Future of Medicine: Digital Health and Wearables in Clinical Care

Watches that record an ECG, rings that track sleep, patches that measure glucose every few minutes, and home blood pressure cuffs that upload readings to a clinic: digital health has moved from the hospital into patients' pockets. For clinicians, this creates both opportunity and noise.

Wellness gadget or medical device?

The first question about any wearable feature is its regulatory status. In the U.S., low-risk products intended only to maintain or encourage a general state of health, such as tracking activity or sleep for fitness, are generally treated as general wellness products, not medical devices. Features intended to diagnose, treat or monitor disease fall under device regulation.

This means a single consumer device can contain both:

• Wellness features (step counts, general sleep estimates) with no medical claim.
• Cleared medical features (for example, an ECG app or irregular rhythm notification) with a defined intended use and evidence behind it.

The evidence: atrial fibrillation as a case study

The best-known large study of a consumer wearable is the Apple Heart Study from 2019. It enrolled a very large number of participants who used a smartwatch app that analysed pulse data for irregular rhythms.

Key findings:

• Only a small proportion of participants received an irregular pulse notification.
• Among notified participants who returned an ECG patch, atrial fibrillation was confirmed in about a third.
• When a notification and a simultaneous patch recording coincided, the notification usually reflected atrial fibrillation on the patch, giving a high positive predictive value for that comparison.

What the study did not show is that screening the general population with smartwatches improves outcomes such as stroke prevention. It was a feasibility and accuracy study, and participants were self-selected. Notifications also led to clinic visits, which carry costs.

The broader lesson applies to every wearable: a detection feature is a starting point for clinical assessment, not a diagnosis, and its predictive value depends on who is wearing it.

Glucose sensors go over the counter

Continuous glucose monitors (CGMs) transformed diabetes care for people using insulin. In March 2024, the FDA cleared the first over-the-counter CGM, intended for adults 18 and older who do not use insulin, such as people with type 2 diabetes managed without insulin and people without diabetes who want to understand how diet and exercise affect glucose. The FDA noted it is not intended for people with problematic hypoglycaemia.

This shift raises new questions for clinicians:

• How to interpret normal post-meal glucose rises in people without diabetes.
• How to avoid unnecessary anxiety or restrictive eating.
• When CGM data should prompt formal diagnostic testing such as HbA1c.

Health systems: from devices to models of care

Wearables matter most when they are part of a clinical pathway rather than an isolated gadget.

Virtual wards and remote monitoring

NHS England's virtual wards allow patients to receive hospital-level care at home, supported by apps, wearables and other medical devices such as pulse oximeters, with clinical teams reviewing the data and visiting when needed. Typical uses include frailty, acute respiratory infection and early supported discharge.

Research cohorts

NIH's All of Us Research Program collects wearable data from participants alongside genomic and health record data, aiming to understand how activity, sleep and heart rate relate to health across a diverse population.

Global strategy

WHO's global strategy on digital health encourages countries to adopt digital health in ways that are people-centred, evidence-based, secure and interoperable, and that improve equity rather than widen gaps.

What wearables can and cannot do today

Risks to manage

• False reassurance and false alarms. Both lead to harm: missed disease or unnecessary tests.
• Alert fatigue. Streaming data from many patients can overwhelm clinical teams unless thresholds and triage are designed carefully.
• Equity. Devices cost money and require smartphones and connectivity. Measurement accuracy may also vary across skin tones for optical sensors.
• Privacy. Consumer data may sit outside health data protections, depending on jurisdiction and terms of service.
• Responsibility. Who reviews the data, how quickly, and who is liable if something is missed must be agreed before a service starts.

A clinician's guide to patient-generated data

1. Identify the device and feature. Is the feature cleared for the purpose, or a wellness estimate?
2. Check the context. Symptoms at the time, activity, device fit and quality of the recording.
3. Confirm with standard tests. A 12-lead ECG, laboratory glucose or HbA1c, clinic blood pressure or ambulatory monitoring.
4. Use trends, not single values. Wearables are often better at showing change over time than absolute accuracy.
5. Document and communicate. Record what data were reviewed and explain clearly what they do and do not show.
6. Agree boundaries. Tell patients whether and how you will review data they send, and what to do if they feel unwell.

Links to telemedicine and AI

Wearables are the sensors of remote care; telemedicine is how clinicians act on the data. See telemedicine after the pandemic boom. AI algorithms increasingly interpret wearable signals, so the validation and bias questions in AI in diagnostics apply here too, as does the broader overview in AI in medicine in 2026.

What students should focus on

The exam-relevant skills are the classic ones, applied to new data sources:

• ECG interpretation, especially atrial fibrillation and its management.
• Diabetes diagnosis criteria and glucose physiology.
• Screening principles and predictive values in low-prevalence populations.
• Pulse oximetry physiology and its limitations.

The MediFlow app includes exam-style questions on these core topics, so you can build the knowledge needed to interpret whatever device a patient brings.

A worked example: interpreting an irregular rhythm notification

Imagine a 45-year-old patient with no symptoms who brings a screenshot of an irregular rhythm notification. A structured approach keeps the response proportionate:

1. Clarify the feature. Was it a notification based on pulse data, or a single-lead ECG recording? Was the feature cleared for this use and for the patient's age?
2. Take a history. Palpitations, breathlessness, syncope, alcohol intake, thyroid symptoms, sleep apnoea symptoms and family history.
3. Assess risk. Blood pressure, diabetes, previous stroke or transient ischaemic attack, heart failure and vascular disease all affect management if atrial fibrillation is confirmed.
4. Confirm. A 12-lead ECG, and if that is normal, ambulatory monitoring when clinically appropriate. The Apple Heart Study shows why: only about a third of notified participants who wore a patch had atrial fibrillation confirmed during monitoring.
5. Explain. Tell the patient what a notification can and cannot mean, and what symptoms should prompt urgent care.

The same logic, identify the feature, contextualise, confirm, communicate, works for glucose traces, oxygen saturation readings and home blood pressure logs.

Questions to ask before launching a wearable-based service

Health systems and practices introducing remote monitoring should answer these questions in advance:

• Which patient group benefits, and what outcome is the service meant to improve?
• Which devices are validated for that group and measurement?
• Who reviews data, during which hours, and with what response times?
• What thresholds trigger action, and how were they chosen?
• How are patients without smartphones or connectivity included?
• How is the data stored, and does it meet health data protection rules?
• How will the service be evaluated, and when would it be stopped?

WHO's digital health strategy frames these as questions of governance, interoperability and equity, not just technology choice.

Outlook

Wearables will keep getting more capable, with more cleared medical features and more integration into care pathways such as virtual wards. The value will come not from more data, but from validated measurements, sensible thresholds, clear responsibility and fair access. Clinicians who can turn a patient's smartwatch screenshot into an appropriate, evidence-based next step will be well placed for this future.