×
G. Thomas Hudson

G. Thomas Hudson

Machine Learning Researcher


Background


About

About

Machine learning researcher working on fundamental multimodal models and applications within healthcare. Specialising in multimodal and health data, developing foundation models and intelligent systems that bridge advanced AI methods and real-world clinical impact.

Work Experience

Work Experience

  • Research Fellow, Centre for AI Fundamentals, 

    Jan, 2026 - Present

    Developing AI fundamentals and their applications to healthcare with Professor Sami Kaski.

    • Developing AI fundamentals including Bayesian inference, human-in-the-loop methods, and foundation models.

    • Applying fundamental AI methods to healthcare.

  • Senior Research Associate, 

    Jan, 2025 - Jan, 20261 year

    Developed methods to monitor the quality of hospital discharge notes, enabling clinicians to communicate more effectively and improving patient outcomes.

    • Worked with clinicians from Salford Royal Hospital and a multidisciplinary team, producing multiple papers and contributing to a spinout company.

    • Built and led a team developing fundamental machine learning methods, leading to multiple papers including a novel multimodal approach presented at ICCV 2025.

    • Mentored Masters and Doctoral students, guiding research directions that led to publications in venues including EMNLP.

  • Senior Data Scientist, 

    Jan, 2025 - Present

    Worked at the intersection of research and industry, building models to analyse electronic health records.

    • Developed methods to automatically triage patient requests in GP practices.

    • Developed methods to detect skin cancer in patient-taken images.

  • Post Doctoral Research Associate, 

    Feb, 2023 - Jan, 20251 year 11 months

    Worked on a project exploring how techniques including large language models can solve veterinary problems.

    • Used LLMs to understand discussions on veterinary message boards.

    • Developed methods to extract diseases from unstructured case discussions and match them to custom veterinary vocabularies.

    • Built tools to identify longitudinal trends in diseases and treatments.

    • Supported productive discussions between veterinary professionals dealing with complex cases.

    • Managed communication and collaboration across an international multidisciplinary team.

  • Data Scientist, 

    Jul, 2018 - Jan, 20234 years 6 months

    Researched how machine learning techniques can solve NLP problems in the anti-money laundering sector, including information extraction from large unstructured documents.

    • Applied research techniques to industry problems.

    • Developed methods for managing and monitoring models in production.

Projects Experience

Projects Experience

  • Automated patient discharge letter analysis

    - Present

    Developed and funded work on automated analysis of patient discharge letters to support cost-effective hospital audits and quality improvement projects.

  • Veterinary clinical NLP

    - Present

    Built LLM-based tools to understand veterinary message-board discussions, extract disease mentions, match them to vocabularies, and identify longitudinal trends.

  • Multimodal models for healthcare

    - Present

    Developed multimodal approaches combining clinical text and images for healthcare applications, including skin cancer classification.

Skills

Skills

  • Machine Learning

    Python

    PyTorch

    HuggingFace

    LLMs

    foundation models

    multimodal learning

    Bayesian inference

    human-in-the-loop AI

    model monitoring

    survival analysis

  • Natural Language Processing

    large language models

    long-document understanding

    information extraction

    medical text de-identification

    clinical NLP

    native language identification

  • Healthcare AI

    electronic health records

    hospital discharge summaries

    clinical communication

    patient request triage

    skin cancer classification

  • Teaching and Research Leadership

    NLP teaching

    workshop design

    student mentoring

    interdisciplinary collaboration

    grant writing

Education

Education

  • Computer Science, PhD, Durham University

    Jan, 2019 - Jan, 2024

  • Computer Science, MEng, Durham University

    Sep, 2014 - Jul, 2018

    Machine Learning

Awards

Awards

  • BCS Chartered Institute for IT Prize , BCS, The Chartered Institute for IT

    Awarded on:

    Awarded for being the top performing student in Computer Science.

Publications

Publications

  • Bridging Survival Analysis and Machine Learning to Improve Healthy Life Expectancy Estimation using PHR Records, npj Digital Medicine

    Published on: Jan 01, 2026

    Benedict Jones, James Harmsworth King, Matthew Watson, G. Thomas Hudson, and Noura Al Moubayed. In press.

  • Generalizable multilingual medical text de-identification using generative instruction tuning, Nature Communications Medicine

    Published on: Jan 01, 2026

    G. Thomas Hudson*, Chenghao Xiao*, Benedict Jones, James Harmsworth King, and Noura Al Moubayed. Joint first authors. In press.

  • Automating the quality monitoring of a hospital discharge summary improvement project utilising large language models, npj Digital Medicine

    Published on: Jan 01, 2026

    G. Thomas Hudson, Benjamin David James, Matthew Watson, Mark Holland, Olivier Gaillemin, Noura Al-Moubayed, Darren Green.

  • Multimodal Models for Skin Cancer Classification using Clinical Free Text and Dermatoscopic Images, Nature Communications Medicine

    Published on: Jan 01, 2026

    Matthew Watson, Thomas Winterbottom, Thomas Hudson, Benedict Jones, Hubert P. H. Shum, Amir Atapour-Abarghouei, Toby Breckon, James Harmsworth King, and Noura Al Moubayed.

  • Everything is a video: Unifying modalities through next-frame prediction, International Conference on Computer Vision

    Published on: Jan 01, 2025

    G. Thomas Hudson, Dean Slack, Thomas Winterbottom, Jamie Sterling, Chenghao Xiao, Junjie Shentu, and Noura Al Moubayed. ICCV 2025.

  • Video Prediction of Dynamic Physical Simulations With Pixel-Space Spatiotemporal Transformers, IEEE Transactions on Neural Networks and Learning Systems

    Published on: Jan 01, 2025

    Dean Slack, G. Thomas Hudson, Thomas Winterbottom, and Noura Al Moubayed.

  • Towards more human-like language models based on contextualizer pretraining strategy, EMNLP/CoNLL 2023 BabyLM Challenge

    Published on: Jan 01, 2023

    Chenghao Xiao, G. Thomas Hudson, and Noura Al Moubayed. Loose Track Winner.

  • Length is a Curse and a Blessing for Document-level Semantics, Proceedings of EMNLP 2023

    Published on: Jan 01, 2023

    Chenghao Xiao, Yizhi Li, G. Thomas Hudson, Chenghua Lin, and Noura Al Moubayed.

  • Can Text Encoders be Deceived by Length Attack?, ICLR 2023

    Published on: Jan 01, 2023

    Chenghao Xiao, Zihuiwen Ye, G. Thomas Hudson, Zhongtian Sun, Phil Blunsom, and Noura Al Moubayed.

  • MuLD: The Multitask Long Document Benchmark , The Language Resources and Evaluation Conference (LREC 2022)

    Published on: Jun 25, 2022

    MuLD is a benchmark of six NLP tasks where inputs consist of at least 10,000 words. It covers task types including translation, summarization, question answering, and classification, with output lengths ranging from a single classification label to outputs longer than the input text.

  • Ask me in your own words: paraphrasing for multitask question answering , PeerJ Computer Science

    Published on: Jun 14, 2021

    This work studies how models trained for decaNLP fail under simple paraphrasing of questions, introduces the PQ-decaNLP corpus, and shows that training with paraphrased questions improves robustness.

  • On the Development of a Large Scale Corpus for Native Language Identification , Treebanks and Linguistic Theories (TLT17)

    Published on: Jan 01, 2018

    Introduces italki, a large corpus for Native Language Identification, and evaluates its impact on machine learning systems for classifying authors' native languages from English writing.