Dr Midhun Parakkal Unni

I am an Academic Fellow in AI for Health at the Centre for Machine Intelligence, University of Sheffield, where I work at the intersection of machine learning, dynamical systems theory, and signal processing. My research focuses on machine learning and mathematical modelling for prediction and mechanistically understanding physiology—particularly in applications involving neuromuscular control and wearable medical technologies.

I have a Ph.D. in Mathematics (Applied) from the University of Exeter, where I created mathematical and AI models to understand Parkinson's gait and predict motor abnormalities. My experience spans both academia and industry, including research at the University of Manchester and TCS Innovation Labs, where I worked on distribution shift problems in machine learning, human-in-the-loop methods, and dynamical systems modelling with applications in human motor control and cardiovascular pathologies.

I investigate uncertainty-aware, interpretable machine learning methods grounded in mechanistic understanding, with a focus on data from wearable sensors and physiological time series. My goal is to create robust AI systems that are clinically meaningful and deployable. I'm particularly interested in developing methods to address healthcare AI challenges such as distribution shifts (changes in data between training and deployment). and incorporating human-in-the-loop techniques. 

I welcome collaborations with researchers and clinicians in the fields of AI for health, dynamical systems, physiological modelling, signal processing, and digital health tools. If your work aligns with these themes, let's connect.