AI Insider: unlocking autoimmune disease

Since his first day at UNSW in 2005, Professor Fabio Luciani has been working with machine learning in his field of systems immunology.

Professor Fabio Luciani in the School of Medical Sciences has been applying computational biology and machine learning to understand how the immune system functions in health and disease, since he first joined UNSW. Throughout his career, his research has combined computational biology, statistics and machine learning, disciplines that now form a key part of modern artificial intelligence (AI).

With a focus on autoimmune disease, one of the biggest challenges in biomedical research, the immune system mistakenly attacks the body's own tissues as a result of both environmental and genetic factors. Prof. Luciani’s work focuses on disentangling the complex factors around what causes these diseases.

“The number of autoimmune diseases is growing and rare ones are being discovered,” said Prof. Luciani. “Common ones such as type 1 diabetes, multiple sclerosis and coeliac disease are on the rise. Roughly one in four adults have an autoimmune disease and none of them are curable. It’s a really big problem in terms of health conditions for our society.”

In the last 10 years, technology such as imaging and genomics has enabled researchers to dissect the mechanisms that are driving the disease.

“We have been working solidly on the application of data analytics through machine learning and AI to dissect those signatures, the molecular mechanisms that could explain what causes autoimmune disease.”

Machine learning vs AI generative models

Prof. Luciani is one of the founding members of the UNSW Cellular Genomics Futures Institute. The institute has been instrumental in advancing single cell genomics, enabling researchers to study DNA, RNA and proteins, at the single cell level. 

“This provides us with an immense amount of data in the context of autoimmune disease. We use machine learning on data sets for genomics, single-cell technology and autoimmune disease all together,” said Prof. Luciani.

Early on, machine learning techniques had to be developed to identify what was needed.

“For example, we were interested in proteins called T-cell receptors. These are proteins that are a very important part of our immune system used to recognise foreign antigens.  We needed a way to identify these T-cell receptors one at a time in each cell in the context of autoimmune disease. So we had to develop the technology to do that,” said Prof. Luciani.

These approaches combined statistical modelling and machine learning, forming part of the broader family of AI methods that has continued to evolve over the past two decades.

“Large datasets were fed into these models and the models would give predictions. The modelling wasn’t generative, they were learning in an informed way.

“Machine learning still uses techniques that enable you to interpret what the machine does. But in large language models and generative models results can be much more difficult to interpret. It gives you a prediction, but it doesn’t tell you how.”

This is the fundamental difference in Prof. Luciani’s lab. He and his team are still keen to interpret what the model does. They want to predict what’s wrong in autoimmune disease but they also want to understand how.

A gamechanger, understanding the shape of proteins

The Nobel Prizes in Physics and in Chemistry, both awarded in 2024, were a major breakthrough for the field of medical research. These two prizes recognised the mathematical models used in AI and how they have been successfully used to predict the structure of proteins, the building blocks of life.

For Prof. Luciani and his team, knowing the shape of proteins in our body has changed how they study autoimmune disease.

“It has helped us in so many ways. From understanding if you’re born with a mutation what’s the effect of that mutation on that protein, to designing new drugs to help with treatment,” said Prof. Luciani.

Originally the team applied machine learning techniques to genomics data. Today their approach combines these approaches with generative AI models that can predict protein structures and interactions.

“We've been studying T-cell receptors from a genomic perspective for many years. Now we're also studying them from a protein perspective.

“We’re looking at their 3D shapes which is really important because they recognise antigens, foreign bodies, viruses and self-proteins, and that’s where autoimmunity comes in.

“No one know what triggers autoimmune diseases. We are combining AI tools to answer the complex question of what happens when you have an autoimmune disease and how do immune cells go rogue,” said Prof. Luciani.

Looking ahead to deliver research with real impact

In January 2026, Prof. Luciani expanded the lab from UNSW to Westmead Institute of Medical Research (WIMR) where he was appointed to the inaugural role of Director of Functional and Immunogenomics Research Strategy.

“Together with my team, I will help drive a major program of work at the intersection of immunology, genomics, bioinformatics and artificial intelligence, with a focus on improving outcomes for immunocompromised patients and transplant recipients,” said Prof. Luciani.

This opportunity will complement his work on acquired mutations which is “exploding” internationally.

"Some immune cells acquire genetic mutations during a person's lifetime rather than inheriting them at birth. These are known as somatic mutations, and we believe they may fundamentally change how immune cells behave, causing them to become 'rogue' and contribute to autoimmune disease."

This idea of hunting for somatic mutations as a driver of autoimmune disease is now being studied by researchers around the world. Prof. Luciani and his team are world leaders in the field but the technology to generate the data where machine learning tools are needed is very expensive.

“This work has been made possible through an interdisciplinary team we have put together. You need clinicians willing to recruit patients, technical assistants in the laboratory, immunology specialists, genomics people who understand how to instruct DNA and RNA at the single cell level and you need machine learning experts. It’s a very skilled team we had to build,” said Prof. Luciani.

What Prof. Luciani particularly likes about AI is that from an equity, diversity and inclusive perspective, the technology can be very helpful.

“We are all hungry for data, this is a perfect way to stop any kind of discrimination. I’m very confident that AI will be a great solution for making things more equitable,” said Prof. Luciani.

Main image: Martina Bonomi (PhD student), Professor Fabio Luciani, Arman Safavi (Postdoctoral researcher), Michael Li (PhD student), Jun Xing (Research Assistant), Anna Liu (Honours Student), Jacky Xie (Research Assistant), Cher Olivo (Honours student).