One size fits no one: why women’s health could be the defining test for AI in healthcare 

I recently joined leaders from healthcare, technology, research and investment at the Milken Institute Asia Summit panel AI’s Health Care Promise: Getting It Right for Everyone. It was a timely discussion about how we can harness the benefits of artificial intelligence while ensuring it improves care for everyone, not just some. 

One theme stood out: the question is no longer whether AI will play a role in healthcare. It already does, but the challenge is making sure it works safely, fairly and transparently for everyone. 

That mirrors what we heard through the National Commission into the Regulation of AI in Healthcare. More than 12,000 patients, clinicians, innovators and healthcare leaders told us they see the potential of AI to improve diagnosis, treatment and access to care.

Their support, however, depends on trust. People want AI to be safe, transparent and accountable, while keeping clinicians at the heart of decisions about their care. 

Perhaps the most interesting insight was the difference in trust between women and men. Research from the Health Foundation found that only 32% of women believe AI will improve healthcare quality, compared with 46% of men. Women were also more likely to say AI may never be safe enough for use in the NHS. 

That’s valuable intelligence. For too long, many women have experienced delayed diagnoses, under-recognition of symptoms and healthcare systems that have not always been designed around their needs. It is no surprise that they ask tougher questions about safety, transparency and accountability. AI must earn their trust.  

Women’s health also represents one of the greatest opportunities for AI. For decades, they have been underrepresented in research, clinical trials and health datasets. AI gives us the opportunity to identify those gaps and help close them. 

I’ve often said that one size fits no one. Healthcare has traditionally been built around averages, yet we know women can experience symptoms, disease progression and responses to treatment differently. AI has the potential to move us beyond those averages and towards more personalised care. 

That opportunity depends on equity being built into the system from the start. 

Algorithms trained on incomplete or unrepresentative data risk reproducing existing inequalities. Systems built on diverse datasets, tested rigorously and monitored continuously could help tackle some of healthcare’s most persistent gaps. 

That is why equity must be a golden thread running throughout the AI lifecycle. We need to ask not only whether a system works, but who it works for, who it works less well for and how we act on these potential health inequalities. This is why it is exciting to hear about sex-specific algorithms, sex-based data disaggregation and a recognition that researchers, clinicians and the public are aligned on getting this right for all. 

The public also sent a clear message about the need for human oversight. People do not want AI replacing their clinicians. They want AI to free up clinicians to deliver better care. Trust increases when healthcare professionals remain involved and falls when human judgement is removed. 

Healthcare’s future is about combining the strengths of people and technology. 

The same applies to regulation. Good regulation creates confidence, and confidence drives adoption. That is why the Commission called for a regulatory system that is fast, safe and trusted, with continuous monitoring as AI evolves in real-world use. 

We already know much of what needs to be done: representative data, patient involvement, transparency and ongoing learning. 

Now we need action. For implementation of the National AI Commission to have fidelity to the vision, we need to move from being a think tank to being a do tank, bringing together researchers, technology companies, regulators, policymakers, healthcare leaders and patients to turn knowledge into change. 

The goal is simple: a healthcare system where no woman waits years for a diagnosis because her symptoms do not fit a traditional model, where AI supports earlier diagnosis and more personalised care, and where trust is earned through transparency and accountability. 

If we get this right, AI could help address some of the most persistent inequalities in healthcare. 

The gaps in women’s health remain one of the greatest unmet needs in healthcare. Closing this gap could also become one of the greatest success stories, when we accept that it’s not a niche application for AI but one of the clearest tests of whether we can use it to deliver equitable healthcare at scale.