Project Spotlight: AI Scribe Implementation
What are AI Scribing products
Ambient scribing products, when combined with Generative Artificial Intelligence (AI) can convert speech directly into structured medical documentation, such as notes and letters. The products are sometimes referred to as ambient scribes or AI scribes and include ambient voice technologies (AVTs) used for clinical or patient documentation and workflow support.
Functions can vary though inputs include speech, user-provided context, external health record information. Outputs include transcripts, summaries, medical letters, population of records, clinical coding and suggestion of tasks.
The Challenge
AI scribes handle sensitive patient data, necessitating strict adherence to data protection regulations and standards. While these tools promise major reductions in documentation burden and cognitive load. They also introduce new types of risks that must be actively understood and managed. The complexity of managing data flows, ensuring patient confidentiality, and maintaining compliance across multiple organisations presents significant challenges.
Our Approach
Testing AVT in the CoDE
As AI scribe technology become increasingly integrated into clinical workflows, ensuring robust processes are in place is paramount.
The CoDE has initially focused its work programme on testing how AVT work in clinical practice – this includes testing error, omission, hallucination rates, unintended changes to workflow and communication as well as practical considerations (such as microphone placement).
AI scribe/AVT – supporting adoption and spread regionally
Unlike some new technologies that are fully tested ‘off line’ or piloted first and as suppliers are offering free versions of their products, various AVT products are already being utilised in clinical practice.
The regional AVT initiative seeks to disseminate findings from the CoDE testing on AVT and provide support and resources that will enable safe and effective use specifically of AI Scribes in practice.
A peer to peer network (being referred to as the AVT Collaborative) has been created with clinical and digital membership from the seven SW systems. This network aims to work together, supported by the regional team and with expert advice and support from clinical safety and information governance experts.
More specifically the following will be provided to systems:
Clinical safety advice and support to ensure organisations comply with DCB0160 standards. This will be achieved by providing the following:
- Digital Clinical Safety support and guidance
- Programme-level hazard workshops and drop-in sessions
- A system-level approach to AVT assurance
- Opportunity to collaborate, sharing and learning from across the region
- Facilitate the input, knowledge and experience from system
Information Governance advice and support to enable the safe, lawful, and transparent use of AVT. This will be aligned with UK GDPR, Caldicott Principles, NHS DSP Toolkit standards. IG Assurance and Compliance Support will include:
- Assistance with AVT-specific DPIAs
- Risk identification and mitigation advice
- Supplier compliance assessment (e.g. DSP Toolkit, ISO 27001)
- Contractual and data processing agreement checks
- Ongoing Advice and Support:
- Access to IG specialists for Q&A
The plan with this (and subsequent work) is to work with the Data Protection Officers and the Clinical Safety Officers to support an agreed approach. We aim to develop shared guidance, templated artefacts, and common approaches that can accelerate safe implementation while maintaining clinical vigilance and organisational accountability.
This approach links back to both the national NHSE AI guidance for AI Scribes, engagement with clinical safety experts, and the experiments taking place at the CoDE (see link below). All experiments commence with research involving clinicians to validate the key challenges that we need to prioritise.
Key Objectives
Develop standardised Data Protection Impact Assessment (DPIA) templates tailored for AI scribe implementations.
Establish clear guidelines for data processing, storage, and sharing, ensuring transparency and accountability.
Create training modules for staff to understand and uphold IG principles in the context of AI technologies.
Why This Matters
By proactively addressing Information Governance, CoDE ensures that AI scribe technologies are deployed in a manner that respects patient privacy, complies with legal standards, and maintains public trust. This initiative supports the broader goal of integrating AI into healthcare safely and effectively.