Research

Scientific rigour at the heart of every model we build.

Profenso works in sectors where decisions matter — healthcare, finance and engineering. That is why our research programme is grounded in peer-reviewed science, transparent methods and a deep partnership with Australia's leading universities.

Based in Queensland, Australia Partnered with UQ, QUT, QIMR Berghofer & UniSC Focus: safe AI for healthcare
Our approach

Why our research matters

When AI is used in healthcare, a wrong prediction can change a person's life. We take that responsibility seriously. Every Profenso project is built with the same standards used in academic research: rigorous testing, methods that can be explained, honest measures of uncertainty, and neural network designs tailored to the problem at hand.

Where possible, our solutions run locally and keep data anonymised — so that safety, privacy and accuracy reinforce one another rather than compete.

Industry & academia

A bridge between research and practice

Scientific research underpins progress in every applied field. We see our role as keeping the bridge between industry and academia open and well travelled. To that end, we maintain close working relationships with major Queensland institutions:

The University of Queensland logo The University of Queensland Queensland University of Technology logo Queensland University of Technology QIMR Berghofer Medical Research Institute logo QIMR Berghofer University of the Sunshine Coast logo University of the Sunshine Coast

Through adjunct and honorary appointments at these institutions, we secure collaborative research grants, offer top-up scholarships and host internship placements for current and prospective students. Our academic standing also allows us to provide industry-based co-supervision of PhD theses.

Focus areas

Where we are currently working

Our research interests sit at two ends of the spectrum: the mathematical and architectural foundations of AI models, and the safe, clinical application of AI in ophthalmology, cancer and mental health.

  1. 01

    AI safety in sensitive sectors

    Building methods and standards that make AI safe to use in high-stakes settings such as healthcare.

  2. 02

    AI vision for glaucoma

    Using computer vision to improve clinical and biological understanding of glaucoma — one of the leading causes of irreversible blindness.

  3. 03

    Explainable AI for vision and language

    Developing the mathematical and analytical optimisations that allow modern AI models to be interpreted and trusted by domain experts.

  4. 04

    Local AI agents for skin cancer

    Applying privacy-preserving, on-device AI agents to improve clinical phenotyping of keratinocyte cancers.

  5. 05

    Multi-agent frameworks for professional work

    Designing systems in which multiple AI agents cooperate to carry out complex, software-based professional tasks under human oversight.

Collaborators

Our scientific collaborators

We work alongside leading Australian researchers whose expertise spans mathematics, genetics, ophthalmology and biotechnology.

Professor Fred Roosta

Professor Fred Roosta

School of Mathematics and Physics, The University of Queensland

Fred is an ARC DECRA Fellow and Lecturer at The University of Queensland, an Associate Investigator with the ARC Centre of Excellence for Mathematical and Statistical Frontiers (ACEMS), and a Distinguished Research Scholar at the International Computer Science Institute (ICSI) in Berkeley, USA.

  • Machine Learning
  • Numerical Optimisation
  • Randomised Algorithms
  • Computational Statistics
  • Scientific Computing
Associate Professor Puya Gharahkhani

Associate Professor Puya Gharahkhani

Genomics, Imaging and AI Laboratory, QIMR Berghofer

Puya leads the Genomics, Imaging and AI Laboratory at QIMR Berghofer and holds a prestigious NHMRC Investigator Grant along with two MRFF grants. His research uses statistical genetics and machine learning to identify disease genes for complex traits — with a particular focus on neurodegenerative diseases of the eye and brain, including glaucoma, macular degeneration, dementia and Parkinson's disease.

  • Statistical Genetics
  • Multimodal AI
  • Ophthalmology
  • Neurodegeneration
Associate Professor Matthew H. Law

Associate Professor Matthew H. Law

Genetics and Skin Cancer Lab, QIMR Berghofer

Matthew heads the newly formed Genetics and Skin Cancer Lab at QIMR Berghofer. He is recognised for his work in large-scale data analysis and the application of statistical techniques to uncover the genetic underpinnings of skin cancer, particularly melanoma.

  • Statistical Genetics
  • Genome-wide Association Studies
  • Polygenic Risk Scores
  • Skin Cancer
Professor Abigail Elizur

Professor Abigail Elizur

Centre for Bioinnovation, University of the Sunshine Coast

Abigail is Co-Director of the Centre for Bioinnovation. She joined the University of the Sunshine Coast in 2004, after a senior research role with the Queensland Department of Primary Industries at the Bribie Island Research Centre. Her major research focus is the application of biotechnology to aquaculture.

  • Biotechnology
  • Aquaculture
  • Applied Genomics
Next generation

Our current PhD students

We co-supervise doctoral candidates whose work pushes the boundaries of interpretable AI, genetic risk prediction and large-language-model phenotyping.

Eslam Zaher

Eslam Zaher

Interpretable AI — Theory and Practice

In collaboration with industry partner Max Kelsen, Eslam's PhD is supervised by Dr Fred Roosta-Khorasani, Dr Quan Nguyen and Dr Maciej Trzaskowski. His work explores the behaviour of black-box models and develops interpretable methods to support the safe adoption of AI in healthcare.

Asma Aman

Asma Aman

AI for genetic risk in glaucoma

Asma's thesis centres on using AI to improve genetic risk estimation in glaucoma. Her research aims to improve the discovery of neurodegenerative genes, deepen understanding of glaucoma's neurodegenerative aspects, and refine predictions for the risk and onset of primary open-angle glaucoma.

Marloes Helder

Marloes Helder

Phenotyping with large language models

Marloes derives accurate phenotypes from large-scale datasets using large language models and genetic analyses. She has automated the identification of keratinocyte cancers from pathology reports, and is now applying language models to medical records to predict additional health outcomes — validating those predictions with genetic methods. She also works on distinguishing in-situ from invasive melanoma using genome-wide association studies and polygenic risk scores.

Work with us

Interested in collaborating, partnering or studying with us?

Whether you are a clinician, a researcher, a prospective PhD candidate or an industry partner, we would love to hear from you.

Get in touch
Previous Research