Mid Cheshire Hospitals NHS Foundation Trust has launched its seven-year digital and data strategic plan to 2033, with a key focus across four domains for the deployment of AI, covering clinical, operational, workforce, and patients, with each domain represented across five stages through the strategy.
On the initial stage of AI deployment, AI will be used to support staff by processing and presenting information, with clinical functionality in summarisation, documentation, and ambient voice technology; operational functionality in reporting and basic analytics; workforce functionality in admin support and transcription; and patient functionality in information delivery and navigation. Decisions will be made by humans only in this stage, the trust states, with governance centring on transparency, privacy, and mandatory human review.
The second stage, named the “recommend” stage, sees the use of AI to provide insights and recommendations in clinical decision support, diagnostics, operational flow prediction and discharge suggestions, workforce productivity insights and task prioritisation, and patient triage prompts and personalised pathways. This will be on a human-in-the-loop basis, with clinical safety assurance, validation, and audit being key parts of governance, according to Mid Cheshire Hospitals.
The theme of human approval is also found in stage three, termed “enable”, in which AI can prepare or initiate actions, but still requires human approval. This may be in pre-populating clinical notes, orders, or care plans; in operational scheduling and pathway optimisation; in workflow automation; and in automated patient communication, the trust notes. Increased clinical safety officer oversight, audits, workflow controls, role-based accountability, and bias monitoring, will all be included in governance.
Stage four named “automate”, AI will be able to execute defined actions within safe boundaries, dependent on quality data. This will be for limited and tightly controlled clinical use cases, with automation of low-risk processes for the workforce, service automation of appointment management for patients, and operational uses such as scheduling, logistics, and command centre actions. Decisions will be subject to human monitoring and overrides, with the existence of set thresholds and fail-safes to promote safe use, Mid Cheshire Hospitals explains.
On the final stage, AI will be “autonomous”, making and executing decisions independently in the future state, the trust notes. In clinical use, this will mean advanced diagnostics and robotics; in operational, it will be used in autonomous system optimisation; and for the workforce, it will see fully automated processes. For patients, it will mean adaptive, continuous care pathways. Decisions will be system-led, meeting policy and regulatory requirements, and governance will include regulatory assurance, continuous oversight, and clear accountability.
Elsewhere, the trust sets out its future state and roadmap, with commitments to optimising core clinical systems, reducing paper processes, expanding virtual care, empowering staff with intelligent tools and AI, and implementing a unified enterprise data platform that integrates data across clinical, operational, workforce, and estates data. It also explores ambitions for the design and delivery of its intelligent hospital, focusing on digital readiness, system integration, data handover, workforce training, and testing of intelligent hospital capabilities prior to go-live.
Wider trend: Health AI
The Health Services Safety Investigations Body (HSSIB) has launched an investigation into the use of ambient voice technology (AVT) in hospitals, following engagement with stakeholders including representatives from national bodies that highlighted potential safety issues. “Intelligence gathered found that adoption of AVT is accelerating while the safety implications are not fully understood,” HSSIB details. “National literature and implementation activity has focused more strongly on efficiency benefits than on patient safety risk. There are also indications that routes for recognising and reporting AI-related incidents are not yet mature enough to provide confidence that emerging risks are being identified.”
Expressions of interest are being sought from manufacturers and Greater London NHS organisations for MHRA’s AI regulatory sandbox for the London region, with projects to be selected to participate in the initial phase deploying tech into live clinical settings. The initiative, London Region I, is set to create a real-world environment for the safe deployment of AI-enabled medical devices, offering the chance to generate evidence of benefits from NHS settings, and promote earlier patient access to new technologies, the MHRA reports.
AI-generated virtual patients are being used in a project led by Kent and Medway Mental Health NHS Trust, to explore how they can support medical students develop essential clinical skills. The 18-month project, a collaboration with Kent and Medway Medical School and the University of Kent, is designed to offer medical students opportunities to practice clinical conversations and key skills in a “safe and structured environment” before moving on to real clinical settings. The trust reports using “realistic virtual patient avatars” capable of interacting with students, as well as insights from people with lived experience of mental illness, to ensure scenarios are as realistic and respectful as possible.
