Why your knee can look fine on paper but feel wrong in motion

How do you know if your knee is genuinely deteriorating, or just having a bad week?

It is one of the most frustrating positions in musculoskeletal care: you have persistent knee pain, you have had a clinical examination, perhaps even an MRI — and the feedback is, essentially, 'nothing too alarming.' Yet the pain persists, the stairs feel wrong, and something is clearly not right.

The explanation, in many cases, lies not in the structure of the knee but in how it moves. A standard clinical exam measures range of motion and checks for tenderness when you are sitting still. An MRI produces a detailed image of bone, cartilage, and soft tissue — but a static one, taken while you are lying flat and not bearing any weight at all. Neither test can show what happens to your knee joint when you are actually walking across a car park, standing up from a chair, or descending a flight of stairs.

What those activities often reveal — invisibly, to the naked eye — are compensatory movement patterns. When a joint is painful or unstable, the body quietly redistributes load to protect it. The hip drops a fraction. The opposite leg bears more weight. The stride shortens on one side. These adaptations feel normal after a while, but over time they place abnormal force through cartilage surfaces that were never designed to take it.

This is the gap between 'your MRI looks reasonable' and 'but why does it still hurt?' — and it is a gap that conventional assessment, through no fault of its own, is not built to close.

How markerless motion capture works — and what '5,000 data points' actually means

Think of a video call running at 30 frames per second — each frame a snapshot of the room. MAI Motion®, the UKCA and MHRA-registered motion analysis platform used at MSK Doctors, captures movement at 120 frames per second, and within each frame it is not simply recording a picture: it is calculating the spatial position of anatomical keypoints simultaneously.

Keypoints are fixed anatomical reference positions — shoulder, hip, knee, ankle, and so on. MAI Motion® tracks 15 of these points at once. At 120 frames per second, that generates approximately 1,800 individual keypoint measurements every second, with no suit, no reflective markers, and no hardware attached to the body at all. The patient walks in and moves naturally; the AI extracts the rest.

Underlying this is a body model called SMPL (Skinned Multi-Person Linear model), which represents each frame not just as a skeleton of 23 joints but as a surface mesh of thousands of points. Across joints, mesh vertices, and the timing of movement between frames, the combined data stream is extremely dense — dense enough to detect compensations that are invisible to the human eye: a knee that loads fractionally later on one side, a hip that drops a few degrees during the weight-acceptance phase of walking. Those patterns carry diagnostic information that no static image and no manual assessment can reliably capture.

All of this resolves to a single output: the Motion Age score. By comparing a patient's movement signature against age-matched population norms, the system produces a reproducible number that reflects functional biological age rather than chronological age — one that can be tracked across appointments to show whether a knee is responding to treatment or continuing to change.

What the scan actually finds — the movement patterns that predict cartilage stress

The joint curves that MAI Motion® generates unfold over time — tracking load symmetry, knee flexion angle, and rotation timing across an entire walking stride or sit-to-stand task, not freezing a single angle at a single moment.

Compensation is a process, not a position. A patient may be entirely unaware they are favouring their left knee — but the movement curves will show it: the right leg bearing weight fractionally longer, the left knee flexing less deeply during loading, the pelvis tilting a few degrees to redistribute force. These are precisely the early signals that manual examination tends to miss.

Two metrics sit at the heart of the analysis: smoothness — how consistently a joint follows its expected arc — and impulse, the cumulative acceleration of the limb through the task. Smoothness tends to degrade before pain reaches a level the patient would report clinically; a knee beginning to guard itself produces a subtly irregular curve well before symptoms visibly worsen.

The choice of movement task turns out to matter. Early clinical evaluation found sit-to-stand considerably more diagnostically sensitive than squat: statistically significant changes in smoothness and impulse emerged across multiple joints during sit-to-stand (p<0.05) where the squat produced far fewer. Those findings used standard-grade video hardware, so they are best understood as indicative of the technique's sensitivity rather than large-scale clinical benchmarks — a distinction worth noting before framing the result.

What the consultant receives is not raw video but an organised set of curves — a movement signature that makes the invisible visible, without replacing the clinical judgement required to interpret it.

From the gait lab to your living room — why the phone matters

Not long ago, capturing a person's gait with clinical precision meant booking time in a specialist laboratory — motion-sensor suits, calibration rigs, rooms full of hardware, and a bill that reflected all of it. Repeat assessments, essential for tracking a recovering knee, compounded the cost and the inconvenience.

The baseline session for MAI Motion® still begins in clinic — around 30 minutes, supervised, establishing the personalised reference data against which every future scan is compared. For patients outside London, that session takes place at MSK Doctors' sites in Sleaford or Grantham, Lincolnshire; London readers are seen through the London Cartilage Clinic arm of the group.

What changes after that first visit is significant. Subsequent re-scans are completed at home via the MAI Motion smartphone app — running the same capture pipeline the clinic uses, not a simplified consumer version of it. The AI watching the patient move in their living room is the same system that generated the baseline. A repeat assessment requires a phone, adequate floor space, and nothing else.

For patients across Lincolnshire and the wider non-London catchment, that removes three traditional barriers in one step: the long drive to a specialist centre for every follow-up, the cost of repeat laboratory appointments, and the enforced gap between assessments while a waiting list clears. Progress can be documented at six weeks or twelve — whenever the clinical question demands it — without either party needing to be in the same room.

How the data changes your treatment plan

MAI Motion® informs the consultant's clinical judgement — it does not replace it. That distinction shapes how movement data feeds into treatment decisions at each stage of a patient's pathway.

A documented case from the MSK Doctors clinical team illustrates the value of serial scanning. After an initial injection, MAI Motion® scans were repeated at six weeks and then twelve. The curves told a clear story: stance time on the right leg moved towards symmetry, the knee flexion arc regained shape through the loading phase — a sign the patient had stopped guarding — and rotation timing normalised. These are objective, reproducible changes rather than an interpretation of how the patient felt on a given morning.

That distinction matters in both directions. When movement curves show measurable change, the consultant has grounded evidence to support continuing the current management approach and to avoid premature escalation to a more invasive stage of care. When the curves remain flat after treatment, the data supports a different conclusion: the current approach is not producing a biomechanical response, and earlier escalation may be warranted. The clinical decision still belongs to the consultant; the data reduces the guesswork surrounding it.

Tracking the percentage change between pre- and post-treatment curves also transforms the language of clinical milestones. 'Are you ready to progress?' shifts from a subjective impression to a question with a measurable answer — one grounded in the patient's own baseline movement signature rather than a generalised expectation. The reference point is always the individual, not a population average.

What to expect from a motion capture assessment

The question of who benefits from a motion capture assessment is, in practice, broader than it first appears. The obvious candidates are patients already managing knee osteoarthritis, those recovering from surgery, and athletes working towards a return to training — all situations where a before-and-after movement signature has direct clinical value. An equally strong case exists for the patient who has symptoms but no clear diagnosis yet: an objective baseline, established before deterioration progresses, gives any future scan a meaningful reference point rather than a single moment in time.

At MSK Doctors, the initial assessment is held at the Sleaford (NG34) or Grantham (NG31) clinics and combines a consultant-led clinical review with the motion capture baseline in a single appointment. No GP referral is required to book, and results are available on the patient dashboard at the end of the session.

The article opened by asking how a patient could know whether their knee was genuinely deteriorating, or simply having a bad week. Motion capture does not replace the clinical judgement, imaging, and history that answer that question fully. What it adds is something that subjective assessment alone cannot provide: a number that changes when the knee changes, and stays flat when it does not.

You can book a consultation without a referral at mskdoctors.com.