Dinny Graham
Senior Research Fellow
Research Interests
Make your next move.
Prepare your application
Review drafts before you send them.
Explore related searches
Articles (10)
Year:
2026
Artificial intelligence applications in biomedicine face major challenges from data privacy requirements. To address this issue for clinically annotated tissue proteomic data, we developed a federated deep learning approach (ProCanFDL), training local models on simulated sites containing data from a pan-cancer cohort (n = 1,260) and 29 cohorts held behind private firewalls (n = 6,265), representing 19,930 replicate data-independent acquisition mass spectrometry runs. Local parameter updates were aggregated to build the global model, achieving a 43% performance gain on the hold-out test set (n = 625) in 14 cancer subtyping tasks compared with local models and matching centralized model performance. The approach’s generalizability was demonstrated by retraining the global model with data from two external, data-independent acquisition mass spectrometry cohorts (n = 55) and eight acquired by tandem mass tag proteomics (n = 832). ProCanFDL presents a solution for internationally collaborative machine learning initiatives using proteomic data, for example, for discovering predictive biomarkers or treatment targets while maintaining data privacy. Significance: A federated deep learning approach applied to human proteomic data, acquired using two distinct proteomic technologies from 40 tumor cohorts across eight countries, enabled accurate cancer histopathologic subtyping while preserving data privacy. This approach will enable the privacy-compliant development of large-scale proteomic artificial intelligence models, including foundation models, across institutions globally.
Year:
2025
The increased use of spatially resolved transcriptomics provides new biological insights into disease mechanisms. However, the high cost and complexity of these methods are barriers to broader application. Consequently, methods have been created to predict spot-based gene expression from routinely collected histology images. Recent benchmarking showed that current methodologies have limited accuracy and spatial resolution, constraining translational capacity. Here, we introduce GHIST, a deep learning-based framework that predicts spatial gene expression at single-cell resolution by leveraging subcellular spatial transcriptomics and synergistic relationships between multiple layers of biological information. We validated GHIST using public datasets and The Cancer Genome Atlas data, demonstrating its flexibility across different spatial resolutions and superior performance. Our results underscore the utility of in silico generation of single-cell spatial gene expression measurements and the capacity to enrich existing datasets with a spatially resolved omics modality, paving the way for scalable multi-omics analysis and biomarker identification.
Year:
2025
Artificial intelligence applications in biomedicine face major challenges from data privacy requirements. To address this issue for clinically annotated tissue proteomic data, we developed a federated deep learning approach (ProCanFDL), training local models on simulated sites containing data from a pan-cancer cohort ( n = 1,260) and 29 cohorts held behind private firewalls ( n = 6,265), representing 19,930 replicate data-independent acquisition mass spectrometry runs. Local parameter updates were aggregated to build the global model, achieving a 43% performance gain on the hold-out test set ( n = 625) in 14 cancer subtyping tasks compared with local models and matching centralized model performance. The approach’s generalizability was demonstrated by retraining the global model with data from two external, data-independent acquisition mass spectrometry cohorts ( n = 55) and eight acquired by tandem mass tag proteomics ( n = 832). ProCanFDL presents a solution for internationally collaborative machine learning initiatives using proteomic data, for example, for discovering predictive biomarkers or treatment targets while maintaining data privacy. Significance: A federated deep learning approach applied to human proteomic data, acquired using two distinct proteomic technologies from 40 tumor cohorts across eight countries, enabled accurate cancer histopathologic subtyping while preserving data privacy. This approach will enable the privacy-compliant development of large-scale proteomic artificial intelligence models, including foundation models, across institutions globally.
Year:
2025
Year:
2024
The increased use of spatially resolved transcriptomics provides new biological insights into disease mechanisms. However, the high cost and complexity of these methods are barriers to broad clinical adoption. Consequently, methods have been created to predict spot-based gene expression from routinely-collected histology images. Recent benchmarking showed that current methodologies have limited accuracy and spatial resolution, constraining translational capacity. Here, we introduce GHIST, a deep learning-based framework that predicts spatial gene expression at single-cell resolution by leveraging subcellular spatial transcriptomics and synergistic relationships between multiple layers of biological information. We validated GHIST using public datasets and The Cancer Genome Atlas data, demonstrating its flexibility across different spatial resolutions and superior performance. Our results underscore the utility of in silico generation of single-cell spatial gene expression measurements and the capacity to enrich existing datasets with a spatially resolved omics modality, paving the way for scalable multi-omics analysis and new biomarker discoveries.
Year:
2024
Year:
2023
Year:
2023
Year:
2023
Year:
2023
Collaborators (36)
CAROLINA ARMENGOL
Principal Investigator- Childhood Liver Oncology group
-
Georgina V. Long
University of Sydney
Robin Anderson
Head
La Trobe University
Graham Mann
Professor
University of Sydney
Yue Cao
University of Sydney
Phillip J. Robinson
University of Sydney
Jia Liu
The University of Sydney Faculty of Medicine and Health
Roger Reddel
Lorimer Dods Professor and Director
Children's Medical Research Institute
Maija Kohonen-Corish
Macquarie University
Adel Aref
Senior Researcher
Children's Medical Research Institute
Wendy A Cooper
-
Zhaoxiang Cai
Conjoint Associate Lecturer
University of Sydney
ANTHONY WEISS
University of Sydney
Reginald V N Lord
Professor and Head of Surgery
University of Notre Dame
Anthony Gill
University of Sydney
Lisa Horvath
Conjoint Professor of Medical Oncology (Genitourinary Cancer)
University of Sydney
Karen MacKenzie
Conjoint Associate Professor
The University of Sydney School of Medicine
David Thomas
Director
State Library of New South Wales
Ellis Patrick
Associate Professor
University of Sydney
Mark Basik
McGill University
Roderick Clifton-Bligh
University of Sydney
Elgene Lim
Clinical Academic/Professor of Medicine
University of New South Wales
John Souglakos
University of Crete
Jean Yang
Professor
University of Sydney
Zaklina Kovacevic
University of New South Wales Science
Heloisa Helena Milioli
University of New South Wales
Michael Gnant
Full Professor
Medical University of Vienna
Jacob George
Professor
University of Sydney
Rebecca C. Poulos
Children's Medical Research Institute
Jennifer Koh
Staff Scientist
Children's Medical Research Institute
Alexander Swarbrick
Senior Principal Research Fellow & Director of the Cancer Ecosystems Program
Garvan Institute of Medical Research
Peter Hains
Team Leader
Children's Medical Research Institute
Richard A. Scolyer
Faculty Member
University of Sydney
Muhammad Shaban
Machine Learning Scientist
Brigham and Women's Hospital
Qing Zhong
Group Leader | Associate Professor
Children's Medical Research Institute
jinman kim
Professor
University of Sydney
Make your next move.
Prepare your application
Review drafts before you send them.