From Smartwatches to Smart Socks: How the New Generation of Health Wearables Is Changing Care

Health wearables are moving beyond step counts and workout tracking. Smartwatches, rings, patches, connected clothing and sensor-equipped socks can now collect information about movement, heart rate, sleep, activity patterns and other physiological signals over long periods. The important shift is not simply that devices can measure more. It is that continuous data can reveal changes in a person’s usual pattern that may be difficult to see during an occasional clinic visit. Recent reviews of wearable health research show growing use in remote monitoring, chronic disease management and clinical research, although evidence and regulatory status vary considerably by device and application.

For dementia care, that distinction is particularly important. A smartwatch may record activity and sleep; a smart home may detect unusual movement or a door opening at night; and a smart sock can potentially capture physiological signals associated with distress or agitation without requiring the wearer to remember to operate a device. These technologies are promising because they can make changes in everyday behaviour more visible to caregivers and clinicians. But they remain monitoring tools, not automatic diagnoses: the quality of the sensor, the algorithm, the person’s baseline and the clinical context all determine what a change actually means.

The wearable is becoming a continuous health sensor

The first generation of consumer wearables largely asked a simple question: how active are you?

Today’s devices can ask much broader questions. Depending on the hardware, they may collect heart rate, movement, sleep-related measurements, skin temperature, oxygen saturation, electrical cardiac signals or other physiological data. Some systems combine those measurements with smartphone applications and cloud-based analysis.

The result is a shift from isolated measurements toward longitudinal patterns.

That matters because a single reading can be difficult to interpret. A person’s heart rate after climbing stairs tells a different story from their resting heart rate over several weeks. A poor night’s sleep may mean little by itself, while a sustained change in sleep and daytime activity could provide useful information for a clinician or caregiver.

A 2025 scoping review of wearables used for clinically relevant monitoring outside hospitals identified 80 eligible studies, illustrating both the breadth of research and the fact that this remains an evolving field rather than a fully established replacement for conventional care.

Why the body location matters

A smartwatch is convenient because people already understand how to wear one. But the wrist is not necessarily the best location for every measurement.

Different body positions provide different signals and different practical advantages:

The key lesson is that a more sophisticated wearable is not automatically a more useful one. The best placement depends on what researchers or clinicians are trying to measure and whether the person will tolerate the device for long enough to generate meaningful data.

Smart socks illustrate the next stage

Smart socks are particularly interesting because they turn an ordinary item of clothing into a sensing platform.

Research into SmartSocks developed for dementia care has explored sensors capable of measuring signals including heart rate, sweat-related changes and movement. The underlying idea is to identify physiological patterns that may accompany distress or agitation, potentially giving caregivers another source of information when a person cannot easily communicate how they feel.

The attraction is partly practical. A wrist device can be removed, forgotten or rejected. A sock may be less conspicuous and can operate as part of an established dressing routine.

Early research should nevertheless be interpreted cautiously. A 2024/2025 report on SmartSocks included very small feasibility samples while examining physiological-data quality and AI-based detection of stress. The researchers reported promising early findings, including pulse-rate measurements with a mean absolute error below 5 beats per minute in one testing condition, but those results do not establish that the technology can reliably diagnose pain, agitation or disease in the wider dementia population.

A separate feasibility study involving people with dementia was designed around 30 care-home residents, with the purpose of assessing acceptability, feasibility and the relationship between sensor data and established measures of distress and pain. That distinction—testing whether a technology works reliably in real care environments rather than simply demonstrating that its sensors function—is crucial.

Dementia may be one of the clearest use cases

Dementia presents a particular challenge for conventional monitoring because important changes can occur between appointments.

A person may sleep differently, become less active, wander at unusual times, spend longer in bed, move differently or show behavioural changes without being able to explain what has changed.

Wearables cannot determine the cause automatically. But they can potentially create a record of what happened before and after an event.

That makes them potentially useful alongside smart-home technology. The broader ecosystem can include bed sensors, motion detectors, door sensors and other non-camera monitoring systems. The Alzheimer’s Society describes technologies such as bed- or chair-absence sensors, door sensors and movement monitoring as forms of assistive or telecare technology that can support independence and alert trusted people to unexpected events.

For readers interested specifically in this development, explores how these technologies fit together inside the home.

The important concept is baseline detection. Instead of asking an algorithm to decide whether an individual is “normal,” a useful system can learn what that particular person’s ordinary routine looks like and flag meaningful deviations.

That approach may prove more valuable than trying to produce a universal definition of normal behaviour.

AI turns measurements into patterns—but interpretation remains the hard part

Sensors generate numbers. Care requires meaning.

Artificial intelligence and machine-learning systems can process large streams of sensor data and look for relationships that would be difficult to identify manually. In principle, an algorithm could detect that a person’s movement, sleep and physiological signals have changed together.

That is where the technology becomes more than a digital diary.

But AI does not eliminate uncertainty. A change in heart rate might reflect anxiety, medication, exercise, pain, illness or a faulty measurement. Reduced movement might indicate fatigue, worsening mobility, a change in routine or simply a day spent indoors.

For dementia care, false alarms can be especially problematic. Too many alerts may overwhelm caregivers and eventually cause them to ignore the system. Too few alerts can create false reassurance.

This is why the clinically meaningful question is not simply whether an algorithm can identify a pattern in a dataset. It is whether that pattern remains reliable in real homes, among different people, over long periods and under changing conditions.

Recent systematic reviews continue to identify small study populations, heterogeneous measurements and early-stage research as limitations in the evidence base for wearable technologies in Alzheimer’s care.

Research into AI-enabled dementia monitoring is also expanding beyond wearables. examines the wider possibility of combining sensor data with AI to identify changes earlier.

The biggest advantage may be passive monitoring

One of the strongest arguments for modern wearables is not that they measure more—it is that they can require less effort from the user.

Traditional health tracking often depends on remembering to enter information, answer questions or perform measurements at specific times. Passive sensors can collect data while a person goes about their normal life.

That can be valuable for older adults and people living with cognitive impairment.

The trade-off is that passive monitoring can also become passive surveillance.

The more continuously a system observes someone, the more information it potentially collects about their movements, routines, physiological state and behaviour. In a dementia-care setting, questions about consent, capacity, family access and data retention therefore become central rather than peripheral.

The Alzheimer’s Society recommends that decisions about technology supporting people with dementia should involve the person themselves where possible and should reflect individual needs and routines.

That principle matters because the technically most sophisticated system may be inappropriate if the person finds it uncomfortable, intrusive or confusing.

A health wearable is not automatically a medical device

Consumers should also pay attention to the distinction between wellness technology and regulated medical technology.

The U.S. Food and Drug Administration’s current guidance distinguishes low-risk products intended for general wellness from products making medical claims. A wearable marketed for general fitness or wellness may fall outside certain medical-device requirements, while a product intended to diagnose, treat, prevent or monitor a specific disease can face a different regulatory pathway.

That distinction matters when interpreting marketing claims.

A device saying that it tracks sleep is making a different claim from one saying that it diagnoses a sleep disorder. Likewise, an algorithm that identifies a change in activity is not automatically capable of determining why the change occurred.

The FDA now maintains a list of authorized sensor-based digital health technologies, including wearable devices such as watches, rings, patches and bands designed for health monitoring outside traditional clinical settings.

Regulation also varies between countries, so authorization in one market should not automatically be assumed to apply elsewhere.

There are some claims consumers should treat with particular caution. For example, the FDA has warned that it has not authorized, cleared or approved any smartwatch or smart ring that independently measures or estimates blood glucose without piercing the skin.

The general rule is simple: a sensor’s ability to measure something is not the same as its ability to diagnose a medical condition.

What readers should check before trusting a wearable

The most useful health wearable is not necessarily the one with the longest specification sheet.

Look instead at the evidence behind the specific measurement you care about. A device may perform well for heart-rate tracking but have weaker evidence for another metric. Accuracy can also change with movement, skin contact, body position and other real-world conditions.

For a health-related purchase, it is worth checking:

• What exactly is being measured?
• Is the feature intended for wellness or medical use?
• Has the relevant function been independently validated?
• What happens when the sensor loses contact or produces uncertain data?
• Does the device require a subscription?
• Where is health data stored and who can access it?
• Can data be exported or shared with a clinician?
• How long does the device operate between charges?
• What happens if the wearer stops using it?
• Is the product available and supported in your country?
• What regulatory authorization applies to the specific health function?

Privacy deserves particular attention because a health wearable can reveal routines even when it is not recording audio or video. Movement data can show when someone sleeps, wakes, leaves home or becomes unusually inactive.

Security therefore needs to be considered alongside accuracy.

For people using technology specifically to support dementia care, the question should also be whether the information improves care enough to justify the monitoring involved.

The future is likely to be an ecosystem, not a single device

The evolution from smartwatch to smart sock does not necessarily mean that one wearable will eventually monitor everything.

The more plausible direction is a network of complementary sensors.

A watch can provide wrist-based physiological and activity data. A sock can capture information closer to the feet and gait. A bed sensor can observe nighttime movement. Door and room sensors can provide environmental context. A phone can supply connectivity and location-related information. Software can combine those streams into a picture of changing behaviour.

This is particularly relevant to dementia care, where context can be as important as an individual physiological measurement.

A change in nighttime movement means something different if a door sensor shows repeated trips outside the bedroom. A reduction in daytime activity means something different if a person has also stopped leaving the house.

The technology therefore becomes more useful as the individual measurements are interpreted together—but also more sensitive from a privacy perspective.

Work on AI-enabled home monitoring is already exploring this direction. looks at how these systems can extend monitoring beyond the body itself.

What matters now—and what still needs to be proved

Wearable health technology has moved well beyond the simple fitness tracker. The underlying sensors are becoming smaller, software can analyse much larger streams of data, and researchers are increasingly testing these systems outside controlled laboratories.

The strongest near-term value is likely to come from continuous observation that complements human judgment rather than from devices attempting to replace it.

For dementia care, that distinction is especially important. A smart sock that notices a physiological change may give a caregiver a reason to investigate. A smartwatch that detects an unusual pattern may provide useful context for a clinical conversation. A network of home sensors may reveal that a person’s routine has changed.

None of those observations, by themselves, explains why the change happened.

That is the unresolved issue that will determine how far the new generation of wearables can go: not whether technology can collect more health data, but whether that data can be turned into reliable, clinically meaningful information without creating unacceptable burdens around false alarms, privacy, cost and constant surveillance. The future of health wearables will therefore depend as much on validation, human-centred design and responsible data use as it does on adding another sensor to the device.