Data Scientist (KTP Associate) for Deep Learning and Image, Audio and Video Processing

This exciting 24 month project is seeking to recruit an enthusiastic KTP Associate to develop novel machine learning solutions that detect Type 2 diabetes and prediabetes by analysing voice patterns, providing a non-invasive and innovative method for early diagnosis. Currently, Type 2 diabetes (T2D) is often diagnosed 5-7 years after symptom onset, leading to severe complications like blindness, kidney failure, heart attacks, stroke, and limb amputation. About 50% of T2D patients develop diabetic neuropathy, which can damage nerves throughout the body, including those controlling the vocal cords, leading to issues like vocal fold paralysis, hoarseness, or vocal strain.
This post is based at DDM Health Ltd., Coventry, and the Associate is also expected to attend project meetings and training events at the university campus in Egham, Surrey.

As part of this KTP project, the Associate will:

• Establish the extent and quality of the survey data the inspectorate holds, using a range of statistical techniques to explore the dataset
• Identify any ethical issues that may limit access to the data and suggest solutions
• Using initiative and creativity to identify areas for research, developing new research methods and extending the research portfolio
• Identify any practical and technical issues that require to be resolved and suggest solutions. Identify any specific limitations on data-sharing (in relation to specific datasets, types of analysis etc.) and/or additional work required to enable wider access
• Pilot, refine and optimise mathematical and computational techniques with limited supervision from the PI
• Take a leading role in overcoming methodological problems that arise during the project lifespan, and keep the PI up to date with the project’s progress
• Formulate and test new hypotheses arising from the data, where relevant
• Provide technical support for designated research area including maintenance of equipment, lab supplies, databases and webpages, ensuring that all work is carried out in accordance with statutory and Royal Holloway regulations as appropriate.
• Set up and perform experiments in close collaboration and consultation with relevant academic/research colleagues.
• Record, analyse and write up results, maintaining a permanent record of the methodologies and the experiments.
• Compose research manuscripts for peer-reviewed journal/conference publication
• Prepare and present findings of research activity to colleagues and at scientific meetings
• Contribute to the preparation of grant applications, research presentations and publications as requested.
• Undergo continued personal professional development

The project aims to develop an innovative AI-driven tool for diabetes and prediabetes detection using voice inputs. Specifically, this project will develop novel deep learning algorithms, and audio and vision transformers with various attention mechanisms, to identify early warnings and prediabetes/diabetes conditions. Weakly supervised and zero/few-shot learning methods will also be exploited to classify unseen conditions without or with limited training samples, in order to tackle data sparsity issues.