Limits...
Neonatal Diagnostics: Toward Dynamic Growth Charts of Neuromotor Control

View Article: PubMed Central - PubMed

ABSTRACT

The current rise of neurodevelopmental disorders poses a critical need to detect risk early in order to rapidly intervene. One of the tools pediatricians use to track development is the standard growth chart. The growth charts are somewhat limited in predicting possible neurodevelopmental issues. They rely on linear models and assumptions of normality for physical growth data – obscuring key statistical information about possible neurodevelopmental risk in growth data that actually has accelerated, non-linear rates-of-change and variability encompassing skewed distributions. Here, we use new analytics to profile growth data from 36 newborn babies that were tracked longitudinally for 5 months. By switching to incremental (velocity-based) growth charts and combining these dynamic changes with underlying fluctuations in motor performance – as the transition from spontaneous random noise to a systematic signal – we demonstrate a method to detect very early stunting in the development of voluntary neuromotor control and to flag risk of neurodevelopmental derail.

No MeSH data available.


Frequency of noise-to-signal transitions distinguishes babies at high risk from typically developing babies. (A) Stationary (inner-quadrant) transitions sorted according to the proportion of times fluctuating within each quadrant before crossing to the other quadrant. Each dot represents a baby (up triangles are TD, circles are PAR and down triangles are HR). Inset is the median across each group. (B) Non-stationary (inter-quadrants) transitions sorted according to the proportion of times crossing across quadrants from the RLQ to the LUQ. The same index used to plot the opposite direction of transitions shows higher variability when crossing from the LUQ the RLQ. Inset shows the median values/group. (C) Median values of noise-to-signal transitions/group during the first visit already distinguish the groups in both the stationary and the non-stationary cases.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5120129&req=5

Figure 4: Frequency of noise-to-signal transitions distinguishes babies at high risk from typically developing babies. (A) Stationary (inner-quadrant) transitions sorted according to the proportion of times fluctuating within each quadrant before crossing to the other quadrant. Each dot represents a baby (up triangles are TD, circles are PAR and down triangles are HR). Inset is the median across each group. (B) Non-stationary (inter-quadrants) transitions sorted according to the proportion of times crossing across quadrants from the RLQ to the LUQ. The same index used to plot the opposite direction of transitions shows higher variability when crossing from the LUQ the RLQ. Inset shows the median values/group. (C) Median values of noise-to-signal transitions/group during the first visit already distinguish the groups in both the stationary and the non-stationary cases.

Mentions: The signatures of fluctuations in motor performance of the babies in Rank 1 group transitioned far more frequently from higher to lower noise and from highly skewed to more symmetrically shaped PDFs than those of babies in the other groups. This is depicted in Figure 4A for the stationary case where the transitions remain within the LUQ or within the RLQ. In this case, the babies are ranked according to the proportion of times that their signatures remained in a “steady-state” within one quadrant or the other. The inset of the Figure 4A depicts the three data-driven groups from the median ranked parameters of physical growth in Figure 3A. Specifically, babies in the first group (data-driven TD) have the highest proportion of remaining steady in the LUQ or the RLQ on average. The data-driven PAR group falls intermediate to data-driven TD and HR groups. The HR group has the lowest proportion of “steady-state” in LUQ or RLQ.


Neonatal Diagnostics: Toward Dynamic Growth Charts of Neuromotor Control
Frequency of noise-to-signal transitions distinguishes babies at high risk from typically developing babies. (A) Stationary (inner-quadrant) transitions sorted according to the proportion of times fluctuating within each quadrant before crossing to the other quadrant. Each dot represents a baby (up triangles are TD, circles are PAR and down triangles are HR). Inset is the median across each group. (B) Non-stationary (inter-quadrants) transitions sorted according to the proportion of times crossing across quadrants from the RLQ to the LUQ. The same index used to plot the opposite direction of transitions shows higher variability when crossing from the LUQ the RLQ. Inset shows the median values/group. (C) Median values of noise-to-signal transitions/group during the first visit already distinguish the groups in both the stationary and the non-stationary cases.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5120129&req=5

Figure 4: Frequency of noise-to-signal transitions distinguishes babies at high risk from typically developing babies. (A) Stationary (inner-quadrant) transitions sorted according to the proportion of times fluctuating within each quadrant before crossing to the other quadrant. Each dot represents a baby (up triangles are TD, circles are PAR and down triangles are HR). Inset is the median across each group. (B) Non-stationary (inter-quadrants) transitions sorted according to the proportion of times crossing across quadrants from the RLQ to the LUQ. The same index used to plot the opposite direction of transitions shows higher variability when crossing from the LUQ the RLQ. Inset shows the median values/group. (C) Median values of noise-to-signal transitions/group during the first visit already distinguish the groups in both the stationary and the non-stationary cases.
Mentions: The signatures of fluctuations in motor performance of the babies in Rank 1 group transitioned far more frequently from higher to lower noise and from highly skewed to more symmetrically shaped PDFs than those of babies in the other groups. This is depicted in Figure 4A for the stationary case where the transitions remain within the LUQ or within the RLQ. In this case, the babies are ranked according to the proportion of times that their signatures remained in a “steady-state” within one quadrant or the other. The inset of the Figure 4A depicts the three data-driven groups from the median ranked parameters of physical growth in Figure 3A. Specifically, babies in the first group (data-driven TD) have the highest proportion of remaining steady in the LUQ or the RLQ on average. The data-driven PAR group falls intermediate to data-driven TD and HR groups. The HR group has the lowest proportion of “steady-state” in LUQ or RLQ.

View Article: PubMed Central - PubMed

ABSTRACT

The current rise of neurodevelopmental disorders poses a critical need to detect risk early in order to rapidly intervene. One of the tools pediatricians use to track development is the standard growth chart. The growth charts are somewhat limited in predicting possible neurodevelopmental issues. They rely on linear models and assumptions of normality for physical growth data – obscuring key statistical information about possible neurodevelopmental risk in growth data that actually has accelerated, non-linear rates-of-change and variability encompassing skewed distributions. Here, we use new analytics to profile growth data from 36 newborn babies that were tracked longitudinally for 5 months. By switching to incremental (velocity-based) growth charts and combining these dynamic changes with underlying fluctuations in motor performance – as the transition from spontaneous random noise to a systematic signal – we demonstrate a method to detect very early stunting in the development of voluntary neuromotor control and to flag risk of neurodevelopmental derail.

No MeSH data available.