Wearable Technology

FRCS Level

The Principle

Every other method in this chapter measures a few repetitions precisely. Wearables measure millions imprecisely.

Which is better depends entirely on the question. For a mechanism, precision wins. For an exposure, volume wins.

Cumulative load is an exposure question, and it is the one clinical variable this chapter keeps naming and never measures. 3.14.4 covers it.

At the Knee

Inertial measurement units combine accelerometers, gyroscopes and magnetometers to estimate segment orientation.

Instrumented insoles measure plantar pressure and give a usable surrogate for ground reaction force. 3.4.2 covers the forces.

Knee angle from inertial sensors agrees with motion capture to within a few degrees in the sagittal plane and poorly in the transverse. 3.16.7 covers the comparison.

Step counts, jump counts and activity classification are reliable and are what most clinical use rests on. 3.15.5 covers jump counting.

Estimating a joint moment from wearables requires a model, so the output inherits that chain. 3.16.5 covers it.

What Changes It

Sensor placement and whether it is reproducible by a patient at home.

Calibration and drift, since orientation estimates degrade over time without correction.

The algorithm converting signal to a clinical quantity, which is often proprietary.

Battery and compliance, which decide whether the data exists at all.

The quantity sought, since counts are reliable and kinetics are not.

Clinical Relevance

Use wearables for exposure and count data, where they are reliable.

Treat a wearable derived joint moment as a model output rather than a measurement.

Expect transverse plane data from inertial sensors to be weak.

Ask whether the algorithm is published, since a proprietary conversion cannot be appraised.

Fellowship Level

At the Knee

The clinically transformative possibility here is not better measurement but continuous measurement, which is a different category.

A single laboratory visit samples one day's walking under observation. A wearable samples the behaviour that produced the problem.

That matters most for the conditions this site describes as loading problems: patellofemoral pain, tendinopathy and arthritic progression. 3.12.9 covers the first.

Validation is uneven. Step counting is well validated, joint angles moderately, and estimated loads barely.

Machine learning is the route by which sparse sensor data becomes a joint level estimate, which makes this page dependent on that one. 3.16.6 covers it.

What Changes It

The quantity being estimated, since validation quality differs by an order of magnitude across them.

Whether the validation was done in a laboratory or in the field, which is a large difference.

Patient compliance, which determines whether the volume advantage is realised.

Sensor count, since more sensors improve accuracy and reduce the chance they are worn.

Data handling arrangements, which are a governance question rather than a technical one.

How It Is Measured

Concurrent validity against laboratory motion capture, which is the standard and tests the device in the condition it is meant to escape.

Field validation against a reference wearable, which is weaker and more representative.

Instrumented implant comparison for load estimates, which is the only direct check and covers replaced knees. 3.14.12 covers them.

Test retest reliability over days, which matters for a monitoring device and is often unreported.

Validating in a laboratory a device whose purpose is to leave the laboratory is the field's standing methodological awkwardness.

Clinical Relevance

Match the claim to the validation, since step counts and load estimates are not equally supported.

Expect laboratory validation to overstate field performance.

Plan for compliance explicitly, since the method's advantage is entirely in volume.

Note the awkwardness of validating a field device in a laboratory when appraising a study.

Consultant Perspective

At the Knee

This is the page where the chapter's recurring complaint could stop being true, so the decision is when to start using these data rather than whether.

On one side, exposure is the variable that most load related conditions actually turn on, and a device a patient already owns measures it. 3.14.4 covers the argument.

On the other, the quantities with clinical thresholds are the ones validated worst, and acting on an unvalidated threshold is worse than acting on none.

What moves the weight is which quantity. Counselling a runner on weekly volume change rests on step counts, which are solid. Setting a joint load limit rests on estimates that are not.

There is also a question nobody in this chapter has to answer and this one does: who holds the data, for how long, and what else it gets used for.

What Changes It

The quantity used, which separates well validated counts from poorly validated loads.

Whether a threshold is being set or a trend is being watched, since trends tolerate inaccuracy.

Patient compliance and the practicalities of a device being worn for months.

Data governance arrangements, which are a clinical responsibility and not an IT one.

Whether the device is consumer or medical grade, which changes both validation and regulation.

How It Is Measured

Validation studies by quantity, which must be read individually rather than taken as a device level claim.

Field reliability over weeks, which is what a monitoring use needs and is rarely reported.

Clinical trials using wearable endpoints, which are increasing and are the evidence this decision needs.

Compliance rates in real cohorts, which are the practical limiting factor.

No knee specific wearable threshold has been validated against an outcome, so any limit given to a patient today is an informed guess.

Clinical Relevance

Use counts and trends now, and wait for validation before using absolute load thresholds.

Read validation by quantity rather than by device.

State that any wearable derived limit given to a patient is currently an informed guess.

Take responsibility for the data governance question rather than treating it as somebody else's.

Weigh the patient's age and physiology, comorbidities, knee demand, psychological state and their own ideas, concerns and expectations before asking someone to wear a monitor for months, because continuous measurement of your own body changes how some people feel about it.