Browse by author
Lookup NU author(s): Dr Stephen McGough
This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).
© The Author(s) 2026. Identifying robust biomarkers for early cancer detection remains challenging, particularly when working with limited or heterogeneous datasets. Here, we present a proof-of-concept deep learning framework for cancer classification using blood-based proteomic profiles. Our approach leverages sample type transfer and synthetic data augmentation to improve performance and generalization across sample types. Models were trained on plasma proteome data from 13,208 pan-cancer cases and 39,806 controls in the UK Biobank. To address class imbalance and enrich the feature space, a convolutional neural network (CNN-Synth) was trained to detect cancer cases using data augmented with synthetic pan-cancer samples generated via a variational autoencoder. Performance was evaluated in an independent saliva-based dataset from a head and neck cancer case-control study (n = 156). CNN-Synth (AUC = 0.88) surpassed models trained without synthetic data (AUC ≤ 0.77). SHapley Additive explanations identified well-known cancer markers as key features. These results highlight the use of sample type transfer and synthetic data augmentation, with further validation needed.
Author(s): Shakeel A, Merriel SWD, Smith J, McGough AS, Suderman M, Abdallah ZS, Yousefi P
Publication type: Article
Publication status: Published
Journal: npj Digital Medicine
Year: 2026
Volume: 9
Print publication date: 08/07/2026
Online publication date: 02/05/2026
Acceptance date: 12/04/2026
Date deposited: 20/07/2026
ISSN (electronic): 2398-6352
Publisher: Springer Nature
URL: https://doi.org/10.1038/s41746-026-02658-7
DOI: 10.1038/s41746-026-02658-7
Data Access Statement: UKB data are available to access by application procedure detailed here: http://www.ukbiobank.ac.uk/using-the-resource/. SensOrPass data, study protocol, and data dictionary are available from the Head and Neck 5000 and Exeter 10,000 research resources. Full application details are available here: https://headandneck5000.org.uk/information-for-researchers/ and https://exetercrfnihr.org/about/exeter-10000/. The code for this study is publicly available on GitHub: https://github.com/MRCIEU/CNN-Synth-cancer-detection-using-deep-transfer-learning-and-data-synthesis. The repository contains the deep learning model for cancer classification using protein biomarkers, trained on UKB datasets. It includes scripts for model training, evaluation, and explainability, as well as the necessary dependencies specified in the requirements.txt file.
Altmetrics provided by Altmetric
