Digital Health Grows—But Policy, Equity, and “Real-World” Evidence Must Keep Up

This week’s research takes a close look at digital health and health technology from multiple angles: who gets to participate in pediatric trials, how well AI-based screening might work at scale, whether wearable sensors deliver usable “real-world” data, and how telemedicine adoption can widen or narrow gaps. The common theme is not simply whether innovations work—it’s whether systems can deploy them fairly, reliably, and in ways that match everyday life.

Guardian permission waivers: a small policy lever with big downstream effects

A review of 49 studies looked at how researchers handle parent or guardian permission when testing digital mental health tools with young people. About 4 in 10 studies allowed youth to participate without guardian permission, most often to protect their privacy. These studies were more likely to recruit participants online and include gender identities beyond male and female, but they also had lower follow-up rates. Because studies with and without guardian permission were otherwise fairly similar, the findings suggest that review boards may be applying the rules differently. Clearer guidance could help researchers balance youth privacy, safety, and access to research.

AI esophageal cancer screening: promising scale-up—plus measurement of tradeoffs

Researchers developed an AI tool called EAGLE that can detect signs of esophageal cancer and precancerous lesions on routine chest CT scans. Tested with more than 80,000 patients across 12 medical centers in three countries, the tool detected most cancers while keeping false alarms low. This could allow health systems to use CT scans already being performed—such as for lung cancer screening—to identify people who may need further testing. The next challenge is making sure the technology works well across different populations and that patients who are flagged can access timely follow-up care.

Wearable gait sensors: strong methods, weaker “real-world” transferability

A review of 43 studies examined how well wearable sensors measure walking in people with Parkinson disease or stroke during everyday life. Researchers found that wearables can provide useful information, but studies often measured walking in very different ways, making their results difficult to compare. Evidence for people with stroke was especially limited. The findings suggest that wearables have promise for monitoring people outside the clinic, but researchers need more consistent measures and testing standards before the data can be widely used to guide care.

Telemedicine adoption and measurement of the digital divide

A nationwide survey in Poland found that about half of respondents had used telemedicine in the past year, most often through phone calls. Digital services such as e-prescriptions were widely used, while video visits and health apps were much less common. However, people with more education and those living in large cities were more likely to use telemedicine. The findings suggest that digital health may make care more convenient for people already well connected, while leaving some rural and less digitally connected communities behind.

The system lesson: digital health needs guardrails, standards, and inclusion

Across these studies, the policy and equity takeaway is consistent. Digital health progress depends on regulatory clarity (guardian permission waivers), careful calibration and tradeoff tracking (AI screening performance), standardized evidence methods that hold up in everyday settings (wearable gait biomarkers), and deliberate inclusion strategies (telemedicine adoption patterns). When health systems treat these as system design problems—not afterthoughts—public health can better ensure innovations translate into benefits for everyone, not just those who already have access.