Balance

Thirty minutes, twice a week, offline. The phone times a real task: catch against a wall, step-ups, carrying a full cup across the room, a ball on a string. Before each one, the goal-plan-do-check self-talk.

Parameters

30 to 60 minutes, one to three times a week, over four weeks to five months in the trials; six weeks is the Time Gates default. The task set changes weekly, because compliance drops by more than 30% after week four in every programme of this kind.

Scored

You self-rate; the phone in a pocket measures sway before and after.

Why this shape

The 2025 randomised-trial meta-analysis gives overall motor skills g = 0.57, balance g = 0.76 and activity performance g = 0.71, against g = 0.15 for process-oriented approaches. The trials ran in ages 4–13 with no adolescent subgroup, so the transfer to teenagers is an assumption we state rather than a finding we own. Goal-plan-do-check is a verbatim self-talk loop a coach can run, which is why it survives translation into an app.

Phone propped up, you in frame. Stand on one leg, then heel to toe, eyes open then closed.

Parameters

Three holds of thirty seconds per leg. Surface, shoes and time of day are standardised by a checklist, because they move the score more than practice does.

Scored

Sway from pose keypoints, and hold time.

Why this shape

Browser pose estimation correlates 0.80–0.91 with marker-based motion capture for limb movements, and phone-accelerometer sway correlates 0.75–1.00 with research accelerometers with retest between 0.41 and 0.83. That is good enough to track your own trend at home and not good enough to replace a supervised assessment - which is exactly how it is labelled.

Movement games that demand shifting weight and reacting, rather than repeating a fixed pattern.

Parameters

10 to 30 minutes, one to three times a week, four to six weeks.

Scored

In-game performance, and the same sway measure before and after the block.

Why this shape

A six-trial meta-analysis (n = 162) gives balance SMD 0.69, motor skills 0.54 and agility 0.65, comparable to neuromotor task training in one head-to-head comparison. Retention beyond a few weeks is unknown, which is why this supplements the coach blocks rather than replacing them.

Touch on-screen targets with your hand, then hold a called pose.

Parameters

Five minutes, speed increasing.

Scored

Pose match against the called position.

Why this shape

It is a plausible bridge between screen and body, with no trial behind it. Graded accordingly and never required.

Walk a line with the phone in your pocket while answering easy spoken questions.

Parameters

Two-minute blocks.

Scored

Accelerometer sway together with answer accuracy.

Why this shape

A scoping review of dual-task training in healthy children and adolescents found some balance gains, inconsistent designs and no best-practice dose. Included because the combination is realistic; graded because nobody has shown it does anything.

Once a week, thirty seconds: bumps, drops, trips, spills, as counts. A trend line and nothing else.

Parameters

Weekly, self-logged.

Scored

A count, shown as a trend, never compared to anyone.

Why this shape

Every other measure in this pillar is a proxy. This one is the thing a family actually cares about, and no validated instrument exists for it, so it is honestly labelled as an unvalidated trend line and treated as the real outcome anyway. Attaching any consequence to it would guarantee under-reporting, so nothing depends on it.