My Topol fellowship problem / project:
AI tools have potential benefits in clinical pathways, but only a few AI models are successfully implemented and utilised in clinical environments. One reason is lack of appropriate trust and confidence in AI model predictions for clinical decision making. The fellowship aims to tackle this challenge by exploring best practices for presenting data from the AI models to clinicians in a user centric way with appropriate clinical and statistical context. The project will investigate issues related to UX/UI design for AI solutions with particular focus on human factors, workflow integration and cognitive biases in decision making. I will develop and test prototypes for presenting contextual information to clinicians alongside AI outputs. The aim here is to improve interpretability of AI outputs whilst also presenting Clinicians with information related to the scope and limitations of the AI model in a particular context. This will enable optimal AI-augmented clinical decision making with appropriate level of trust and confidence in AI model predictions. The principles and processes developed during this fellowship will feed into future roll out and local governance of AI tools and training for clinicians, enabling safe and appropriate use of AI tools for Clinical decisions making.
About me
I am a Bioinformatician working in the scientific computing team at Leeds Teaching Hospitals where we are helping deliver robust digital solutions to transform clinical workflows. With a background in electronics engineering and Masters in Biomedical Engineering and Computer Science and I have accumulated several years of experience working in various roles related to health-tech within the NHS.
My current focus is on developing tools and setting up digital infrastructure that is powering the AI-Augmented clinical workflows of the future. My interests include clinical implementation of AI with a particular focus on decision support systems, data engineering and data pipeline development, Machine learning Ops (MLOps) and Software as a Medical device.