Completed Intern Projects

P62 - NHS Monitor Corpus

Samuel Hollands

Scraping and curation of NHS website data to create a corpus of representive text data

Published Report

Graph-based Web Scraping Text Data Corpus

P61 - Understanding Fairness and Explainability in Multimodal Approaches within Healthcare

Sophie Martin

Pipeline to compare the impact on fairness of using a fusion model versus a single modality model

Published Report

Fairness MultiModal Python

P53 - Transforming Healthcare Data with Graph-based Techniques using SAIL DataBank

Chris Tomlinson

This project explored topological and semantic graph embeddings of relevant healthcare ontologies to enhance performance on downstream tasks.

Published Report

Representations of data Embeddings Python

P52 - Exploring Process Mining with East Midlands Ambulance Service

Alex Coles

In collaboration with East Midlands Ambulance Service, this work explored using Process Mining techniques to better understand the processes within the service.

Published Report

Process Mining Ambulances Python - PM4Py

P51 - Investigating Privacy Concerns and Mitigations for Language Models in Healthcare

Vicky Smith

An initial exploration of privacy risks in healthcare language models, including privacy-preserving techniques applied before or after model training, and evaluating their effectiveness with privacy attacks.

Published Report

NLP Foundation Models Python

P43 - Enriching Neurology Patient Information using MedCAT

Aizaan Anwar

In collaboration with Lancaster Teaching Hospitals NHS Foundation Trust and Lancaster University, this work explored how to evaluate the embedding space generated for automated clinical coding tasks in Neurology.

Published Report

NLP Neurology Python - MedCAT

P42 - Including Mortality in Hypergraphs for Multi-morbidity

Zoe Hancox

Building on previous hypergraphs work (P34) that can extract the impact of predecessor and successor diseases on disease progression pathways, this work looked to include an implicit relationship to demographics and consider the impact of mortality.

Published Report

Representations of Data Hypergraphs Python - Numba

P41 - NHS Synth

Harrison Wilde

This project focused on building a package for generating useful synthetic data, audited and assessed along the dimensions of utility, privacy and fairness. It gives the ability to experiment with different model architectures to find which are the most promising for real-world usage.

Published Report

Synthetic Data VAE Python

P34 - Using Hypergraphs to Investigate the Impact of Comorbidities

Jamie Burke

In collaboration with Swansea University and the SAIL Databank, this work focused on the generation of hypergraphs for investigating the individual and joint impact of comorbidities on a patient pathway.

Published Report

Representations of Data Hypergraphs Python - Numba

P33 - Exploring Large-scale Language Models with NHS Incident Data

Niall Taylor

In collaboration with the NHS England patient safety data team, an investigation into how to produce a useful and valid representation space when training a language model for a healthcare task.

Published Report

NLP DeCLUTR Python - Transformers

P32 - Predicting the Impact of Health Inequalities - Diabetes

Stephen Richer

In collaboration with East Suffolk and North Essex foundation trust to apply a suite of data science techniques to a large population health data including both primary and secondary care data.

Published Report

RAP Code PHM Data Python - OSNnx

P31 - Txt-Ray Align Continued

Sarah Hickman

This project sought to identify the clinical application, pipeline and validation metrics for this work.

Published Report

Multi-modal Validation in Healthcare Python

P24 - Using LIME to explain facial disease classification

Anwesha Mohanty

Application of Local Interoperable Model-agnostic Explanations to an InceptionV3 classifier looking at a Rosacea

Published Report

Model Explainability LIME Python

P23 - STM for survey data

Anna Linton

The development of an R code for investigating the topics found in free text survey data using a technique that monitors both the content of the responses but also the metadata.

Published Report

RAP code Topic Modelling R

P22 - Txt-Ray Align

Dekai Zhang

An investigation of extracting insight from multi-modal text and imaging data using contrastive learning.

Published Report

Multi-modal Contrastive Learning Python

P21 - SynthVAE Continued

David Brind

Building on SynthVAE - focused on non-Gaussian input data, hyperparameter tuning, improving the codebase and starting to consider how fairness in the created data can be assessed and implemented.

Published Report

Synthetic Data VAE Python - PyTorch, Opacus

P14 - Model Class Reliance

Elizabeth Dolan

Investigating the use of MCR to identify the value of including commercial sales data in respiratory predictions

Published Report

Model Explainability Commercial Data Python - mcrforest, SHAP

P13 - NHS Text Data Exploration

Beth Rushton-Woods

Using a pre-defined toolset this project looked to understand how to ingest NHS.UK text data into a curated form.

Published Report

NLP Weak Supervison Python - scispaCy

P12 - SynthVAE

Dom Danks

Initial creation of a variational autoencoder with differential privacy for generating single table tabular gaussian data.

Published Report

Synthetic Data VAE Python - PyTorch, Opacus

P11 - SynPathDiabetes

Tiyi Morris

Exploration work into incorporating learning in to a pathway simulator for diabetes.

Published Report

Simulation Patient Pathways Python