3 min read

·

Jun 5, 2026

--

Share

Press enter or click to view image in full size

Photo by Bermix Studio on Unsplash

In 2019, a study published in The Lancet Digital Health described how a machine learning model predicted the onset of atrial fibrillation — an irregular heart rhythm that causes strokes — from smartwatch data, before patients had any symptoms at all.

That’s not science fiction. That’s a feature that now ships with consumer wearables.

We are living through a genuinely strange moment in medicine. The devices people carry in their pockets or strap to their wrists are collecting physiological data continuously — heart rate, sleep architecture, blood oxygen levels, even electrocardiograms in some models. This data stream, processed correctly, has real clinical potential. Processed incorrectly — or handed to people without the context to interpret it — it also has real potential to cause harm.

Consumer health wearables are good at detecting trends. If your resting heart rate has been quietly climbing over three months, that’s signal worth paying attention to. If your sleep data shows you’re getting less deep sleep than you were six months ago, and you’ve been feeling more fatigued, that correlation is worth bringing to a doctor.

What they’re not good at is diagnosis. A heart rate spike during a stressful meeting looks identical to a cardiac arrhythmia in raw data. A low oxygen reading from a poorly positioned sensor looks identical to one from genuine hypoxia. The technology generates data; it doesn’t generate meaning. That part still requires a human — ideally one with clinical training — to interpret it in context.

This is the gap that the digital health industry is still figuring out. Apps that present health data without clinical framing are, in some cases, generating unnecessary anxiety. There’s a growing body of evidence on “cyberchondria” — health anxiety amplified by excessive engagement with health-tracking technologies — and it’s a real clinical concern.

The strongest evidence for digital health tools is in chronic disease management. For patients with type 2 diabetes, continuous glucose monitors paired with structured coaching have produced outcomes that rival or exceed conventional management in several trials. For hypertension, app-based medication reminders and blood pressure logging have improved adherence significantly in studies conducted across sub-Saharan Africa, where the burden of uncontrolled hypertension is among the highest in the world.

These tools work because they extend the reach of the healthcare system into the spaces where patients actually live. A consultation happens once every few months. An app is available at 2am when a patient is wondering whether their symptoms warrant attention.

Here’s the version of this conversation that rarely makes it into mainstream tech journalism: digital health tools, as they currently exist, are largely designed for people who already have good healthcare access. The countries with the most to gain from digital health — where clinician-to-patient ratios are the most stretched, where chronic disease burden is rising fastest — are also the ones where smartphone penetration is lower, internet connectivity is less reliable, and the regulatory frameworks for digital health products are least developed.

This is not a technology problem. It’s a design and political will problem. The technology to build offline-capable, low-data health tools exists. The clinical expertise to design them appropriately exists. What’s missing is sustained investment in building them for the places that need them most.

That gap is closing. But slowly.