Digital Health Podcast- Royal Society of Medicine

Annabelle speaks with physician-scientist and Google research lead Dr. Jake Sunshine about bringing loss-of-pulse detection from research concept to a regulated feature on consumer wearables.
Jake shares the journey from early laboratory experiments to real-world deployment, and how his team designed an evidence strategy where no clear playbook existed—combining simulation, opportunistic clinical validation, and early regulatory engagement. It’s a powerful reminder that innovative technologies often require equally innovative approaches to generating evidence.

In this episode, we explore the emerging phenomenon of AI-induced psychosis and the hidden risks of conversational AI. Dr Josh Au Yeung joins Annabelle to discuss how large language models — designed to engage and please — can unintentionally reinforce harmful beliefs, especially for vulnerable users. We dive into the technical reasons behind this, the societal implications, and the importance of monitoring, education, and safeguards to keep users safe.

YouTube is now one of the most influential sources of health information in the UK.
In this RSM Podcast episode, Annabelle speaks with Vishaal Virani, a clinician leading YouTube’s UK health strategy, about how the platform approaches medical accuracy, tackles misinformation, and elevates trusted sources like the NHS.
We explore how algorithms shape visibility, Youtube's “health shelves” and what they mean for credibility, and what healthcare can learn from digital creators about clarity and communication.
If health journeys now begin on YouTube, we discuss: how do we safeguard medical truth at scale?

As AI in healthcare becomes more autonomous, the question of who is legally responsible when it makes mistakes becomes urgent. In this episode, Annabelle speaks with Majid Hassan to explore the liability gap in the UK and the practical steps organizations and developers can take to protect themselves.
Topics covered include:
The difference between clinical negligence and product liability

Why fully autonomous AI complicates accountability, especially when clinicians are taken out of the decision loop

The concept of a “liability gap” in UK law, and how proving defect or causation in AI is difficult

Practical steps for healthcare organizations: vendor due diligence, compliance roadmaps, audit trails, staff training, and insurance coverage

Steps for AI developers to demonstrate due diligence, risk management, and adequate insurance

Lessons from the EU’s revised product liability directive and its potential implications for UK law

Autonomous AI is often described as a red line in healthcare — something that should only ever assist clinicians, never act independently. But is that really where we are today?
This is the first episode in a series examining autonomous AI. Annabelle is joined by regulatory expert Dr Hugh Harvey to explore the foundations of autonomous AI in healthcare. Together, they unpack what autonomy actually means, where it already exists in clinical practice, and why current regulatory frameworks struggle to address systems that operate without a human in the loop.
The conversation covers risk classification, clinical evidence, post-market surveillance, liability, hospital governance, and why regulatory approval alone is not enough to guarantee trust or adoption.