Supervision
I am currently supervising three master’s students across multiple institutions, each engaged in innovative research projects within the fields of cardiovascular research and data science for health and social care. Their work spans a variety of pressing healthcare challenges, from the analysis of incomplete diabetes data using advanced machine learning models, to the creation of visualisation tools for hybrid closed-loop systems, and the identification of key trends in Scottish accident and emergency activity. Together, these projects aim to apply cutting-edge statistical and machine learning techniques to real-world healthcare data, with the goal of improving patient outcomes and healthcare system efficiency.
Maximilian Walden MScR in Cardiovascular Research
with S.Forbes and M.Vallejo
Previous work on data, from the OpenAPS diabetes platform, has resulted in a suite of generalised linear models (GLMs) and sophiscated machine learning models being created for prediction of blood glucose levels. OpenAPS was developed by a dedicated community of individuals, named #wearenotwaiting, who sought to provide a more flexible and customisable approach to diabetes management. The data however is not always complete, and this means that the models are built and trained on a limited subsection of the data. Max’s project looks to identify the proportions of missing data in the variables which have been deemed to be significant, with the aim of then applying existing statistical and machine learning imputation approaches to this. The work will then progress towards the building and application of a transformer based system for imputation.
Dremon Salas MSc in Data Science for Health and Social Care
with S.Forbes
With the development of new statistical and machine learning models for the prediction of blood glucose levels, within a hybrid closed-loop system such as OpenAPS, there is a need to visualise both the data used and the results of the newly implemented models. This is order to allow for users to have a better understanding and hence better interpretation of what the system is doing. Typically it has been found that better interpretation of a medical-system by the user leads to better disease management. Dremon’s project looks to create bespoke visualisations which will show algorithm processes, comparing their similarities and differences. Additionally the work will look to create a set of user references and dashboard tools that can be accessed by the appropriate users.
Jason Alacapa MSc in Data Science for Health and Social Care
with K.Banas
The ability to identify key trends in accident and emergency (A&E) activity data within Scotland, is of a high priority, and is crucial for understanding the dynamics of healthcare utilisation, resource allocation, and patient outcomes. Current linear models help to understand this dynamical shift that occurs, but do not currently incorporate the potentiality of non-linear trends of attendances. Jason’s project therefore looks to develop and evaluate supervised machine learning models that predict whether patients attending A&E will meet the 4-hour waiting time standard using Scottish A&E data from 2018 to 2023. This research will provide evidence of key-influencing social and demographic factors and also looks to investigate the impact of the COVID-19 pandemic on A&E waiting times.