Why two people with the same MRI can move completely differently

Your scan comes back showing moderate osteoarthritis. The person in the next consulting room has an almost identical report — same grade, same joint, similar age. Yet they are managing a gentle daily walk, and you can barely get down the stairs by afternoon. The imaging tells both of you what is structurally wrong. It cannot tell either of you why your experience is so different.

The answer lies not in the picture of the joint but in the way the whole body has learned to move around it. Over months or years, without any conscious decision, every person in pain develops a private set of adjustments — a shift in how weight is distributed, a subtle change in stride, a compensatory lean that takes load off the sore side. Research published in Gait & Posture by Van Rossom and colleagues (2023) confirmed that knee and hip osteoarthritis patients each develop a distinct biomechanical fingerprint: a personalised, often unconscious movement strategy that shows up clearly in the way they walk and transfer weight, but rarely shows up on a scan.

That fingerprint is the missing piece. Standard imaging shows what a joint looks like at rest. It cannot show how your whole body is loading, compensating, and adapting — moment to moment, under the actual demands of daily life.

How MAI Motion captures 5,000 data points every second

The process takes no more than a few minutes and requires nothing from the patient except ordinary movement. No reflective markers are taped to the skin, no wearable sensors are clipped to the body, and no laboratory suit is needed. The patient simply walks, squats, or rises from a chair while a camera records.

That ordinary footage is what MAI Motion® — a UKCA and MHRA-registered AI platform — works from. Rather than the language-based AI familiar from consumer tools, it uses computer vision: algorithms that read the geometry of movement from standard 2D video much as a structural engineer reads stress in a bridge from its flex. Each second of recording is converted into a three-dimensional body mesh, rebuilding joint position and alignment across every plane of movement. The result is approximately 5,000 data points per second — enough information from a single squat lasting under ten seconds to fill a standard spreadsheet many times over.

What makes this clinically useful is not the volume alone but the consistency. Fifteen skeletal keypoints are tracked frame by frame, generating the same objective measurements regardless of who performs the assessment or when it is repeated. A published study by Armstrong, Wen, and colleagues in the Journal of Arthritis (2022) documented this approach as a low-cost method for detecting and tracking knee osteoarthritis — grounding the technology in peer-reviewed clinical application rather than proof-of-concept alone. The shift from a clinician's trained eye — valuable but inherently variable — to reproducible, timestamped numbers means that change over time can be measured rather than estimated.

The compensation patterns hidden inside knee and hip pain

The body is a skilled improviser. Long before pain becomes severe enough to prompt a clinic visit, the musculoskeletal system has already begun redistributing load — quietly, automatically, and in ways that follow a predictable anatomical logic.

In knee osteoarthritis, the primary problem is excessive pressure across the inner (medial) compartment of the joint. The body's instinctive answer is to shift that pressure elsewhere. Patients often lean slightly towards the sore knee as they walk — a lateral trunk lean that reduces the adduction moment across the medial compartment. Stride direction shifts too: turning the foot outward changes the angle at which ground forces travel through the knee. The knee also tends to stay straighter through the gait cycle, sacrificing normal shock absorption to avoid the loading arc where pain is greatest. Most people are entirely unaware that any of this is happening.

Hip osteoarthritis produces a different signature. The gluteus medius — the muscle that holds the pelvis level during single-leg stance — weakens, and the body compensates by leaning the trunk towards the affected side as each foot strikes the ground. This is the Duchenne limp: a pelvic drop on the unaffected side, offset by an ipsilateral trunk lean on the painful one. Hip extension shortens, and the lower back compensates with increased anterior pelvic tilt and extra lumbar movement.

Both patterns are effective in the short term. The difficulty, as Zeni (2014) established — and as Van Rossom and colleagues (2023) confirmed through measurable gait fingerprints across both conditions — is that load does not disappear; it relocates. The lean that protects a painful medial compartment quietly overloads the outer compartment, the contralateral knee, or the lower back. Secondary pain sites are not random: they are biomechanically predictable from the compensation pattern itself.

What the data finds that MRI and clinical exam miss

Even the most attentive clinician watching a patient walk a hospital corridor is working from a compressed signal — stride pattern, obvious asymmetry, general posture. That trained perception carries real diagnostic value. What frame-by-frame computational analysis adds is a different layer: the micro-variations that accumulate across dozens of repetitions and are too granular for any observer to track simultaneously.

The most specific biomarker evidence published so far comes from within the MSK Doctors research group, with independent replication across other centres an area of active work. With that context in place, the findings are clinically meaningful. A clinical trial documented in Musculoskeletal Regeneration Medicine identified two statistically significant markers (p<0.05) for knee pain status, both measurable from standard markerless video: the smoothness of the knee flexion curve during a squat, and the cumulative acceleration of elbow flexion during a sit-to-stand. Both capture the quality, not merely the range, of a movement — something a standard consultation almost never records. Armstrong, Wen, Lee and colleagues' 2022 paper in the Journal of Arthritis grounded this in a peer-reviewed clinical context, documenting markerless capture as a low-cost method for detecting and monitoring knee osteoarthritis.

Reproducible numerical biomarkers address something that matters to anyone reviewing the same patient over months: subjective clinical observation varies between appointments and between clinicians. Stance time symmetry, flexion curve shape, and rotation timing — measured consistently by the same system — provide a common reference point, making treatment response measurable rather than impressionistic.

Where the picture becomes fuller still is in integration with 3D volumetric MRI. MAI Motion® correlates functional motion data directly with structural imaging, so a consultant sees not only what a joint looks like on a scan but how the patient actually loads it during movement. A moderate cartilage lesion reads differently against a background of well-distributed loading than against a persistent antalgic lean — and that composite view is what neither assessment alone can provide.

Motion Age: what a single number tells you about how you move

All that accumulated kinematic data — built from hundreds of keypoints tracked across dozens of repetitions — resolves, in MAI Motion's output, into a single figure: a Motion Age.

The concept is deliberately analogous to a fitness age score, but grounded in movement science rather than wellness marketing. MAI Motion benchmarks each individual's movement signature against a population of people at the same chronological age, then expresses the result as a biological movement age. A Motion Age higher than chronological age indicates that movement quality has declined relative to peers — characterised by loading asymmetries, reduced range, or less fluid transitions between phases. A younger Motion Age reflects movement that is well-distributed and mechanically efficient.

What makes this practically useful is not the single number but the trend it generates over time. Every re-scan is compared against the individual's personal baseline and all previous results. For patients managing knee or hip pain at a distance from the clinic, the home app makes this longitudinal record genuinely practical: a follow-up scan does not require a journey to Sleaford or Grantham. For those in rehabilitation after injection therapy or a procedure, a falling Motion Age provides an objective correlate of recovery that sits alongside self-reported pain scores rather than replacing them.

Within the MSK Doctors Regen PhD programme, Motion Age is paired with a 32-biomarker blood panel as part of a broader functional baseline, though the longitudinal tracking output is available independently of that programme.

What a movement assessment involves at MSK Doctors

Knowing that a joint's movement fingerprint is measurable is one thing; understanding what an assessment involves in practice is another.

At MSK Doctors' clinics in Sleaford (NG34) and Grantham (NG31), no GP referral is required. The assessment itself is brief and non-invasive: the patient performs a small number of ordinary movements — walking, squatting, sitting and standing — in front of a standard camera. Nothing is attached to the body. The process takes minutes.

What follows is not an automated printout. The movement data is interpreted by a consultant alongside clinical history, a physical examination, and, where relevant, MRI findings. Motion capture informs the consultation rather than supplanting it — shifting part of the conversation from impression to measurement while keeping clinical judgement at its centre.

The output then has a longitudinal life. Between visits, follow-up re-scanning through the home app means that recovery or progression can be tracked without a return journey. For patients travelling from outside Lincolnshire, assessment and consultation are typically combinable in a single visit. For those based in London, equivalent assessment is available through the London Cartilage Clinic.

To book without a referral, visit mskdoctors.com.