed and existing AI forecasting tools will all be delivered through the Faculty Platform – Faculty’s best-in-class, secure, open data science environment for building and deploying AI software.