Description du poste
Location: Highfield Campus Fixed Term until 31/01/2029 Overview We are seeking a Research Fellow to join LungSight, a multidisciplinary project developing low-cost, non-invasive audio-visual AI for earlier identification of chronic lung disease. The project aims to support a shift from reactive, clinic-based diagnosis towards proactive home screening using everyday devices such as smartphones, tablets and home computers. Based in the School of Engineering at the University of Southampton, you will lead the research engineering clinically informed acoustic features for fine-tuning machine learning models of respiratory health. You will work with large-scale real-world recordings of speech, breathing and cough, developing acoustic features that are robust to noise, demographic variation and different recording environments, predictive of early-stage chronic respiratory conditions and explainable in terms of the natural history of disease. You will also contribute to development of machine learning models and to the integration of audio with visual and clinical information. LungSight is a multi-university collaboration also involving Manchester Metropolitan University, the University of Cambridge, the University of Leicester, the University of Leeds, and the University of Sheffield alongside NHS and third-sector partners. This post will work particularly closely with a complementary PDRA at the University of Sheffield, whose research will focus more strongly on development of machine learning models while the Southampton post has a stronger focus on acoustic features and clinical knowledge mapping. The role offers opportunities for high-quality publications, open research outputs, clinical and stakeholder engagement, and career development at the interface of AI and healthcare. Main duties and responsibilities Develop and evaluate acoustic feature sets predictive of respiratory health using speech, breathing and cough recordings. Curate, pre-process and analyse large-scale real-world audio datasets, including sensitive healthcare and helpline recordings processed within appropriate secure data environments. Fine-tune and validate models on labelled respiratory datasets and clinically relevant downstream tasks, assessing accuracy, generalisability and clinical utility. Investigate potential confounding factors, demographic bias, background-noise effects and effects of recording environment, to support feature robustness and equitable model performance. Contribute to development of self-supervised models, working closely with the Sheffield PDRA who will lead complementary machine learning research; investigate how engineered physiological features can be combined with learned representations from the Sheffield machine-learning work. Collaborate with project researchers to develop multimodal approaches combining acoustic, visual and clinical information for respiratory disease screening. Work closely with researchers across the LungSight consortium, including Manchester Metropolitan Sheffield, Cambridge, Leicester and Leeds Universities, as well as clinical, NHS, patient/public and third-sector partners; participate in project meetings, joint research activities and dissemination. Lead and contribute to high-quality research publications and presentations, support project reporting, and contribute to the supervision and development of students where appropriate. You should hold a PhD (or equivalent research experience) in Computer Science, Electronic Engineering, Speech/Audio Processing, Machine Learning, Biomedical Engineering or a closely related discipline, with research experience in speech, audio, acoustic signal processing or closely related time-series analysis. Working at the University of Southampton: Check out the staff benefits and why you should join us at The University of Southampton. Further details: Job Description and Person Specification Closing Date: Wednesday 16 December 2026 Interview Date: To be confirmed