This site has been optimized to work with modern browsers and does not fully support your version of Internet Explorer.

Evaluation of AI tool to identify patients at high risk of dementia

Summary

Routine clinical, laboratory and brain imaging tests are typically performed when older adults are admitted to hospital with the results recorded in the electronic patient record (EPR). To enhance care for those at risk of cognitive decline, a team at the University of Oxford is developing the Oxford Digital Biomarkers for Dementia (OxDBD), an AI-driven tool that leverages the EPR data to produce a risk score and identify those at high short-term risk of dementia. By enabling early intervention and personalised care plans, OxDBD aims to significantly improve patient outcomes.

Health Innovation Oxford and Thames Valley carried out a feasibility study and engaged with key stakeholders working in the dementia pathway to explore the potential utility, level of acceptance and barriers to adoption of the OxDBD risk score to identify older adults at risk of cognitive decline and dementia.

What is the challenge?

In England 70% of hospital bed days are taken up by adults aged over 65 following unplanned admissions. These patients are often severely unwell and are particularly vulnerable to deconditioning, leading to longer hospital stays and increased risk of complications. This situation places immense pressure on healthcare resources, straining hospital capacity and impacting the quality of patient care. The challenge lies in effectively managing this growing population while reducing extended periods of hospitalisation and unnecessary readmission, which not only affect patient outcomes but also burden an already overstretched healthcare system. Addressing this issue is critical for improving care of older patients and optimising hospital efficiency.

What did we do?

Health Innovation Oxford and Thames Valley carried out a comprehensive feasibility study to evaluate the acceptance and potential challenges of adopting the OxDBD risk score into EPRs. This AI-driven tool aims to identify patients at risk of cognitive decline using routinely collected clinical and imaging EPR data. To gather valuable insights, the study engaged 17 expert stakeholders, including GPs, acute care consultants, geriatricians, old age psychiatrists, managers and commissioner from 14 NHS organisations across England.

Through in-depth discussions and thematic analysis, the team identified key factors that could impact the tool’s implementation, such as clinical workflow integration, training needs and potential resistance to AI in diagnostics. This collaborative effort highlighted both the enthusiasm for innovative dementia care solutions and the practical barriers that must be addressed to ensure widespread adoption. The study provided crucial insights to refine OxDBD and tailor its rollout to meet the needs of healthcare professionals and patients alike.

What has been achieved?

Stakeholders have shown a positive interest in the potential of the OxDBD risk score in alerting them to patients at risk of cognitive decline who might benefit from evidence-based care plans tailored to individual risk profiles. Additionally, they noted that the OxDBD risk score may help raise awareness about dementia amongst non-specialist clinicians and prioritise patients for access to diagnosis and post-diagnosis support. The study also offered valuable insights into challenges relating to successful implementation. The insights gathered lay a foundation for future development efforts, guiding the path towards more effective and efficient care delivery for individuals at risk of dementia.

What people said

“The feasibility study has provided invaluable insights into the views of a wide range of stakeholders across the UK. This information not only supports the case for our AI tool, but, crucially, it will enable barriers to implementation to be identified and addressed early in tool development thereby substantially enhancing the likelihood of successful adoption and commercialisation.”

Sarah Pendlebury, Professor of Medicine and Old Age Neuroscience, University of Oxford, Consultant Physician and OUH Clinical Lead for Dementia/Delirium, Oxford University Hospitals

What next?

A study is currently ongoing at Oxford University Hospitals NHS Foundation Trust and Royal Berkshire NHS Foundation Trust to develop a digital algorithm for the OxDBD that will use data collected routinely during admission to identify patients at high short-term risk of dementia.

Contact: Florence Serres, Project Manager florence.serres@healthinnovationoxford.org