Supine lying center of pressure movement characteristics as a predictor of normal developmental stages in early infancy

Abstract

BACKGROUND:

Absent or abnormal fidgety movements in young infants are associated with subsequent diagnoses of developmental disorders such as cerebral palsy. The General Movement Assessment (GMA) is a qualitative clinical tool to visually identify infants with absent or abnormal fidgety movements associated with developmental stage, yet no quantitative measures exist to detect fidgety activity.

OBJECTIVE:

To determine whether a correlation exists between quantitative Center of Pressure (CoP) measurements during supine lying and age.

METHODS:

Twenty-four healthy full-term infants participated in the Institutional Review Board-approved study. Participants were placed supine in view of a GoPro camera on an AMTI force plate for two minutes. Spontaneous movements were evaluated by three trained raters using the GMA. Traditional CoP parameters (range, total path length, mean velocity, and mean acceleration of resultant CoP) were assessed, and complexity of each of the resultant CoP variables (location, velocity, and acceleration) was calculated by sample entropy. Linear regression with Pearson correlation was performed to assess the correlations between the CoP parameters and adjusted age.

RESULTS:

Nineteen infants were deemed fidgety per the GMA and were included in further analyses. All Sample entropy measures and range of resultant CoP had significant correlations with adjusted age ( $p<$ 0.05). Sample entropy of resultant CoP decreased with increasing age while range of resultant CoP increased with increasing age.

CONCLUSION:

The results suggest that complexity of CoP and range of CoP are good predictors of age in typical developing infants during the fidgety period. Therefore, an approach using these parameters should be explored further as a quantifiable tool to identify infants at risk for neurodevelopmental impairment.

Keywords

Center of pressure supine lying sample entropy baby neurodevelopment fidgety

1. Introduction

Qualitative evaluation of spontaneous movements is often utilized to predict neurodevelopmental impairments in young infants. One such tool is Prechtl’s General Movement Assessment (GMA), which has been established as a predictor of cerebral palsy (CP) diagnoses [1, 2, 3]. The GMA classification is age-dependent since the characteristics of spontaneous movements evolve with age [4]. The classification of general movements via GMA is based on the quality of the infant movements focused on complexity, fluency, and variation during supine lying positioning [5]. The GMA classifies general movements as “normal”, “absent”, or “abnormal” based on the impression of the clinicians. Thus, the current GMA method is purely observational and qualitative.

Fidgety movement is a type of general movement characterized as small movements of the neck, trunk, and limbs at moderate speed with variable acceleration in all directions [1, 6, 7]. Fidgety movements may be seen as early as 6 weeks post-term but are usually predominantly observed from 9 to 20 weeks post-term. If the fidgety movements are never observed from 9–20 weeks post-term, infants are classified as absent fidgety on the GMA. Additionally, the fidgety movements which are moderately or greatly exaggerated in amplitude, speed and jerkiness are considered as abnormal fidgety movements. Absent or abnormal fidgety movements have been associated with subsequent diagnoses of developmental disorders such as CP [1, 2, 3], and the GMA is the current clinical tool for identifying these infants early in life. Early intervention can optimize motor, cognition, and communication outcomes with promotion of learning and neuroplasticity as well as prevent secondary impairments and minimize influence of complications [8], yet the GMA is solely based on clinicians’ observation, with poor inter-rater reliability [9]. Although the GMA shows high sensitivity and specificity (95% and 96%) of fidgety movement assessment [1], it still provides limited discrete values rather than continuous values. Thus, a robust and objective quantification of general movements in the fidgety age of infancy is necessary to provide a refined assessment tool for better identification and classification of the risk for neurodevelopmental disorders and would benefit from early intervention.

Entropy measures of infants’ center of pressure (CoP) have previously been utilized to identify abnormal movement patterns in early infancy [10, 11, 12, 13]. Sample entropy (SampEn) quantifies the structure of variability and randomness in time series data [14]. Supine lying entropy measures have previously been used to determine differences in CoP movement patterns between full term vs. pre-term babies [10, 11]. SampEn measures of CoP have also been used to distinguish sitting postural control between infants with typical vs. delayed development [12]. A recent study reported that infants at high risk of developmental delay showed smaller supine lying SampEn values of spontaneous leg movement, as measured by wearable accelerometers [13]. Supine lying SampEn of infants, therefore, may be a feasible tool to assess abnormal infant movements related to potential developmental disorders. The purpose of this study was to determine whether quantitative CoP measurements during supine lying were associated with age in healthy full-term infants. We hypothesized that the complexity of CoP (SampEn) during supine lying would be significantly correlated with age in infants with normal fidgety movement.

2. Methods

After recruiting caregivers of potential participants via flyers, phone screen interviews were performed to see if their infant would qualify for this Institutional Review Board-approved study. Infants were excluded if they had any orthopaedic or neurological conditions or had received vaccinations within two weeks prior to data collection. Infants included in the study were born $>$ 37 weeks gestation and between 5–95th percentile in height and weight [15]. Infants’ ages were adjusted for pre- or post-maturity from a term pregnancy (40 weeks) using the chronological and gestational ages, per GMA guidelines [16]. As the movement pattern changes over time and is determined based on post-menstrual age, it is important to correct for all infant’s age born at less than 40 weeks gestational age, for accurate assessment for their movement patterns [6]. For example, a baby born at a gestational age of 38 weeks is 2 weeks premature. If the baby’s actual age is 20 weeks, then the corrected age would be 18 weeks old. Twenty-four healthy full-term infants (age: 16.3 $\pm$ 5.6 weeks; 10M, 14F; length: 65.5 $\pm$ 4.6 cm; weight: 6.4 $\pm$ 1.2 kg) were enrolled after caregivers consented and signed a permission form as approved by the University of Arkansas for Medical Sciences Institutional Review Board. Infants were placed supine on an AMTI force plate (Watertown, MA; 1000 Hz) in orthogonal view of a GoPro camera (San Mateo, CA). Data was collected for 2 minutes (Fig. 1). A force platform is the most commonly used type of force transducer in human movement science, considered as a gold standard for measuring 3-D ground reaction forces (GRF). The force platform incorporates stain gages to determine a single 3-D GRF vector. The location of the force vector usually called center of pressure (COP) as the force vector results from a distribution of forces to a contact area on the surface of the force platform. Thus, COP is located within (x, y) coordinate pair of the force platform [17]. The camera was used to record infants’ movements while lying on their back on the floor (force platform). Three trained raters (TJ, NK, and BM) assessed the video by observing spontaneous movements of the infants and using GMA. One- or two-minutes representative data are sufficient to observe normal fidgety movement [18]. Infants with any discrepancies on the GMA scoring between the three raters were considered as abnormal fidgety. Nineteen infants (adjusted age: 14.7 $\pm$ 4.5 weeks; 9M, 10F; length: 59.9 $\pm$ 2.7 cm; weight: 6.2 $\pm$ 0.8 kg) were deemed fidgety per the GMA and were included in further analyses.

Figure 1.

Test setup (force plate – dashed outline).

2.1 Data analysis

Raw CoP data from the force plate was down sampled to 200 Hz, with the middle 100 seconds of CoP data used for analysis. The CoP data were filtered using a low pass 4 ${}^{\text{th}}$ order Butterworth filter with a cutoff frequency of 5 Hz [19]. Infants whose length (i.e. height) exceeded the longest dimension of the force platform (45 cm $\times$ 50 cm) were placed diagonally on the force platform and thus the resultant CoP value (RCoP) was calculated using anterior-posterior and medial-lateral CoP data. Complexity of RCoP parameters (RCoP, RCoP velocity, RCoP acceleration) was calculated by SampEn. Numerous studies have used SampEn to assess variability in different time-series signals such as heart rate, gait and postural control, and electroencephalogram (EEG) data [20, 21, 22, 23]. SampEn quantifies the regularity or predictability of a given time-series [24]. Specifically, SampEn calculates the probability that a sequence of data points, having repeated itself within a tolerance $r$ for a window length $m$ , will also repeat itself for $m+1$ points, without allowing self-matches [14]. Smaller sample entropy values indicate greater regularity and less randomness. In this study, SampEn was calculated from the COP time-series by using the algorithm by Richman and Moorman, 2000:

$\displaystyle\text{SampEn}({m,r,N})=-\ln\left[{\frac{C^{m+1}(r)}{C^{m}(r)}}\right]$

Where $N$ is the number of COP data points, and $C^{m}$ is the number of matched pairs with length $m$ within tolerance $r$ , for every possible window vector. Two entropy parameters: $r$ and $m$ were set at 0.1 and 3, respectively, selected within appropriate theoretical ranges [25]. In addition, range of RCoP, total path length of RCoP, mean absolute RCoP velocity, and mean absolute RCoP acceleration were assessed using custom MATLAB (R2019a; Natick, MA) code.

2.2 Statistical analysis

There were seven dependent variables: SampEn of RCoP, RCoP velocity, and RCoP acceleration (complexity parameters); range, total path length, mean absolute velocity, and mean absolute acceleration (traditional CoP parameters). Descriptive statistics were calculated for each RCoP variable (Table 1). Correlations between the CoP parameters and adjusted infant age were assessed using Pearson’s $r$ . Simple linear regressions were used to establish predictive models for RCoP variables that were found significantly associated with adjusted age. The level of statistical significance for all tests was set at $\alpha<$ 0.05. Statistical analyses were performed using SPSS ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ (version 23; Chicago, IL, USA).

3. Results

Significant correlations were observed for SampEn of RCoP ( $r=$ 0.501, $p=$ 0.027) and range of RCoP ( $r=$ 0.609, $p=$ 0.006) with adjusted age (Table 1). However, other variables such as SampEn of RCoP velocity, SampEn of RCoP acceleration, total path length, mean absolute velocity, and mean absolute acceleration were not significantly associated to the adjusted age (Table 1). In the linear regression model, SampEn of RCoP and range of RCoP were significant predictors, explaining 25% and 37% of the variance in adjusted age, respectively (Fig. 2).

Table 1
Correlations between CoP parameters and adjusted age

Variable	Mean $\pm$ SD		Correlation (r)	$p$
SampEn of RCoP	0.112	$\pm$ 0.053	0.501	0.029
SampEn of RCoP velocity	0.257	$\pm$ 0.047	0.162	0.506
SampEn of RCoP acceleration	0.290	$\pm$ 0.044	0.573	0.548
Range of RCoP (cm)	5.36	$\pm$ 3.27	0.609	0.006
Total path length of RCoP (cm)	367	$\pm$ 124	0.197	0.418
Absolute mean RCoP velocity (cm/s)	1.86	$\pm$ 0.82	0.132	0.589
Absolute mean RCoP acceleration (cm/s ${}^{2}$ )	27.8	$\pm$ 13.3	0.043	0.862

Figure 2.

Correlations between adjusted infant age (weeks) and A) Sample Entropy of RCoP and B) range of RCoP.

4. Discussion

The purpose of this study was to determine whether quantitative CoP measurements during supine lying were associated with age in healthy full-term infants. Complexity of RCoP parameters (RCoP, RCoP velocity, RCoP acceleration) and traditional CoP parameters (range, total path length, mean absolute velocity, and mean absolute acceleration of RCoP) were calculated and then correlations between these CoP parameters and infant age were assessed. Infants in the fidgety period exhibited lower complexity with increasing age. Interestingly, SampEn of RCoP was significantly associated with infant age. A non-linear data analysis approach has been previously used to identify infants at high risk for developmental delay. Dusing et al. reported that full-term infants at 1 to 3 weeks of age had greater entropy values of CoP (higher complexity) compared to pre-term infants of the same age during supine lying, which indicates the pre-term infants (who have a higher risk of developmental delay compared to full-term infants) exhibited lower complexity of their CoP movement [10]. Deffeyes et al. also used entropy measures of CoP to detect differences between infants and toddlers with typical vs. delayed development and reported greater entropy values for infants with typical development during sitting postural control [12]. Furthermore, Smith et al. found that infants with typical development showed increased SampEn of spontaneous leg movements compared to infants at risk of developmental disorders [13]. These previous findings support the idea that the entropy measure of CoP can be a useful tool to detect infants at high risk of developmental disorders. Our results also indicate that this approach might be useful to develop a screening tool for neurodevelopmental disorders in early infancy (fidgety period: $\sim$ 6–22 weeks) [26] aimed at early diagnosis and interventions, but this needs to be further explored.

In addition to the nonlinear SampEn parameters, range of RCoP was a significant predictor of infant age while other traditional CoP parameters appeared not to be associated with infant age. Range of RCoP had the highest correlation ( $r=$ 0.609; Table 1) with infant age among the dependent variables and infants’ CoP movements tend to be greater with increasing age. In order to determine if range of RCoP was simply related to increased body size which occurs as infants age, we performed a post-hoc calculation of the correlation between infant height and range of RCoP. An insignificant correlation was found ( $r=$ 0.320, $p=$ 0.182), indicating that infants’ length did not affect the CoP results as a potential confounder. In general, as babies grow by 26 weeks (6 months) of age, reflexes tend to disappear and be replaced with cortex-directed voluntary movements including head control, posture, and various type of locomotion [27]. Greater range of RCoP with increasing age may reflect this concept of the development stage of infants, indicating that infants develop their movements to be more gross and regular (evidenced by lower SampEn) movements such as changing posture, kicking and initiation of rolling from supine to prone. Therefore, range of CoP is a highly correlated parameter with age and should be included in developing a diagnosis method when considering using a multivariate approach.

Footnotes

Acknowledgments

The research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under Award Number P20GM125503.

Conflict of interest

None to report.

References

Prechtl

Einspieler

Cioni

Bos

Ferrari

Sontheimer

. An early marker for neurological deficits after perinatal brain lesions. Lancet. 1997 May 10; 349(9062): 1361–3. doi: 10.1016/S0140-6736(96)10182-3.

Olsen

Marschik

Spittle

. Do fidgety general movements predict cerebral palsy and cognitive outcome in clinical follow-up of very preterm infants? Acta Paediatr. 2018 Feb; 107(2): 361–2. doi: 10.1111/apa.14126.

Datta

Furrer

Bernhardt

Hüppi

Borradori-Tolsa

Bucher

, et al. Fidgety movements in infants born very preterm: Predictive value for cerebral palsy in a clinical multicentre setting. Dev Med Child Neurol. 2017 Jun; 59(6): 618–24. doi: 10.1111/dmcn.13386.

Chen

García-Betances

Liming

Cabrera-Umpiérrez

. Smart Assisted Living Toward An Open Smart-Home Infrastructure. Springer International Publishing; 2019.

Einspieler

Bos

Libertus

Marschik

. The general movement assessment helps us to identify preterm infants at risk for cognitive dysfunction. Front Psychol. 2016; 7: 406–14. doi: 10.3389/fpsyg.2016.00406.

Prechtl

. General movement assessment as a method of developmental neurology: New paradigms and their consequences. The 1999 Ronnie MacKeith lecture. Dev Med Child Neurol. 2001 Dec 18; 43(12): 836–42. doi: 10.1017/s0012162201001529.

Einspieler

Prechtl

. Prechtl’s assessment of general movements: A diagnostic tool for the functional assessment of the young nervous system. Ment Retard Dev Disabil Res Rev. 2005; 11(1): 61–7. doi: 10.1002/mrdd.20051.

Novak

Morgan

Adde

Blackman

Boyd

Brunstrom-Hernandez

, et al. Early, accurate diagnosis and early intervention in cerebral palsy. JAMA Pediatr. 2017 Sep 1; 171(9): 897. doi: 10.1001/jamapediatrics.2017.1689.

Bernhardt

Marbacher

Hilfiker

Radlinger

. Inter- and intra-observer agreement of Prechtl’s method on the qualitative assessment of general movements in preterm, term and young infants. Early Hum Dev. 2011 Sep; 87(9): 633–9. doi: 10.1016/j.earlhumdev.2011.04.017.

10.

Dusing

Kyvelidou

Mercer

Stergiou

. Infants born preterm exhibit different patterns of center-of-pressure movement than infants born at full term. Phys Ther. 2009 Dec 1; 89(12): 1354–62. doi: 10.2522/ptj.20080361.

11.

Dusing

Thacker

Galloway

. Infant born preterm have delayed development of adaptive postural control in the first 5 months of life. Infant Behav Dev. 2016 Aug; 44: 49–58. doi: 10.1016/j.infbeh.2016.05.002.

12.

Deffeyes

Harbourne

Dejong

Kyvelidou

Stuberg

Stergiou

. Use of information entropy measures of sitting postural sway to quantify developmental delay in infants. J Neuroeng Rehabil. 2009 Aug 11; 6: 34. doi: 10.1186/1743-0003-6-34.

13.

Smith

Vanderbilt

Applequist

Kyvelidou

. Sample entropy identifies differences in spontaneous leg movement behavior between infants with typical development and infants at risk of developmental delay. Technologies. 2017 Sep 2; 5(3): 55. doi: 10.3390/technologies5030055.

14.

Richman

Moorman

. Physiological time-series analysis using approximate entropy and sample entropy. Am J Physiol Circ Physiol. 2000 Jun; 278(6): H2039–49. doi: 10.1152/ajpheart.2000.278.6.H2039.

15.

CDC.gov, 2006. WHO growth standards – https://www.cdc.gov/growthcharts/index.htm.

16.

Spittle

Boyd

Inder

Doyle

. Predicting motor development in very preterm infants at 12 months’ corrected age: The role of qualitative magnetic resonance imaging and general movements assessments. Pediatrics. 2009 Feb 1; 123(2): 512–7. doi: 10.1542/peds.2008-0590.

17.

Gordon

Robertson

Caldwell

Hamill

Kamen

Whittlesey

. Research Methods in Biomechanics. Second ED. Human Kinetic; 2014.

18.

Ricci

Einspieler

Craig

. Feasibility of using the general movements assessment of infants in the United States. Phys Occup Ther Pediatr. 2018 May 27; 38(3): 269–79. doi: 10.1080/01942638.2017.1395380.

19.

Hansen

Wei

Shieh

J-S

Fourcade

Isableu

Majed

. Sample entropy, univariate, and multivariate multi-scale entropy in comparison with classical postural sway parameters in young healthy adults. Front Hum Neurosci. 2017 Apr 26; 11: 206. doi: 10.3389/fnhum.2017.00206.

20.

Kang

Chen

. EEG entropy analysis in autistic children. J Clin Neurosci. 2019 Apr 1; 62: 199–206. doi: 10.1016/j.jocn.2018.11.027.

21.

Huijben

van Schooten

van Dieën

Pijnappels

. The effect of walking speed on quality of gait in older adults. Gait Posture. 2018 Sep 1; 65: 112–6. doi: 10.1016/j.gaitpost.2018.07.004.

22.

Terada

Kosik

Johnson

Gribble

. Altered postural control variability in older-aged individuals with a history of lateral ankle sprain. Gait Posture. 2018 Feb 1; 60: 88–92. doi: 10.1016/j.gaitpost.2017.11.009.

23.

Weippert

Behrens

Rieger

Stoll

Kreuzfeld

. Heart rate variability and blood pressure during dynamic and static exercise at similar heart rate levels. PLoS One. 2013 Dec 13; 8(12). doi: 10.1371/journal.pone.0083690.

24.

Pincus

. Approximate entropy as a measure of system complexity. Proc Natl Acad Sci U S A. 1991; 88(6): 2297–301. doi: 10.1073/pnas.88.6.2297.

25.

Stergiou

. Nonlinear analysis for human movement variability. CRS Press; 2016.

26.

de Vries

Visser

Prechtl

. The emergence of fetal behaviour. I. Qualitative aspects. Early Hum Dev. 1982 Dec; 7(4): 301–22. doi: 10.1016/0378-3782(82)90033-0.

27.

Ollendick

Schroeder

. Encyclopedia of Clinical Child and Pediatric Psychology. Springer US; 2012.

Supine lying center of pressure movement characteristics as a predictor of normal developmental stages in early infancy

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSION:

Keywords

1. Introduction

2. Methods

2.2 Statistical analysis

3. Results

Table 1 Correlations between CoP parameters and adjusted age

Footnotes

Acknowledgments

Conflict of interest

References

Table 1
Correlations between CoP parameters and adjusted age