University of Queensland
Using artificial intelligence (AI) to prevent mental illness
I am an academic researcher with training in clinical psychology, neuroscience, engineering, and medical neuroimaging. My work uses mathematical and engineering approaches such as graph theory and artificial intelligence (specifically machine learning) to predict who is at risk to develop a mental illness. I am particularly interested in complex trauma, treatment-resistant depression, and mental illness resulting from brain injury such as stroke or concussion. I use magnetic resonance imaging (MRI) scans from people with mental illness as input for machine learning. The computer then trains itself to identify patterns in the scans that can accurately classify people into those with and without a mental illness. AI can detect subtle forms of these patterns in people who are at risk but have not yet developed a mental illness. These people can then benefit from preventative interventions, which is much more effective than treating people already affected by mental ill-health.
Mental illness is a significant health issue in Queensland, as it is in many parts of the world. According to the latest available data from the Queensland Government, around 20% of Queenslanders experience a mental health condition each year. This includes conditions such as depression, anxiety, bipolar disorder, schizophrenia, and eating disorders, among others. According to a report by Deloitte Access Economics, commissioned by the Queensland Mental Health Commission in 2020, mental illness cost the Queensland economy around $13.5 billion in 2018-19. This included direct costs of $3.2 billion and indirect costs of $10.3 billion. The report also found that the burden of mental illness falls disproportionately on younger people and those living in regional and remote areas of Queensland. For example, young people aged 15-24 accounted for around 40% of the economic burden of mental illness, while people living in remote and very remote areas accounted for around 13% of the burden. The report highlights the need for increased investment in mental health services and support, as well as a focus on prevention and early intervention to reduce the economic burden of mental illness in Queensland. This is where my research comes in: Most physiological illnesses can be identified through tests, such as blood samples, MRI scans or physical examinations. However, for mental illness, we still rely on self-report measures and outdated criteria to diagnose patients. This is extremely subjective and often does not lead to accurate diagnoses, so that patients end up having to see several clinicians until the correct diagnosis can be made. Even once someone receives the appropriate diagnosis, treatment is often time-intensive, expensive and does not work for all patients. It is therefore extremely important to diagnose patients early, if possible even before they develop severe symptoms. To date, we do not have any reliable biomarkers to do this. AI is an exciting new tool, which is able to detect patterns and information in neuroimaging data that we were previously unable to identify. What motivated me to start this research was to use this novel technology to help more people with mental illness, cut the costs for them, and to receive better, early treatments. While I am currently working with clinicians and clinical populations in Queensland, I am hopefully that this technology will eventually help people all over the world.
I am currently leading a small group of undergraduate, honours and PhD students. The majority of these student are young women from diverse cultural backgrounds and sexual identities. I make it a priority to teach my students coding and basic machine learning approaches and openly encourage them to ask questions and be persistent when they feel they get stuck with their code. In my experience, students excel rapidly if they feel safe enough to make mistakes. Making mistakes is so important to learn and develop, make truly ground-breaking discoveries, and build self-esteem in a still very heavily male dominated space. My goal is to teach my students in such a way that they become independent and confident so that they will soon exceed me in STEM-related skills, and I am happy to say that a lot of them are well on track to achieve this. I also believe that it is incredibly important to share knowledge by making code and data available for other people to use and I encourage my students and others at conferences or scientific meetings to do so. While I am a strong believer in sharing knowledge, I have an even stronger passion for equity, diversity and inclusivity. I identify or belong to several marginalized groups myself, which has made my journey incredibly difficult. I am trying to be a role model for my students and other women in science by openly talking about my struggles and providing a supportive environment for everyone to share their experiences. I am convinced that many women today still struggle in STEM fields because they do not feel safe enough to express themselves. They are made to believe that they have to fit into certain stereotypes. I believe there is nothing more toxic for the future of science and technology. People’s diversity can teach us much more than outdated norms, and STEM can only benefit from diverse perspectives and experiences. I teach this as part of many domestic, institutional, and international equity, diversity and inclusivity committees and have great success in motivating young women to join these initiatives and communities.
I am excited to tell you about several STEM promotion or engagement activities that I have undertaken over the last couple of years: I am the chair of the Brain Art committee at the Organization for Human Brain Mapping (OHBM), the largest international conference for human neuroimaging studies. Every year we organise a large Exhibition at the conference that brings together scientists and artists interested in the brain. The exhibition is a highlight and has been growing exponentially ever since we started it in 2016. Exhibiting artists are invited to attend the conference and engage with the scientific community. This has led to several successful projects such as the publication of an art catalogue, scientific publications on the importance of integrating art and science, and the commercial use of the art by external stakeholders. As part of the competition, we have an area for children, where they can colour in images of the brain, build brain models and work on puzzles that teach them different areas of the brain and what their functions are. This has been a huge success with parents and children alike, which has led to an increase of people attending the conference with their children. We have even started to provide childcare and educational courses for children, to allow more women to attend the conference.
On a local level, we run the annual Australian Brain Bee Challenge at the University of Queensland. It is a competition for high school students in year 10 to learn about the brain and its functions, learn about neuroscience research, find out about careers in neuroscience and to dispel misconceptions about neurological and mental illnesses. Winners get to spend some time in a research laboratory, which I have hosted in the past. This is an exceptional opportunity to engage with schools and high school students. Many of these students tell us that they did not expect that a career in STEM can be so exciting, and many stay in touch for us for research projects.