The Digital
Colleague

Innovation journeys in health are rarely short or quick. On the day ByMany meets the team, Frazer, who has worked in medical imaging for three decades, tells us that she first became inspired by the possibilities of automating breast cancer screenings more than 20 years ago while working in the United States. ‘It’s only recently that we are finally seeing the longitudinal datasets and computing power we need become readily available.’ Gerda Evans, who joins us as a patient advocate, was diagnosed with breast cancer in 1989 and has been working in the community ever since. Both also speak with deep fondness and respect for John Hopper from the University of Melbourne, who sadly passed in 2024 but was a key team member and is remembered as a scientist passionate about the evidence, always willing to challenge preconceived ideas and let the data tell the story. We are joined by Katrina Kunicki, who brings both an understanding of radiology and an infectious startup business mindset, as well as energetic AI and data specialists Carlos Peña-Solorzano, Michael Elliott, and Davis McCarthy, all from the St Vincent’s Institute of Medical Research.

What emerges from our conversation is a portrait of collective achievement in action. Rather than a lone genius or individual breakthrough, there is evidence of diverse expertise—from radiography to business strategy, from clinical practice to AI development—combining to tackle one of healthcare’s most pressing challenges. The team’s approach reflects a deeper truth about innovation: when facing truly complex problems, no single perspective suffices. The radiographer-turned-MBA, the clinical leader with technological optimism, the data scientists seeking domain expertise – each brings essential pieces to the puzzle. And the full picture they are trying to put together: to save more lives with better accuracy, and faster, more efficient breast cancer detection for all women.

The challenge is a sobering one, centred around a disease with which many of us are likely to have a connection. Breast cancer is the most common cancer diagnosed in Australian women and the second leading cause of cancer death. The great hope, however, lies in the likelihood that if medical professionals can catch it early in the breast, before it spreads to lymph nodes or distant organs, the chances of a cure are very high. Early detection is vital, but current screening is a ‘one-size-fits-all approach’ with inaccuracies, growing demand, and reliance on highly trained humans. Radiology staff currently use a process called batch reporting, where they often read 150 mammograms in 150 minutes, expertly looking for small, subtle changes and signs of early cancer. Ninety-five per cent of those are normal, so a highly trained expert is sitting there marking images as normal, normal, normal, with the occasional evidence of cancer being flagged. Uniquely, to improve accuracy in breast cancer screening, there are always two radiologists reviewing each mammogram, and if they differ in their assessment, it goes through to a third. It is therefore always ‘diagnostics as a team’.

The hypothesis is that AI can join this team, providing at least some layer of screening and detection more quickly and affordably. Michael builds on this, making a strong case for AI to be part of the process, particularly in the initial screening rounds. ‘Often, the discussion is around AI taking more of the menial tasks away from humans. It just happens that in this field, the menial task is still a highly skilled one. But we feel that in the early stages, where perhaps only 1–5% of those images have any risk of breast cancer, AI can perform very well, and as we refine down to the point where 50% of those images are showing cancer risk, and we want human engagement, we can get far more efficient.’

Davis is then quick to point out other benefits of the AI teammate, noting that where humans, constrained by time, are often giving binary readings – ‘detected’ or ‘not detected’ – AI can ‘give a score for each image between 0 and 1. What we are finding is that there is a lot of information in that score, which is a game-changer.’ The team has discovered that the AI score, if below the threshold for cancer, is a better predictor of future cancer risk than any known measures. Helen now envisions a personalised approach to screening based on AI risk scores that can save many lives. ‘We are simulating new screening protocols with AI risk scores where, for example, women in the top 2% of AI risk scores receive a personalised prescription that might include annual screening or use of contrast mammography or MRI.’

The possibilities are very exciting, though Katrina, who is more used to the speed of the startup world, offers a sobering reality check: ‘With services like cancer detection, we must acknowledge the need for an appropriate balance of validation and risk management.’ In healthcare, where patient safety intersects with technological innovation, every step forward requires careful navigation through regulatory frameworks, clinical validation, and organisational change. Of course, behind the algorithms and implementation challenges are real people whose lives could be transformed by earlier, more accurate detection. There is a rush to expedite technology, but a patience to do so in a deeply human way. Cancer survivor and community advocate Gerda is part of the group that supports this fine balance. In teams navigating uncertainty, acknowledging doubts and challenges is as important as maintaining optimism. It’s this juggle—enthusiasm tempered by realism, vision grounded in practical experience—that enables sustainable innovation.

Despite the challenges, or perhaps because of them, the team’s enthusiasm remains undimmed. Helen’s optimism is contagious: ‘I think there’s never been a more exciting time in healthcare. We now have the ability to do incredible things and to really shift the needle and advance the human condition.’ This shared sense of purpose is the team’s North Star, and Katrina echoes this sentiment: ‘That’s what keeps me going and why I came to the project, because I believed in the problems we were trying to solve. When the going gets tough, you’ve got to focus on that.’ ‘We’ve all learned so much,’ Helen continues, ‘even if we were to stop now.’ But stopping isn’t on the agenda. The team’s commitment to continuous learning exemplifies how groups navigate uncharted territory, not through perfect planning, but through iterative discovery over long periods. This learning extends beyond technical knowledge. It encompasses understanding how to implement AI in clinical settings, how to manage change in complex healthcare systems, and how to maintain momentum when progress feels glacial.