← Cloud analytics

Gross motor growth charts

Motor development plotted on age-normative reference curves, read like a physical growth chart

220 motor metrics modelled against age
1227 h of free playtime from 92 infants in the normative cohort
91% of the metrics generalised to an external clinical cohort

The same recording that produces a motor score can be placed on age-normative reference curves for 220 motor metrics — how much time in each posture, how often the infant changes posture, how much of the movement is locomotion — so that an individual child's development is read the way a paediatrician reads length and weight against a growth chart.

The charts were built once, from typically developing infants recorded repeatedly at home, and every new recording is measured the same way and positioned on them as a centile and an age-adjusted z-score (Figure 1A). Because each metric has a chart of its own, the result shows not only whether an infant is ahead or behind, but in what (Figure 1B).

AHow the charts are built and read

BOne infant on four of the charts

Figure 1. A For each of the 220 metrics, the normative recordings were grouped into one-month age bins from 5 to 19 months and a beta distribution was fitted in each bin; its parameters were smoothed across age, and the centile curves were read from the fitted distributions. B Four of the 220 charts — the share of free playtime spent lying, in crawl posture, sitting and standing — with the same infant as in the BIMS figure. At 10.4 months the infant was still lying for 88% of playtime, where the median infant of 10 months lies for 7% and the 97.5th centile is 47%, and sat for 10%, at the bottom of the sitting chart; a clinical AIMS assessment placed them below its 5th centile at the time (filled points). By 12.2 months lying, crawl posture and sitting were in the typical range. Standing followed later: near the bottom of its chart at 12.2 months, above the median by 15.5. Data: Airaksinen et al., Science Translational Medicine 2026, data files S2 (centiles) and S4 (participant id_09).
Show the data for panel B
Age (months)LyingCrawl postureSittingStandingAIMS at the time
7.096.4%0.3%2.9%0.4%below 5th centile
8.489.2%1.0%9.7%0.1%below 5th centile
10.487.9%0.7%10.2%1.2%below 5th centile
12.20.3%31.9%61.2%6.6%at or above 5th centile
13.97.5%14.7%47.2%30.7%at or above 5th centile
15.51.9%8.0%27.8%62.4%at or above 5th centile
18.30.7%4.5%48.2%46.6%at or above 5th centile

What you get

Per recording
A centile and an age-adjusted z-score for each of the 220 motor metrics, including the BIMS.
Per child
The individual trajectory across repeated recordings, against the normative curves.
Coverage
Postures, movements, transition dynamics, locomotion types and activity counts: time in each category, bout rates and lengths, left–right balance, and signal magnitude.

What you need

Garment
MAIJU suit with a fitting size.
Sensors
Four Movesense Flash or Movesense MD sensors.
Successful recording
From 30 minutes of detected free playtime with at least three sensors working; ages 5 to 19 months, the range the charts cover.

How it is derived

The normative charts were built from 580 successful at-home recordings — 1227 hours of free playtime — in 92 typically developing infants aged 4 to 19 months, each recorded repeatedly, on average 1.9 months apart. Every recording went through the same automated analysis: free playtime detected, posture and movement classified, and 220 metrics computed from the result (Science Translational Medicine 2026). The metrics are drawn from the posture and movement profile and from the raw signal, and they describe time in each category, how often bouts occur and how long they last, left–right balance, and movement magnitude.

Each metric was modelled with a beta distribution per one-month age bin, a two-parameter family that fits skewed values on a fixed scale; the parameters were smoothed across age and the centiles read from the fitted distributions. Comparison with a Gaussian model gave qualitatively the same results. The paper reports, for every metric, how strongly it changes with age and how steadily an individual infant keeps to their centile over time — which is what decides which metrics are most useful for following one child.

In a clinical cohort

The charts were tested on an external cohort of 39 infants with 201 measurements, recruited from a clinical follow-up programme that took in any infant with an observed developmental delay or abnormal neurological status. Eleven of them, with 60 measurements, had severely delayed gross motor development by the AIMS.

Of the 220 metrics, 201 (91%) did not differ between the normative cohort and the external cohort's infants with typical or mildly delayed development, so the charts carried over to a new, clinically relevant population. In the infants with severe delay, 120 metrics (55%) differed significantly from the normative cohort (Science Translational Medicine 2026).

Against physical growth charts

An earlier study compared this approach directly with the charts paediatricians already use. The motor measure fitted its developmental model as tightly as length, weight and head circumference fit theirs (R² = 0.99), with a single-measurement spread of 1.4 months against 1.5 for length and 1.9 for weight and head circumference (eBioMedicine 2023).

What typical development looks like

The charts are also a description of typical gross motor development at home. Lying falls from 92% of playtime at 6 months to 3% at 11 months. Prone rises from 29% at 5 months to 44% at 7 before the infant moves up to crawl posture, sitting and standing. Sitting rises to about 40% of playtime near the first birthday and then levels off, while standing keeps increasing, to 56% at 19 months. A new skill does not simply replace the old one: infants still spend about 7% of playtime in crawl postures at 16 months, long after they have learned to walk (Science Translational Medicine 2026).

Available on

Publications

All research →

Request a quote