1. IntroductionPremature infants are born at a gestational age ranging from 22 to 37 weeks, with a body weight of up to 2500 g [1]. Recent studies have reported that pre-term infants are more prone to neurodevelopmental impairment (NDI) than full-term infants [1,2,3]. Moreover, there is evidence that the occurrence of cerebral palsy (CP) is associated with pre-term birth and extremely low birth weight in infants [4,5,6]. The prediction of NDI (including CP) in pre-term infants is challenging, as it requires the use of diagnostic methods and tools to recognize early symptoms (occurring during the first 3 months of life) that would indicate later development of NDI [7]. The results of the review of the literature showed that Prechtl’s general movements assessment (GMA) is currently the most evidence-based clinical approach for the prediction of motor impairment in high-risk infants [7,8]. GMA is a standardized functional assessment based on the subjective visual assessment of young infants’ general movement (GM) patterns using the Gestalt perception phenomenon, whose advantages are both the ability to perceive complex entire patterns or configurations and sensitivity to individual components [9,10,11,12]. The presence of normal GMs, especially fidgety movements (FMs) at 3 to 5 months after term, especially if they co-occur with other smooth and fluent movements, are very likely to show neurologically normal development [11,12]. Conversely, the presence of abnormal GMs, in general, indicate an increased risk for later neurological dysfunction [9,10]. While poor repertoire GM patterns [PR] are not early markers of adverse neurological outcomes, the consistent cramped-synchronized GMs patterns [CS] at pre-term and term age, and the absence of FMs at post-term age, are early signs of an atypical outcome, mainly CP [12,13,14].GMA presents high reliability for predicting CP in high-risk infants, with a reported sensitivity of 98% (95% confidence interval, CI 74–100%) and specificity of 91% (95% CI 83–93%) [15,16]. Even though GMA currently presents the highest value for predicting NDI, it is based on a subjective visual assessment. With advancing computer science, multiple systems based on sensor technology have been adapted for the identification of GMs in young infants [17,18]. There are several study articles on monitoring the body movements of infants based on either direct sensing using body-worn miniaturized movement sensors [19], indirect sensing based on visual sensors (2D or 3D video) [20,21,22,23], or both [24,25]. While the obtained results of GMA based on motion sensors appear very promising, none are sensitive enough to determine such movement parameters that are able to distinguish normal from abnormal GMs [17,18,19,26]. Moreover, according to the results of a review of “automated movement recognition technologies to assess infant movement” performed by Marcroft et al., the applicability of these techniques is limited by difficulties attaching special markers to the very small limbs of infants [27]. In recent years, center-of-pressure (CoP) data have been used for the analysis of infant movements [28,29]. COP is defined as the point where the ground reaction force is applied, and it represents the center point of the entire pressure of the body in the ground contact surface, and the analysis of CoP trajectory and CoP area are common in the study of human postural control [30]. Although, the CoP analysis of these studies was sufficient to screen the infants grouped by birth status (pre-term/term) [28,29] these studies did not relate to any observational clinical classification of high risk for CP according to the GM concept. Although, several studies compared various CoP metrics between infants grouped by future motor outcome (typically developing infants vs. infants with or at risk for CP), using force plates, they were used in infants over one year and assessed sitting postural control [31,32].

To the best of our knowledge, the validation of CoP in young infants (up to 3 months) grouped by future motor outcome (typical/impaired postural control) and measured using a force platform in a horizontal (supine or prone) position has not been published thus far.

Taking into consideration the close relationship between postural control and movement [33], it can be hypothesized that abnormal GMs are accompanied by atypical postural control. Thus, the current study attempted to identify typical vs. impaired postural control in pre-term infants in horizontal positions (i.e., supine and prone) using the force platform. Using clinical GMA as a reference, the present study aimed to characterize a new posturometric test based on CoP movement analysis in terms of design and construct validity for the detection of postural control disturbances in pre-term infants. For this purpose, comparative studies were carried out between pre-term infants who presented normal FMs (indicating normal future motor outcomes) vs. absent FMs (indicating later neurological dysfunction). 3. Results

Participants were included in either the tested group (infants presented absent FMs) or the control group (infants presented normal FMs) based on the GMA. The inter-assessment agreement of stratification between absent FMs and normal FMs was perfect in both term and post-term GMA (ICC = 0.996–1.00 and ICC = 0.996–1.00; 95% confidence interval) (ICC = 0.986–1 and ICC = 0.985–1, respectively).

Statistical analyses revealed significant differences between the infants from the tested group and the control group for all CoP parameters describing spontaneous CoP displacement in the supine position (Table 3). This was true for both types of posturometric indices, i.e., for those based on CoP shifts (SPL, VmaxCoP) and those based on the surface area of the CoP (ACoP, MCoPx, and MCoPy) (Table 3). Furthermore, the main postural parameter describing spontaneous sway of CoP (SPL) was twice as short in infants from the tested group compared with controls. Additionally, it was also observed that infants who presented absent FMs (tested group) showed a half smaller range of spontaneous CoP displacement, on average, in both the linear direction medial-lateral (MCoPx) and anterior-posterior (MCoPy) than those who presented with normal GMs (control group) in the supine position (Table 3). Although the differences between groups (tested vs. control) in the main postural control parameters describing spontaneous sway of CoP, such as SPL and VmaxCoP, in prone were not as great as in supine, they differed significantly (Table 4). However, no statistically significant differences were found between parameters describing the surface area of the CoP, such as ACoP, MCoPx, and MCoPy between infants with absent (study group) and normal (control group) FMs (Table 4). To determine the overall accuracy of the new posturometric test, the AUC values were used. AUC values greater than 0.9 were considered to be outstanding discrimination, those from 0.8 to 0.9 were excellent, and those from 0.7 to 0.8 were acceptable discrimination, while values less than 0.7 represented nonacceptable discrimination. A value of 0.5 means random discrimination [36]. The highest discrimination values between the tested group and the control group were shown for both CoP parameters describing spontaneous sway in the supine position, i.e., SPL and VmaxCoP (outstanding discrimination; AUC > 0.9). Nevertheless, the parameters describing the surface area of the CoP in the supine position (ACoP, MCoPx, and MCoPy) had an excellent discriminant value for the normal and absent FM patterns (AUC > 0.8) (Table 5). Considering the characterization of the postural control parameters in the prone position observed that the discriminant value for postural parameters describing spontaneous sway of CoP, such as SPL and VmaxCoP, was acceptable discrimination (AUC > 0.7), while it was nonacceptable and statistically insignificant for parameters describing the surface area of the CoP, such as ACoP, MCoPx, and MCoPy (AUC = 0.65) (Table 5). The best cut-off point (highest sensitivity while maintaining highest specificity) for individual variables with a significance test are presented in Table 5 and Figure 1. The ROC curves and the optimal cutoff point for individual postural parameters in two conditions of examination in supine (A) and prone (B) positions are presented in Figure 3. The diagonal line represents no discrimination, while the curves represent the sensitivity and specificity of individual postural parameters at different cutoff points. 4. Discussion

The design and construct validity of the new PT for the recognition of postural control disturbances in pre-term, high-risk infants was determined by comparing its outcomes with the results of the assessment of general movements (GMA) for the early recognition of neurological deficits in pre-term infants.

The main finding of our study was the recognition of the correlation between abnormalities in postural control measured by new PT in pre-term infants with their absent FM patterns at 12–14 weeks post-term age. The comparison analysis between pre-term infants with absent FMs vs. pre-term infants with normal FMs in PT in a supine position presented altered postural control parameters in infants with absent FMs, such as (1) significantly shorter sway path length, (2) significantly slower velocity of CoP displacement, and (3) a significantly smaller range of spontaneous CoP displacement in both linear directions, i.e., medial-lateral and anterior-posterior in comparison with normally developing pre-term infants (controls). Differences between groups (tested vs. control) regarding the main postural control parameters describing spontaneous CoP sway and area in the prone position were not as evident as in supine.

To assess whether the new PT data were able to discriminate between infants who presented absent FMs (study group) from infants who presented normal FMs (control group) the Receiver Operating Characteristic (ROC) curve was applied. The outstanding discriminant capacity (AUC > 0.9) of both CoP parameters describing spontaneous sway in the supine position, i.e SPL and VmaxCoP, and excellent discriminant value of the parameters describing the surface area of the CoP in the supine (ACoP, MCoPx, and MCoPy) in examination in supine position between groups confirmed that new PT has the ability to discriminate between pre-term infants at risk for neuromotor deficits, i.e., those who presented absent FMs and normally developing pre-term infants, i.e., these who presented normal GMs.

The strength of this study is the use of the pressure-sensitive platform based on very sensitive transducers and equipped with a large tabletop, resembling a changing table for babies adapted for the safe examination of infants in horizontal positions. So far postural control in children has been assessed by means of posturography, i.e., measurement of spontaneous CoP displacement using a force platform. However, the force platforms (e.g., Kistler, AMTI) are limited in their clinical utility for posturometric assessment of infants because they are not sufficiently adapted to the specific anthropometric characteristics of infants in the supine position, such as small body size and very low body weight.

Validation of CoP measured using a force platform in a horizontal (supine or prone) position has not been published thus far. Moreover, only a few studies have been concerned with the recognition of postural control of infants in the first months of life using CoP methodology [23,28,29,31,32]. Two studies by Dusing and co-workers assessed the displacements of CoP in the supine position at an early stage of infants’ development using a pressure-sensitive mapping method, but it was related to the comparison between pre-term and full-term infants [28,29]. There were recognized that premature infants exhibited more stereotypic patterns of movement, resulting in larger, but repetitive, CoP excursions than full-term infants. However, both examined populations (pre-term and full-term infants) as well approach to the CoP movement analysis in the above studies were different than ours. Therefore, these results cannot be directly compared. Probably, most similar to the presented study is the work of Støen and colleagues who analyzed supine infant movement relative to an observational clinical classification of high risk for CP according to GMs concept, where one of the measures was the standard deviation of the movement of the centroid over the duration of the recording [23]. Støen and colleagues have reported that infants with absent fidgety movements, i.e., infants at risk for motor impairment demonstrated greater variability in the centroid movement i.e., greater instability during movement [23]. The above study quantified the magnitude of the variability in postural control using root mean squared of the CoP displacement, while our study was based on spontaneous sway of CoP and velocity of the CoP displacement evaluation, so these results cannot be directly compared with our findings.

Due to the lack of reference for the results of posturometric measurements of supine postural control in pre-term infants up to three months of age, the obtained results in our study are difficult to compare with those found in previous studies. This study is a feasibility study and reports pilot findings; therefore, it is difficult to draw definite conclusions.

Nevertheless, the current results showed that the presence of abnormalities in postural control in preterm infants as measured by new PT correlates with their absent FMs pattern at 12–14 weeks post-term age. Based on these preliminary results, it can be concluded that the new PT in the supine position based on measurement of the CoP displacement has been demonstrated to be valid and can be a particularly revealing indicator for the development of postural control abnormalities in pre-term infants.

Further construct validity and reliability studies of presented PT are needed to provide some evidence that pre-term infants at risk for neuromotor deficits who present abnormal GMs and normally develop pre-term infants, i.e., these patients who presented with normal GMs differ in the nature of the development of postural control.

5. Study Limitations

This study has several limitations that must be considered and addressed in follow-up studies. First, due to the fact, that commonly used for CoP displacement measurements force plates (e.g., Kistler, AMTI) are not sufficiently adapted to the specific anthropometric characteristics of infants, such as body size and first of all very low body weight of the pre-term infant, the custom-built force plate used in our study has not been compared with state-of-the-art force platforms. Second, due to large differences in sampling frequency (our force plate 50 Hz vs. sensory mats only 5 Hz), it has not been compared with pressure-sensitive mats (PSM). Third, the number of participants in our study seemed relatively small; however, the low prevalence of CP (and thus abnormal GMs), which remains at 2–3 per 1000 live births, was a significant limitation in recruiting a greater number of infants with consistently presented abnormal GMs in the tested group for this study.