I graduated from Imperial College London with an MSci Physics degree in 2015. I then worked for Goldman Sachs from 2015 to 2020 gaining a range of experience from project management to process modelling and business decision management. I particularly enjoyed building efficient and transparent decision models for transaction reporting. From here my interest in data analytics and machine learning grew. In 2019 I completed the Applied Data Science course at the University of Oxford before joining the Interactive AI CDT at the University of Bristol in 2020.

As a PhD student at the University of Bristol my current research explores time series classification with healthcare data. I am supervised by Dr. Zahraa Abdallah and Professor Raul Santos-Rodriguez.

Research Interests

I am interested in applying machine learning methods to time series healthcare data in order to obtain valuable insights. Healthcare data comes with a range of challenges which includes missingness, noise and multi-modality (doctors notes vs. accelerometer data!).

Recently, I have been working with Electroencephalography (EEG) data to detect early stage Parkinson’s Disease. Our method involving a novel transformation of the EEG data achieved high performance on early stage classification. This work was presented at the Health Intelligence workshop at AAAI 2024, please read this pre-print for further information on this work. We also had a paper accepeted at the upcoming Machine Learning and Signal Processing (MLSP) 2024 conference where we investigate connectivity metrics combined with regional statistics for classification. Currently we are exploring deep learning methods for early stage Parkinson’s classification including convolutional neural networks, reservoir networks and are working towards a novel acrhitecture for EEG classification. Please see projects and publications for further details.

I have also collaborated on the ContinUous behavioUral Biomarkers Of cognitive Impairment (CUBOId) project. This project aims to identify behaviour patterns that can detect dementia early and better monitor the progression of the disease via a sensor system installed in the home. Acceleration data and approximate location are recorded via a wrist worn wearable and passive sensors throughout the home record data on temperature, humidity, light and electricity.