When a watch or patch reports a number about your body, what happened between the body and the screen? A sensor first detects a physical signal; software processes it; then a person or care team interprets the result. Each link can introduce uncertainty. A wearable may be useful for tracking or self-management without being able to diagnose disease, and a promising measurement is not automatically evidence that acting on it improves health.
A sensor measures a signal, not a diagnosis
Digital health includes wearable devices with sensors, apps, and communication systems that move health information among users, caregivers, and professionals [2]. A sensor detects a physical quantity—such as motion, light, pressure, or a chemical signal—and converts it into an electrical or digital reading. That reading is a proxy for something about the body, not the thing itself. Software may filter noise or infer a pattern, but the inference depends on how well the signal represents the condition of interest.
For example, a wearable patch paired with a learning model has been developed to detect wheezing through tiny vibrations. The technology may help classify respiratory sounds, but potential to speed diagnosis is not the same as showing that every user’s device can diagnose asthma or chronic obstructive pulmonary disease [2]. Keep the distinction clear: measurement, classification, diagnosis, and improved health are separate steps.
Ask what was compared
Evidence for a wearable claim should identify the intended users and the comparison used. A small clinical study of a smartphone app that measures heart rate from facial color changes found measurements highly accurate compared with current monitoring methods [2]. That finding concerns heart-rate measurement in the studied context. It does not by itself establish detection of every irregular rhythm, reliable performance across populations, or better health outcomes.
Clinical trials are research studies that test how well medical approaches work in people. Their purpose can differ: a study may test treatment, screening, prevention, or supportive care [1]. For a wearable, ask whether investigators tested the sensor against a reference measure, whether the algorithm classified the signal correctly, and whether care decisions based on the output helped. A result at one stage cannot substitute for the next.
Look for the full data pathway
A wearable’s information may be sent to an app, a health record, or a clinician. That pathway can make repeated monitoring possible, and connected glucose monitors and insulin pumps illustrate how a sensor can be linked to a treatment device [2]. The sensor detects glucose levels; a signal is transmitted; the pump adjusts insulin delivery. This is a particular system for a particular use, not evidence that all wearables make treatment decisions safely or independently.
Ask how often the device samples, what happens when the signal is missing or noisy, and who reviews an alert. A continuous stream can capture changes missed by a single appointment, but more data do not guarantee better interpretation. The value depends on whether the information is accurate enough, reaches someone able to act, and changes care in a beneficial way.
Match the evidence to the promise
A marketing claim may emphasize convenience, earlier detection, or prevention. Each is a distinct outcome. If a device claims to monitor, the evidence should address measurement performance. If it claims to detect a condition, the evidence should address classification in the intended population. If it claims to prevent complications, studies need to examine what happens after the alert and whether outcomes improve. Trials are designed to answer different research questions, so their design must fit the claim [1].
Wearables can support monitoring and self-management, and some may help clinicians follow chronic conditions over time [2]. But a device’s usefulness is specific to its sensor, software, intended users, and workflow. Read the claim as a chain: bodily signal, recorded data, interpretation, decision, and outcome. Evidence that supports one link should not be stretched to cover the whole chain.
Explore the physics: an explanatory diagram
Sources & further reading
- Clinical Trials: What Are the Phases? National Cancer Institute
- Digital Health NIBIB
Prepared for the independent PSI Center reading library. About edition dates and source checking.
This explainer describes scientific principles. It does not assess individual treatment choices.
