Beneath the streets of London, on a sunny April day, the brightest minds in healthcare gathered at the Outernet to debate where medicine is heading. Here is what the industry needs to take away
Words by Isabel O’Brien
1. The real driver of AMR
Lord Ara Darzi, Director of the Institute of Global Health Innovation, opened with a bombshell that cast the AMR crisis in a new light. While the dwindling antibiotic pipeline is partly to blame, with the companies developing antibiotics falling from 20 to just 12 in the last two decades, he also pointed to diagnostic failings.
A clinician suspecting sepsis has around one hour before the death risk rises, while a definitive microbiological diagnosis can take between 24 and 72 hours. That often leaves doctors with one option: guess intelligently. “That guess, repeated a billion times a year, is the engine of resistance. It is not negligence. It is not ignorance. It is the inevitable consequence of building medicine on a diagnostic infrastructure that was never fast enough.” The approach save lives, but could be claiming them in tandem.
Darzi added that AI diagnostics are already reaching 99% accuracy at the point of care. Therefore, the real fix, he suggested, may not be faster laboratories, but upgrading healthcare infrastructure so AI tools can be integrated more widely.
2. AI drug discovery: plenty to go around
Reid Hoffman, LinkedIn co-founder and investor in AI companies, made the case that AI drug discovery is not a “winner-takes-all” market. Unlike the software sector, and perhaps even more traditional R&D, he argued the landscape has room for multiple winners across different diseases and approaches. “Human bodies have many issues,” he said, welcoming as many players into the space as possible.
He was more critical of how slowly AI is being adopted in clinical settings. Hoffman argued that any clinician not using AI as a second opinion is “bordering on malpractice”. He referenced a friend who was misdiagnosed at a US hospital, used AI to seek a second opinion and switched care team. They were later told they would have died within two hours without intervention. If that view takes hold, the knock-on effects for how healthcare is delivered could be far-reaching.
3. Why health misinformation is hard to crack
Dr Deborah Cohen, author and Senior Visiting Fellow, LSE Health, offered one of the clearest explanations of the day, detailing why health misinformation is so insidious.
Today, influencers build parasocial relationships with viewers, through car-seat chats, replies to comments and day-in-the-life videos, creating genuine feelings of familiarity and trust. The result is patients arriving at clinics influenced not by strangers online, but by people they feel they know personally.
One US OB-GYN told Cohen that the first 10 minutes of a 20-minute appointment are now spent undoing what patients have seen online. Regulators, she noted, “are interested in sickness – they’re not interested in wellness”, leaving much of the online health space largely unchecked. Meanwhile, 55% of Americans now turn to social media for health information.
4. What if the next depression treatment isn’t a drug?
Kathleen Fisher, Chief Executive Officer of the Advanced Research and Invention Agency (ARIA), and ARIA programme director Jacques Carolan touched on one of the most exciting scientific frontiers in neurology. Deep brain stimulation, originally developed for Parkinson’s disease, has already shown potential in treating depression, addiction and epilepsy.
Backed by a £69 million investment, ARIA is now testing whether brain circuits can be modulated non-invasively using wearable devices, removing the need for surgery altogether. Fisher and Carolan said it would take around seven years to reach proof of principle. But if successful, the implications for drug-based treatment models in depression, addiction and chronic pain could be significant.
5. Why AI integration in healthcare will take time
Ming Tang, Chief Data and Analytics Officer, NHS England, had a simple answer to why AI has not yet transformed healthcare: the data is not ready. “You cannot drive AI unless you’ve got an integrated dataset.”
In healthcare systems like the NHS, patient data is often stuck in disconnected silos with incompatible standards and infrastructure. The NHS’s federated data platform, which began rolling out in 2024, is an attempt to solve that challenge, giving innovators something solid to build on rather than fight their way into.
But Tang pointed to a second problem that infrastructure alone cannot fix. The NHS produces world-class health innovators, including in AI, only to lose many of them overseas. “We have such fantastic innovators in the NHS, but they all go abroad because they can’t quite get the traction here.” Fix the data infrastructure, she argued, and you also begin to fix the conditions pushing top talent elsewhere.
