Electronic health records contain a rich history of the patient journey with huge potential for improving direct patient care, service delivery and research.
However, electronic health records can often be difficult to access and contain incomplete or unstructured information such as large amounts of text, which is challenging and time-consuming to analyse.
Over 80% of NHS patient data is locked in unstructured, siloed records (e.g., clinical notes, letters). Manual extraction is slow, costly, and error-prone, delaying insights and burdening staff.
How are we meeting this challenge?
The National Institute for Health and Care Research (NIHR) Maudsley Biomedical Research Centre has led a collaboration with key partners including UCLH Biomedical Research Centre (BRC) and King’s College Hospital to develop CogStack, a search engine, analytics and visualisation tool, to unlock important data from electronic health records, with alerts to support clinical decisions and healthcare research.
CogStack can search any clinical data and, using our artificial intelligence (AI) and natural language processing (NLP), it can assign codes based on particular words, diagnoses or treatments.
This short video describes how CogStack implements best-of-breed enterprise search, natural language processing, analytics and visualisation technologies to unlock the health record and assist in clinical decision making and individual direct patient care.
CogStack integrates with hospital systems such as Epic in South London and Maudsley and national initiatives like NHS Sub-National Safe Data Environments, enabling multiple benefits to:
- Patients: Fewer missed diagnoses, safer care pathways (e.g., reduced “lost-to-follow-up” incidents).
- Clinicians: Reduced administrative workload, actionable insights via alerts/dashboards.
- NHS: Cost savings (e.g., streamlined workflows, and interoperable data for research.
- Researchers: Access to structured, searchable datasets (e.g., >1B+ SNOMED annotations from SLAM records).
Collaboration and achievement
CogStack, developed by the NIHR Maudsley and UCLH BRCs with GSTT, KCH, UCL and KCL, leverages AI and NLP to harmonize unstructured data across the health system. It has collaborated with NIHR Applied Research Collaboratives (ARCs), NHS AI Lab, and NIHR Mental Health Mission/TRC to accelerate trial recruitment, for example its use led to recruitment that was 8x faster for COVID-CNS project.
The freely accessible CogStack platform has been deployed in local NHS Trusts at King’s College Hospital, South London and Maudsley NHS Foundation Trust, Guy’s and St Thomas’, and University College London Hospitals, triggering lasting change for both physical and mental healthcare. It has now expanded its reach and so far it has been deployed in 11 NHS trusts and 3 Sub-National Safe Data Environments (SDEs)
Others who have adopted the system include international health and care providers and academic groups, including Monash Partners Academic Health Science Centre, Australia who were awarded AUS$1.92million for local deployment.
Looking to the future
The goal is to expand to all NHS Trusts, NHS England’s SDEs and new National Data Service, providing AI-based NLP text structuring (annotation) services using existing models. The team would like to use CogStack to help tackle inequalities through identitying biases from audits, for example in delays of dementia diagnoses biases. Now CogStack is formally launched as a university spinout there are also plans to scale revenue models (e.g., automated coding, life science partnerships).
Impact
- Enhanced data efficiency and savings of £1.2 m per year in 2018 for King’s College Hospital by detecting thousands of missing records of fracture clinic procedures in only 30 minutes.
- Accelerated medication reviews which saved more than two hours of work per pharmacy review at South London and Maudsley NHS Foundation Trust using CogStack AI to read and summarise records.
- Improving patient safety at University College London Hospitals NHS Foundation Trust by providing alerts for patients who are ‘lost-to-follow-up’ in their healthcare journey, identifying 100’s of such events to be actioned in the gastroenterology clinic.
References
Awards
