NHS England has published its sepsis modern service framework, putting forward a digital roadmap with recommendations and timelines for digital infrastructure including the single patient record, the Federated Data Platform, EPRs, machine learning, AI, wearables, and digital diagnostics.
The framework outlines how the NHS will work to improve sepsis care over the next ten years, focusing-in on priority actions, and helping to inform those involved in planning and commissioning services for those with, or at risk of, sepsis. The ambition is to “reduce death, life-threatening complications and long-term impact from sepsis and severe infection in all patient groups by at least 25 percent by 2035”, NHSE shares.
Over the next year, focus will be on commissioning a new national infection and sepsis audit with digital data collection and reporting as a core mechanism, it states. Minimum digital information standards for early warning systems will be established to ensure consistent implementation across EPR systems, variation in enhanced and critical care capacity will be examined using national data, and guidance on sepsis, infection, antimicrobial resistance, and deterioration will be digitally curated and made easier to access. “These actions create the data backbone (audit, standards and interoperability) without which later AI, diagnostics and monitoring ambitions cannot be delivered at scale,” NHSE states.
Short-term digital improvements for the next two-to-three years will look to develop and validate sepsis risk prediction and stratification tools; linking primary, secondary, and community datasets to understand pathways, delays, outcomes, and inequalities; building on the Community Deterioration Management improvement programme to explore the use of digital early warning systems; and introduce diagnostic tools with digital integration to clinical records.
Longer term ambitions for the next ten years cover the use of the Federated Data Platform for the large scale analysis of infection and sepsis data, flagging patients at risk of sepsis in the single patient record, and enabling advance care plans or treatment escalation plans to be shared through shared care records. Machine learning will be used in personalising predictions, diagnosis, and treatment, and AI will be used in clinical decision support systems, with wearable technologies helping to detect deterioration earlier.
Wider trend: AI in health and care
East Kent Hospitals University NHS Foundation Trust has started a pilot of an AI tool capable of analysing routine clinical information and identifying infection risk early. MEMORI, developed in collaboration with teams at the trust and embedded in Sunrise, has reportedly been offered for trial at no cost as part of a long term partnership with Sanome focusing on clinical AI safety. The tool has been rolled out in the Harvey ward at Kent and Canterbury Hospital. The AI tool is said to analyse individual patient data including observations, medications, and demographics, and using what it has learnt from thousands of other patients to predict an infection risk score. The system also offers alerts if patient risk level changes.
An AI system is being used to colour code internal anatomy during surgery at London North West University Healthcare (LNWH) NHS Trust, said to help surgeons to see subtle structures that might be hard to distinguish with the eye. The Eureka system analyses the surgical field in real-time, overlaying colours onto structures such as nerves and connective tissue. Surgeons are able to choose between keeping the overlay constant, or having it pulse intermittently during different stages of a procedure. LNWH shares hopes that use of Eureka will enhance precision, reduce risk of accidental injury, and help to support surgeons during complex procedures.
The government of Alberta has employed Claude AI to locate and fix cyber security vulnerabilities across government systems, said to assess 466 million lines of code, implement fixes, and run continuous security review. Claude Code’s Opus and Sonnet models have been used to analyse systems in all 27 provincial ministries, covering 1,280 applications and 3,400 code repositories, most of which the government notes “has never undergone a systematic security review”, and with accumulated technical debt estimated to cost into the billions of dollars.
