Experimental design.

(A) Measures taken in the BRIGHT project used in this work. The measuring tape represents anthropometric measures, the brain represents fNIRS FC and the test represents the cognitive flexibility assessment. (B) Schematic representation of the spatial layout of the fNIRS array. Sources are marked with red stars, detectors are marked with blue circles, channels are marked with grey lines and numbered with black circles. The channels/optodes used as a reference for the tragus are highlighted in green. (C) Participants undergoing fNIRS testing at different ages. Image reproduced with permission from (95).

Linear mixed models results showing FC that displayed a statistically significant change with age.

(A) Significant results of the linear mixed model, blue indicates connections that decreased with age, red indicates connections that increase with age. (B) Mean and standard error of the mean (SE) of the functional connections that changed with age. Error bars are 1 SE. (C) Violin plot showing the mean ± SD (red circles and lines) and the individual variability (coloured dots) of the FC that showed a change with time.

FC that significantly changed with age

. Results are displayed in terms of estimated betas, standard errors and p values that survived Bonferroni correction.

Associations between FC, early growth and later cognitive flexibility.

(A) Significant positive associations are in green, significant negative associations are in orange, and non-significant associations are in blue. * indicates regressions that survived FDR correction for multiple comparisons. (B) Schematic representations of the early FC connections shown to predict cognitive flexibility in preschoolers. Significant positive associations are in green, significant negative associations are in orange.

Results of the regression analyses of the effect of ΔWLZ on FC at 24 months

. Significant positive associations are in green, significant negative associations are in orange, and non-significant (NS) associations are in blue; * indicates regressions that are still significant after correcting for HCAZ at 7/14 days and ** indicates regressions that are still significant after correcting for neonatal HCAZ and WLZ. Regressions that survived FDR correction for multiple comparisons are in bold.

Demographic information per each age.

Schematic representation of the fNIRS array and the functional connections tested.

(A) Each dot represents a channel, colours on the left plot corresponds to colours on the right on the baby’s head. 6 sections. (B) The 21 connections tested in the linear mixed models. Interhemispheric connections are in orange, intrahemispheric connections are in green, fronto- posterior are in blue, and crossing interhemispheric connections are in yellow.

Linear mixed models result showing FC that displayed a statistically significant change with age.

(A) Results of the linear mixed model, blue indicates connections that decreased with age, red indicates connections that increase with age. (B) Mean and SE of the functional connections that changed with age. Error bars are 1 SE.

An example scatterplot of the association between

ΔWLZ and FC at 24 months. Scatterplot of the association between ΔWLZ between birth and 5 months and frontal interhemispheric connectivity at 24 months. The black line represents the line of best fit.

fNIRS preprocessing steps.

The column “infants included in the analyses” in Table 3 refers to those participants whose data survived these preprocessing steps.

The effect of different thresholds for GVTD for motion detection and minimum valid data after pre-processing on data inclusion.

(A) Percentage of data included (left) and seconds of data included (right) per participant. Each line represents an infant, the black line represents the mean value. These graphs are reported from the 12 months sample as example. (B) Number of infants included in the LMM by varying STD threshold for GVTD and minimum length of valid data at different ages.

Strength of correlation between FC in the first and last portion of data.

This graph is reported from the age 12 month sample as example.

FC that significantly changed with age (fNIRS pre-processing without global signal regression). Results are displayed in terms of estimated betas, standard errors and p values that survived Bonferroni correction.

Characteristics of included and excluded participants and seconds of data included in the analyses at each age. WD=withdrawn, D=deceased, MV=missed visit, DD=developmental delay, NIRS not undertaken = the participant was assessed but did want or could not perform the NIRS assessments, FC not undertaken = the participant was assessed with other NIRS task, but not FC, Fussed out = the participant wore the headband and the FC acquisition had started but the participant showed signs of fussiness soon after the start of the acquisition, MP=missing pictures of the headband placement, EM=missing event markers, TI=technical issues during the NIRS testing session.