Designing health technology people can trust

When someone opens a health app, they are often carrying a question they have not yet asked anyone else. It might be about a late period, a symptom that will not settle, or a pregnancy they confirmed that morning. What they type in that moment is some of the most personal information they will ever give a piece of software.

That is why we think trust in health technology is won or lost in places most people never see. A calm colour palette and a friendly tone help, but they do not decide whether someone comes back. Three things do. The first is how the product handles their information. The second is how honest it is about what it knows. The third is whether it points them to a person when a person is what they need.

Data practices are part of the product

The World Health Organization made this point plainly in its first guideline on digital health interventions, published in 2019. People need to be assured that their data is safe, and that looking for information on sensitive topics such as sexual and reproductive health will not put them at risk. For a product in women's health, that is close to the whole brief.

The record across the industry gives people reason to be careful. A peer-reviewed analysis of 30 menstrual tracking apps, with nearly 200 million downloads between them, found that their privacy policies often did not cover the reproductive data the apps depended on to work. A user could read every word of a policy and still not learn what happened to the information that mattered most to them.

So we treat data decisions as design decisions, made early and reviewed often. In practice that comes down to a handful of commitments.

• Ask only for information a feature actually needs, and explain why at the moment you ask.
• Put privacy settings where people can find them, write them in plain language, and make them easy to change later.
• Keep the policy and the product in step, so that what the app does matches what the policy says.
• Treat deleting data as something people can do themselves, not a request they have to chase.

The Principles for Digital Development, endorsed by more than 300 organisations working in global development, name this directly as one of nine principles: "Establish people-first data practices". We find it a useful test. If a data decision would surprise the person it affects, it probably needs another look.

Be honest about what the product knows

Health rarely fits a template. Cycles vary from month to month and from person to person. Symptoms overlap. Every pregnancy follows its own course. A product that presents every prediction in the same confident tone teaches people to trust it most in exactly the cases where it is weakest.

We would rather a product said less and meant it. A cycle prediction can be shown as a likely window instead of a single date. An unusual entry can be logged without the app raising an alarm or brushing it aside. Educational content can describe common experiences without implying that anything different is a problem, and it can say clearly when a change is worth raising with a health professional.

None of this is technically hard. It does require a team to decide, on purpose, that being accurate about uncertainty matters more than looking impressive in a demo. That decision has to be made again with every new feature, because the pressure to sound certain never goes away.

Know when to hand over to a person

Digital tools can extend care. They cannot be the whole of it. In the same 2019 guideline, WHO described telemedicine as a valuable complement to face-to-face care that cannot replace it entirely.

For a health product, the hand-over is one of the most important moments to design. When should the product suggest that someone speaks to a midwife, a doctor or a pharmacist? How does it say so without frightening them, and without hiding the suggestion in small print? What happens after they make contact, and does the product make the next step easier or harder?

We think these questions deserve the same care as any sign-up flow. A product that knows its limits and makes the next step simple is more useful than one that tries to answer everything itself. It is also safer.

This is why Obaa+, the women's health app we are building, pairs its tracking and education features with structured consultations with certified midwives. Information helps people understand what is happening. A qualified person who can listen to the specifics of their situation helps them decide what to do about it.

Trust is something a team practises

Trust built carefully in the interface can be undone by a single careless release. That makes it an organisational habit as much as a design quality.

The NHS service standard, which sets expectations for digital health services in England, includes three points that could summarise this article. Teams should "create a secure service which protects people's privacy", "make your service clinically safe", and "define what success looks like and be open about how your service is performing". None of these is finished once and ticked off. Each needs an owner and a regular review.

The questions get sharper as health products add automated features. WHO's guidance on the ethics and governance of artificial intelligence for health asks that ethics and human rights sit at the heart of how these technologies are designed, deployed and used. That is a high bar, and the right one. Any feature that interprets someone's health data on their behalf should be held to it.

At Luna, we try to meet that bar by keeping healthcare thinking, product design and engineering in the same conversation from the start. A privacy question should not wait until the end of a build. A clinical question should go to people qualified to answer it.

An open conversation

We do not expect any team to get trust right on the first attempt, including ours. Obaa+ is still in development, and we expect our thinking to sharpen as we test it with the women it is for.

If you are building, funding or delivering digital health services and working through the same questions about data, uncertainty and the hand-over to care, we would like to compare notes. Get in touch with the Luna team.