Fellowship Bio

I am a haematology registrar based in Addenbrookes Hospital, Cambridge. My interests include medical haematology, transfusion medicine and implementing AI tools in laboratory diagnostics. I am also the East of England representative for HaemSTAR, a national haematology registrar audit network.

Fellowship Project

Deployment of AI-driven application to fast classify blood cells towards automated reporting of blood films

Barts Health NHS Trust

The manual assessment of blood film morphology is a labour intensive and time-consuming process. To speed up the diagnostic process, the Barts Life Sciences team have developed a machine learning model for the detection of blast cells on digital blood films. This model uses a novel machine learning architecture and was developed in collaboration with the BloodCounts! consortium. My role in the project was twofold. I needed to develop a validation pipeline to test the performance of the model against real world data. I also had to review the regulations required to deploy an in-house model. To test the model, I collated blood films selected at random from films that were submitted for medical reporting. I drafted a scanning protocol, digitised the films and manually identified and labelled features of interest. The validation dataset containing the digitalized blood film features was analysed to understand how different morphologies can impact model performance. Future work may include expanding the training and validation datasets to fine tune the model’s performance and functionality. Simultaneously, I have been in close discussion with various stakeholders to understand the clinical pathway and to see how an AI tool could be best incorporated into the current workflow. I reviewed the local trust guidelines and performed a literature review of the legal and governance frameworks for AI regulation. I drafted a guideline that summarised some of the regulatory requirements and local processes required for the deployment of an in-house model within the Barts trust.

Fellowship Testimonial

This fellowship has been an incredible experience that has expanded my worldview and allowed me to take part in cutting edge AI model development. I am grateful to my supervisors Dr Suthesh Sivapalaratnam and Dr Concetta Piazzese and the Barts Life Sciences team for their guidance. This experience has developed my technical skills such as coding, and my understanding of the strengths and limitations of AI technologies. I have learnt to critically evaluate AI tools and better appreciate the challenges involved in its implementation in clinical practice. Other highlights from the fellowship include the monthly AI workshops and the chance to meet and brainstorm with my other clinical fellows. I hope to use my experience to help bridge the gap between AI development and clinical implementation, and to leverage my new skills to find innovative solutions to clinical problems.