A Fall-Risk Score Is Not Enough

A score is a snapshot. It compresses a complex, time-varying pattern into a single value at a single moment — and in doing so it discards the dimension that often carries the most information.

On any one day, these two people can produce the same score. Read as single numbers they look identical. Read as sequences they are telling completely different stories — and only one of them warrants attention.

Population norms are useful. They give a reference for what typical movement looks like across a group, and they are often the only reference available at a first encounter.

But people differ enormously in their normal movement, for reasons that have nothing to do with risk: height, footwear, habitual pace, prior injury, occupation, terrain. Someone can sit outside a population norm permanently and be entirely stable, while someone comfortably inside it can be moving away from their own established pattern.

Longitudinal analysis is only as good as the stability of what it tracks. The gait descriptors that repay repeated measurement include:

• Walking speed
• Cadence
• Stride variability
• Left/right asymmetry
• Posture through the gait cycle
• Balance-related indicators
• Change in the temporal structure of the gait pattern

How these are derived from a walking observation is described in From Walking Video to Movement Intelligence.

A worsening feature should not automatically trigger a medical conclusion. A trend is evidence that something changed; it is not evidence of why.

Any of the following can move a movement feature without indicating clinical deterioration:

• Temporary fatigue — time of day, sleep, recent exertion
• Injury — including minor, unrelated and self-limiting ones
• Rehabilitation — where change is the intended outcome
• Environment — surface, footwear, available walking distance
• Measurement quality — a degraded capture producing a degraded feature
• Device or camera differences — a change of instrument, not of person

This is why capture conditions belong in the record alongside the measurement. A trend computed across inconsistent instruments and settings is not a trend in the person — it may simply be a trend in the measurement setup.

Framed that way the output has a clear and modest job: direct limited professional attention toward the people whose movement has changed. It orders a queue for review. It does not make a determination, and it does not replace the judgment of the clinician or care team who does.

This is the shape FallRisk is built around, and the reason repeated, consistent capture matters more than any single high-effort assessment. The wider MobilityCare family applies the same longitudinal logic across rehabilitation, post-operative movement and elderly care.