Associate Professor Zongyuan Ge is at the forefront of using AI to transform healthcare. In conversation with DHCRC he shares his view on AI as a ‘teammate’ for clinicians and the exciting projects he’s working on right now.
You’ve built a career at the intersection of AI, digital health, medical imaging and machine learning – what drew you to this field, and what continues to fascinate you about it?
My interest in medical AI really solidified around 2018, back when I was at IBM Research in Melbourne. I was working on a project focused on melanoma skin cancer, and that’s when it clicked for me. I saw firsthand how AI could potentially transform the entire healthcare industry – not in an abstract way, but in a way that provides real, tangible benefits to the public.
AI has reshaped almost every industry in recent years – how would you assess its adoption in healthcare? What’s working, what’s lagging, and where do you see the biggest gains still to come?
We’ve definitely seen a wave of interest and some real progress in this space, but there’s still a long way to go. Implementing medical AI is quite complex – it’s not just about the technology, but also about ethics, clinical trials, and regulatory approvals. So, getting an AI model into real clinical use is a much more complicated process than in most other fields. That said, I’m quite hopeful. Companies like Tempus AI and Optain are doing great work to move the field forward. They really understand both the market needs and how the healthcare system operates, which makes a big difference when it comes to real-world adoption.
What are some of the most pressing challenges (or biggest opportunities) facing the health sector today?
I think the biggest challenge is also the biggest opportunity: closing the ‘bench-to-bedside’ gap. The research community publishes thousands of papers, many claiming to ‘beat human doctors’ at one specific task. But very few of these models ever make it into a real clinic workflow. This highlights a key issue with the tone of the AI community. The goal shouldn’t be to compete with doctors, it should be to collaborate. The biggest opportunity is in building ‘clinical assistants.’ The current generation of AI is fantastic at augmenting a doctor’s ability and catching things they might miss, but it’s not ready to operate fully autonomously. The future is about cooperation, not replacement.
A big focus for the DHCRC is building a workforce ready to use and embrace AI and digital tools. From your viewpoint, what skills will the health workforce of tomorrow need, and how are you helping to build that culture among your researchers?
The health workforce of tomorrow needs to be ‘bilingual’ – they need to speak the language of medicine and the language of data. I’m actually seeing this firsthand in my own team, and it’s incredibly encouraging. I’m fortunate to have several PhD candidates who are also medical doctors (MDs). They come into the lab, learn the deep technical skills of coding and AI modelling, and then they take that knowledge back into the clinic. They become our most valuable ‘champions’ and collaborators, because they were part of the development process and understand how to best integrate these tools. This new generation of clinician-scientists is exactly what we need to accelerate adoption. It reinforces my core belief: this future won’t be built by AI scientists alone. It absolutely requires a multidisciplinary effort.
Tell us about your work with the Digital Health CRC and how it’s helping to translate research into real-world healthcare outcomes?
Our work with the DHCRC is incredibly exciting. We’re essentially trying to unlock the eye as a ‘window’ into a person’s whole-body health. We’re building AI models that can detect signs of systemic diseases, like cardiovascular and kidney disease, from a simple, non-invasive retinal photograph. By combining advanced AI with these images and longitudinal health data, we can spot risk factors that might otherwise be invisible. The ultimate goal is simple: to make screening for major risks like CKD and CVD far easier, cheaper, and more accessible for everyone.
What’s next on your to do list in digital health / What are you most looking forward to in 2026?
That’s a great question, and it gets right to the heart of my current work. Most medical AI today is very specialised. It might be excellent at analysing an eye scan or a skin lesion, but it can’t see the connection between them. The problem is, medicine doesn’t happen in a silo – a patient is a whole person, not just a single organ. My ‘what’s next’ – and what I’m most excited about for 2026 – is a project my recent Viertel Bellberry Medical Fellowship is supporting. We’re building what we call a ‘Unified Phenotype Foundation Model’ (UPFM) to close this exact gap. Instead of looking at isolated data points, it learns the fundamental patterns of human health by integrating all of a patient’s information over time. This includes everything from eye scans and skin images to brain signals, lab results, and electronic health records. The goal isn’t replacement. The goal is to build a collaborative ‘teammate’ for clinicians – an AI that sees the complete picture. This unified view will enable truly proactive healthcare: helping clinicians catch diseases earlier, predict risk more accurately, and personalise treatments for major conditions like cardiovascular disease, skin cancer, and epilepsy.
