Room D229, School of Engineering of Warwick, Coventry, CV4 7AL
Since graduating with BSc in Theoretical Physics from Swansea University in 2014 I have worked as a software developer across several sectors creating scalable cloud-based solutions. From 2017-2019 I was employed at tech start-up working on using smartphones as a non-invasive medical device. A key area of research for the company was to predict blood glucose from a photoplethysmogram (PPG) signal obtained from the smartphones image sensor. The mobile app would allow users to take a “reading” which would be analysed and provide the user with some feedback on physiological parameters such as heart rate, heart rate variability, blood pressure, and respiratory rate.
During my time there I became very interested in artificial intelligence techniques and how they can be applied to biomedical signals for the purpose of non-invasive detection of parameters that would otherwise require an invasive procedure to detect, such as blood glucose.
My PhD project, supervised by Dr Leandro Pecchia and Dr John Hattersley, is titled “Artificial intelligence for non-invasive hypoglycaemia detection and prediction using wearable sensors”.
This project aims to develop an AI model to detect episodes of abnormally low blood glucose levels (hypoglycaemia) in diabetic patients by analysing the electrical activity of the heart measured non-invasively by a wearable sensor.
In collaboration with the Human Metabolism Research Unit at the University Hospitals Coventry and Warwickshire, we will recruit diabetic patients to participate in studies. The protocol will involve the wearing of a continuous glucose monitor (CGM) and electrocardiogram (ECG) device for continuous measurement and collection of blood glucose levels and ECG data. We will then use the data collected from the studies to create and evaluate AI models to predict episodes of abnormal blood glucose in diabetics.
Hypoglycaemia is a harmful complication of diabetes and insulin-dependent diabetics are more susceptible as interventions aim to achieve normal blood glucose levels. Symptoms can vary greatly and include, but not limited to, trembling, palpitations, sweating, dry mouth, confusion, seizures, coma or even death if untreated.
The early detection of a hypoglycaemic episode non-invasively has clear benefits. Upon detection of a hypoglycaemic episode a network connected device could alert the user, a nominated individual, carer, or healthcare provider to the status of the wearer. Such an alerting system enables intervention as the earliest moment, allowing for improved outcomes.
I am also interested carbohydrate-restricted diets for weight management and diabetes control.