The Hurdles Ahead: Navigating the Challenges of AI in Healthcare

While the promise of AI in healthcare is immense. But for every opportunity, there are equally significant hurdles. Leaders across the system know the reality: the journey from innovation to impact is not straightforward.

At our recent AI in Cancer Research panel, the conversation turned to the roadblocks standing between potential and progress. Here, we recap the critical challenges slowing AI's healthcare revolution.

Unsurprisingly, many of these hurdles revolve around data, trust, and the intimate nature of healthcare. The insights aren’t about algorithms or data models, but about leadership, because these challenges can’t be solved by technology alone.

The Data Dilemma: Silos, Fragmentation, and Quality

Maybe the biggest hurdle to AI's advancement in healthcare is the fragmented nature of health data, which is currently "locked up in a million different silos". Creating a fragmented system that is inefficient and often unsafe.

Even in advanced precision medicine programs, processes remain "extremely labour-intensive" due to the need to securely manage and analyse vast amounts of sensitive, identifiable data like genomes.

The quality of data is paramount; our panel stressed that poor quality or biased training data leads to poor outcomes. Real-world healthcare data is "messy" and often "not perfect," making it difficult to train AI tools effectively outside of highly curated research environments.

For leaders, the challenge is clear: building the infrastructure and governance that allows data to flow securely across systems, while maintaining quality. Without this, the AI revolution will remain stuck in pilot projects instead of transforming patient outcomes at scale.

Trust, Ethics, and Regulation

Technology only works if people trust it. The deployment of AI tools requires a strong foundation of trust and robust ethical frameworks.

Large-scale AI failures, such as Epic Systems' sepsis prediction failing in 67% of cases and Google's Verily struggling to read 21% of diabetic retinopathy images, highlight the dangers of premature or flawed implementation. And build fear and distrust with the public.

Leaders must ensure a “human in the loop” approach, where clinicians remain accountable and patients stay informed. Consent, transparency, and clear ethical frameworks are not optional. They are prerequisites for adoption.

In Australia, regulation lags behind. While ethical principles exist, robust frameworks tailored for healthcare do not. This leaves innovators uncertain and investors wary. Encouragingly, national efforts are underway to train ethics committees to be “AI-ready,” but leadership from both industry and government is essential to accelerate progress.

Systemic Resistance and Risk Aversion

Perhaps the toughest hurdle is cultural. Healthcare organisations are naturally risk-averse, especially when patient data and lives are at stake. Approvals for AI initiatives move slowly, and building secure environments demands significant upfront investment.

But there’s also a deeper misalignment. Too often, health policy is built around cost containment rather than optimising care. AI has the potential to lower costs and improve outcomes, but if leaders remain locked into short-term cost control, progress will stall.

The leadership challenge here is reframing the conversation: shifting from defensive cost management to proactive investment in systems that deliver both quality and efficiency. It requires courage to change entrenched mindsets, but without it, healthcare risks missing the opportunity AI offers.

The Path Forward

While AI holds incredible promise for healthcare, the path forward is complex. Addressing data fragmentation, building robust ethical and regulatory frameworks, fostering public trust, ensuring high-quality training data, and overcoming systemic risk aversion are key.

Success will not come from technology alone. It will come from leaders who:

  • Build the infrastructure for secure, high-quality data.
  • Champion transparent, ethical frameworks that earn public trust.
  • Push for regulation that enables innovation while safeguarding patients.
  • Reframe investment decisions around outcomes, not just costs.

The insights from this event underscore the need for a cautious yet determined approach, prioritising responsible development and collaborative efforts to unlock AI's full, safe potential in revolutionising healthcare.

Thanks to our panel for sharing their thoughts:

  • Mark Cowley – Deputy Director, Children’s Cancer Institute & Associate Professor, UNSW Medicine
  • Bill Petch – Chief Executive Officer, Crohn’s Colitis Cure
  • Lloyd Prescott – Chief Executive Officer, Southern Star Research
  • Dr. Yagiz Alp Aksoy –Clinical Fellow, Biomedical AI Centre, Centenary Institute & Doctor at the Royal North Shore Hospital
  • Erin Evans, CEO of Intelligen