According to the hypothesis of Adolph et al [16], motor dexterity development will be accompanied by a decrease of intraindividual variability. Those children who are faster with the pegs will show less intraindividual variability, and this result will be reflected in medium- or high-level correlations between placement and removal times of each hand and the SDs.

To prove this hypothesis, we have designed a study which could determine the variability of movement times during a playing activity comprising the removal and placement of pegs in a pegboard. A total of 39 children were interviewed, 21 boys and 18 girls. The children were aged between 25 and 41 months (mean=34.03, SD 3.96) and attended a day care center for 0- to 3-year-olds, located at the university campus on the northern side of Madrid.

Once the school agreed to participate in the study, a letter was sent to the families of the 3 groups of children in the upper grade (2- to 3-year-olds), with a summary of the project and a reference to a more detailed document at their disposal in the school to obtain informed parental consent from each family (see [33] for a detailed description on ethical cautions). The project had been previously approved by the university Commission of Ethics in Science. The children were interviewed in the school setting, in a quiet room by one of the researchers, while another researcher videotaped the child-pegboard interaction and a third one supervised the registration of the activity information. Every child was invited to play with the board and move 5 of the 10 pegs, finding out how the lights would change. After this familiarization activity, the test started, and each child was invited to move the 10 pegs from their original position in row A (the row positioned facing the professional) to the empty holes in row B (the row positioned facing the child; first trial). Once finished, the child was asked to perform the same task with the other hand (second trial). A snapshot of the videotaped activity can be seen in Figure 1.

Given that the traditional pegboards do not allow to measure the intraindividual variability during its use by a single child, we have designed and developed a prototype of a sensor-based electronic pegboard. The physical design of the pegboard proposed in this paper has been based on the current designs used for dexterity tests to maintain the reliability of the tests as much as possible. The design of the board is based on a series of requirements derived from (1) the experts’ needs related to the assessment tests (eg, the mobility-driven design of the device or its size), (2) the children’s interactions with this type of tools (ie, the necessity of a user-friendly simple interface and interaction without any special learning requirement for its use), and (3) the technological limitations and its costs. These requirements have been compiled from different sources such as the literature on pegboard tests, experts’ knowledge of the matter, and the documentation on the available sensor and communications technologies in these environments. Considering these requirements, the proposed electronic pegboard design has been developed as a modular system comprising the pegboard itself, a data collector module, and a user interface accessed through mobile or desktop devices (an example of this interface is shown in Figure 2).

The modules are independent and communicate between each other using a wireless communications system. The full components overview and its communications can be seen in Figure 3. Further details on the modules’ design can be found in Multimedia Appendix 1.

To determine the reliability of the pegboard, an additional study was carried out. In this study, the total trial times of the activity of 17 children with the pegs were recorded manually by an expert in manual dexterity using a stopwatch and was compared with the total times obtained electronically. The children were aged between 30 and 41 months and completed 2 trials, one with each hand. Hereafter, the measurements were analyzed, comparing the manually registered with those using the electronic pegboard, finding a high Pearson correlation between the manually and the electronically measured times of the first (r=0.998; P<.001) and the second trials (r=0.997; P<.001).

The electronic pegboard has enabled the study of intraindividual variability as the system allows for recording 10 placement times (for each peg) and 9 removal times (for pegs 2 to 10, excluding the first one) in a single trial, instead of 1 single time measurement for the total amount of pegs. In addition, it is possible to obtain the individual profiles of task performance for each child. In the following paragraphs, results concerning the intraindividual variability and individual profiles are reported.

Specifically, we have recorded the mean and intraindividual values (IIV) for each child performing the activity. The values have been classified depending on the hand used for the movement and the type of movement (placement or removal of pegs) in 4 groups:

The IIV value for each child is calculated as the SD of the times measured during the activity for each group. These values along with the mean time taken for each movement can contribute to determine if there is a relationship between the intraindividual variability, measured as the time taken to move the pegs, and the dexterity (the mean speed of placement or removal for each hand).

A correlation analysis among age, the 4 mean measures of dexterity (mean RP, mean LP, mean RR, and mean LR), and the corresponding IIV values was performed. Results (see Table 1) show various negative correlations with age (eg, older children are faster at placing pegs with the right hand). Moreover, positive correlations between mean values and IIV for each hand and movement were found, from r=0.46 (P=.003) to r=0.72 (P<.001).

The increased measurement possibilities with the electronic pegboard makes it possible to obtain individual profiles of the children showing the degree of variability in their performances [15]. As an example, Figure 4 shows placement times of the 10 pegs with both the right and the left hands of 2 children, both 30 months old. The mean time of Child A’s activity with the left hand is 2174.40 millisecond and the IIV value is 728.30 millisecond; with the right hand, mean=1418.78 millisecond and IIV=272.31 millisecond. Child B’s mean time for task performance with the left hand is 2089.78 millisecond and IIV=463.20 millisecond; with the right hand, mean=1725.00 millisecond and IIV=416.04 millisecond.

For instance, by comparing these results for the 2 children, the differences in their profiles can be observed. Child A shows a lower mean and variability of the right hand compared with the left hand, which possibly reflects the further development of a right manual preference. However, Child B displays more comparable means and very similar variabilities of both hands, and therefore no clear manual preference.

The pilot study carried out with children has proven to be useful to study the intraindividual variability in children. On the basis of Adolph et al’s [16] hypothesis, a lower intraindividual variability was expected in individuals with lower placement and removal times as their motor control increases. Our results show this relation in 3-year-old children, thereby confirming the initial hypothesis.

Nevertheless, some limitations can be observed in this preliminary study. A first limitation, as a consequence of its exploratory nature, is the low number of participating children. Although significant results were found, this low number of participants did now allow applying Bonferroni corrections, which is necessary when the number of correlations is high, as the results showed. Second, the age ranges of the participants were reduced, which might have caused lower correlations because of the increased homogeneity of the studied population. A third limitation lies in the absence of additional testing of the motor development level and manual dexterity of the children, as well as their manual preference, even though this may not be apparent yet. These issues are being addressed in subsequent studies.

From a technological point of view, we have presented in this paper a novel sensor-based pegboard design, which shows interesting features for the motor development in children. The study carried out can be seen also as a validation test of the tool design, as the 2- and 3-year-old children have been able to perform all the requested activities without any added difficulty derived from its design.

In future research, we will work on the integration of the pegboard within an ecosystem of similar sensor-based wireless devices that contribute to centralized data storage and analysis from different activities and perspectives, and therefore, improve the development assessment task. This would facilitate early interventions preventing children from more severe difficulties in manual activities, including handwriting. Moreover, this instrument seems promising for more precise diagnoses of eventual motor difficulties not only in children with atypical development, but in adults with Parkinson disease as well [11-15,34].