Blog | Shaping the future of clinical AI

EMIS Web® was developed by GPs for GPs, and the quality and integrity of the patient record has always been, and remains, one of the defining characteristics of our clinical systems.

The advent of Artificial Intelligence (AI) means the quality of the patient record has never been more important, as the performance of Large Language Models (LLMs) is influenced by the quality and completeness of the data they are trained on.

Ambient Voice Technology (AVTs) relies on LLMs to perform their functions, and yet there are concerns that using standalone AVTs and ‘cutting and pasting’ into the patient record is leading to large quantities of less structured data being added, risking the degradation of clinical record quality.

Our commitment: A deeply integrated AVT for primary care

We’re committed to providing a deeply integrated AVT that’s tuned to primary care records and templates, facilitating structured data capture in a way that will not only maintain, but enhance data quality. In turn, this will allow agentic LLMs — for functions such as clinical decision support — to continuously improve, creating a virtuous cycle of quality improvement. It also creates an environment that can facilitate the high-quality testing, training and clinical evaluation of native, partner AI and third-party technologies for rapid deployment in the NHS.

To do this safely we’ve developed a robust governance framework for the evaluation and ongoing monitoring of AI technologies. Our approach follows a structured process to ensure AI tools are accurate, effective and safe.

First, we use data science methods to rigorously test their accuracy and precision. Next, clinicians review these outputs to assess how well the AI performs complex, higher-level tasks. We then apply clinical effectiveness protocols to validate the AI in real-world settings. Finally, we continuously monitor model performance across different populations and geographies, ensuring the tools remain generalisable, effective and safe over time.

This is in line with NHS initiatives to accelerate access to AI tools by providing rapid evidence synthesis for scientific, clinical safety, regulatory, legal and compliance submissions.

About the author

Dr Lucy Mackillop

Chief medical officer - Data and Research

Dr Lucy Mackillop is the Chief Medical Officer for Data and Research, and Caldicott Guardian at Optum and a practising Consultant Obstetric Physician. With a career deeply rooted in both clinical practice and academic research, she also holds an honorary senior clinical lectureship with the Nuffield Department of Women’s and Reproductive Health at the University of Oxford. Lucy is a Fellow of both the Royal College of Physicians and the Royal College of Obstetricians and Gynaecologists. At Optum, she combines her clinical expertise with a passion for data-driven innovation, playing a pivotal role in developing solutions that enhance patient care and streamline healthcare workflows.