Analysis of Adolescents’ Head to Shoulder Region during Tablet Use from Sagittal and Frontal RGB Images

Abstract

As schools go digital, the use of tablet computers is increasing. Concerns are raised that the extensive use of tablets and the associated bent-over posture may negatively affect the individual’s health. In order to analyse the possible effects of prolonged tablet use on physical health, a detailed analysis of the posture during tablet use is needed so that appropriate preventive measures can be taken to prevent degenerative changes. Therefore, the aim of this study was to measure and report the posture of 56 students while working with a tablet computer and compare it with an upright posture. Sagittal and frontal images were used for measurements of the subjects’ postures while seated, using the tablet, and in a neutral sitting position looking straight ahead. The body position during tablet use was recorded in two different user configurations: tablet flat on the table and tablet in individual freely chosen user configuration. After appropriate annotation of the data, the following parameters were evaluated in different planes. The craniovertebral angle (CVA), head tilt angle (HTA), and forward shoulder angle (FSA) are measurements that describe the extent to which the head bends forward and downward and how the shoulders are aligned in the sagittal plane. On the other hand, the head shoulder angle (HSA), lateral head tilt angle (LHTA), and trunk flexion angle (TFA) are angles measured in the frontal plane, which indicate the degree of head tilt and trunk bending to the right or left side. The measurement results clearly showed that the use of a tablet had a pronounced effect on the positions and rotations of the participants’ head, neck, and shoulders. This was evident through strong deviations observed in the angles measured between the sitting straight posture and the postures while using the tablet. For example, depending on the body posture class, the mean CVA values were 45.76° for straight sitting posture, 28.25° for holding the tablet individually posture, and 26.04° for the posture adopted while using a tablet placed flat on the table.

1. Introduction

In today’s world, smart devices have entered all fields of life and are used by all age groups. At 3 to 5 years old, the average daily use of smartphones and tablets is around 115 min per day [1]. Even though most parents are concerned that their children might spend too much time on screens [2], the active duration of daily smartphone and tablet use increases as children continue to grow. This is underlined by Mullan’s research [3], which showed that in the UK and the US, around 80% of young people aged 8–18 have access to a tablet. As the digitization of school education continues, it is expected that the use of tables will increase. Parents’ concerns about the health implications of extensive use of smart devices have been the subject of several studies and are well-founded. In addition to possible negative effects on cognitive and social development and mental health [4,5,6,7], constant use of smart devices could also have a negative impact on children’s physical health.

Numerous studies have explored the relationship between prolonged poor posture and the occurrence of back and neck pain associated with the use of smartphones and other portable devices. Lu et al. [8] indicated muscle fatigue after poor shoulder and cervical posture and speculates on its effects on possible pathologies. One objective of the study of Azevedo et al. [9] was to investigate whether postural angles and stable static balance parameters are related to the manifestation of back pain. Through their analysis, they found a statistically significance between the postural angle of the thoracic spine and the manifestation of back pain in children and adolescents. This risk increases as the angle of thoracic kyphosis also increases, although it is relatively small. However, due to the invalid cervical spine measurement of the tool used, they were unable to analyse a possible association between cervical postural angles and a manifestation of back pain. Straker et al. [10] compared the posture and muscle activity of 18 children aged 5–6 years using a tablet computer, a desktop computer and paper technology. They found that tablet computer use resulted in significantly different musculoskeletal strains than desktop computer use. Tablet computer use was associated with less neutral spinal posture, increased shoulder blades elevation, and greater upper trapezius and cervical erector spinae activity.

Promoting awareness of posture from an early age is crucial in preventing the development of poor postures in adulthood [11]. Studies have been conducted to investigate the sitting postures of school students in order to understand their effects. In this manner, Domljan et al. [12] identified main working postures to define them as notable criteria when designing school furniture for the future. Therefore they analyzed the upper, lower, and whole body movements and occasional movements and postures of 18 students from the second to eighth grade. The results showed large differences in the subjects’ behavior and their habits of using the task chair and table in terms of age, gender, time of day, subjects studied, tasks, and teachers’ behavior. The relationship between the sitting posture of school children on school furniture and changes in the spine was analyzed by Araújo et al. [13]. The main findings of this study indicated gender-specific correlations between the posture adopted by adolescents aged 10–16 years and the presence of postural changes such as head forward, thoracic kyphosis and pelvic tilt. Concerning school furniture and posture, Goncalves et al. [14] found that the inclined surfaces of the table towards the children and the seat forward are alternatives to be considered in order to eliminate tensions and improve children’s well-being.

It is important to note that existent studies that report a causal relationship between poor posture and musculoskeletal disorders lack clear evidence. Swain et al.’s umbrella review [15] examined the association between specific physical exposures and low back pain, as well as the level of causality demonstrated by previous systematic reviews. Their findings indicated a lack of consensus on the causality of physical exposure in relation to low back pain, and while an association was observed, it did not provide a definitive causal explanation for low back pain. Due to the absence of conclusive evidence regarding causality, further studies are needed to understand this issue better and enhance individuals’ quality of life through education on poor postural habits.

Gaining basic knowledge about the current posture of children using tablets in the school environment will help to guide recommendations on how long children should spend in screen-based activities [16] and how to work with the tablet in an ergonomic position and how to prevent musculoskeletal disorders caused by prolonged forward displacement of the neck and head [17]. Specific muscular disorders include the upper cross syndrome, that is among others associated with nerve pain due to poor posture of the cervical spine, muscular imbalances and increased strain on the intervertebral discs, shoulder and back pain [18,19,20], that has to be avoided.

To quantify a human’s postures in different scenarios, various methods are available, such as a visual observation method, a plumb-line method, goniometry, photographic and a radiographic methods [21]. However, to the best of our knowledge, the existing studies utilized the observations for posture measurements from a sagittal perspective only, and not from a combination of sagittal and frontal images. Human movement involve multidimensional motion across multiple planes. When the head and neck are tilted forward while using a tablet, it is important to recognize that the movement is not solely limited to a tilt around a single body axis. In addition to the tilt observed in one axis, there may also be a concurrent tilt occurring around another body axis. In particular, the combination of rotation and flexion movements is suspected of causing degenerative changes in the spinal structures [22,23].

Therefore, the purpose of our study was to provide a fundamental description and comparison of the recorded postures adapted by students of an upper-intermediate secondary school in upright sitting position and during tablet use by applying by measuring the recorded postures in photographs from two different perspectives: sagittal and frontal. The findings presented in this study can serve as a foundation for fellow researchers interested in exploring postural changes in students’ upper bodies when using tablets. The lateral and frontal posture characteristics we reported can be utilized to construct biomechanical simulation models, offering more profound insights into the posture’s impact on tablet users’ spinal musculoskeletal system. Furthermore, the results of our study contribute to raising posture awareness among students while working with tablets. Educators can use our findings to promote the health and well-being of young individuals by creating guidelines for ergonomic tablet use in educational settings.

2. Materials and Methods

As our goal was to accurately assess and report on various aspects of children’s posture while working with the tablet, data collection includes quantitative measurements. To evaluate the posture of student’s head, neck, and shoulders while using a tablet, we recorded the volunteers with conventional cameras strategically positioned to capture images from different viewpoints. Subsequently, the taken images were annotated by multiple experts who were trained to identify and mark pre-defined landmarks. To determine the relevant posture measurements, custom scripts were employed to automatically analyze the annotated images.

To describe the collected data, various descriptive statistics were calculated to provide a summary of the posture characteristics of the participants. Measurements such as means, standard deviations, medians, percentages, and correlations were used in this analysis. Furthermore, correlations were examined to determine the relationships between different measurements among captured postures, in order to provide insights into the strength and direction of associations between the respective body parts that characterize the given posture. The data collected and reported in this study establishes a baseline for further investigations in this area.

2.1. Participants

For this study, participants were recruited at a German Gymnasium (upper-intermediate secondary school) through an invitation letter that also included a consent to participate section. Fifty-six students between the age 10 and 19 have participated at the study, and students with scoliosis or cervical implants were excluded from the study. The demographic characteristics of the participants are documented in

Table 1. Participant demographic characteristics (n = 56).

Variable Name	Number/Average	SD	Max	Min
Male participants	31	–	–	–
Female participants	25	–	–	–
Age (years)	12.13	2.10	19	10
Height (cm)	163	10.0	190	143
Weight (kg)	70.6	13.1	92	29

.

2.2. Data Collection Procedure

To analyse the subjects’ posture while using the tablet, sagittal and frontal views were taken using RGB cameras. Following postures were recorded for different tablet holding configurations:

• tablet placed flat on the table (posture class = “tablet flat”);
• holding tablet in an individually freely chosen user configuration (posture class = “tablet individually”);
• neutral straight sitting posture without using a tablet (posture class = “straight sitting”).

During the recording of the straight sitting position, the subjects were asked to look straight ahead at a red dot with a diameter of approx. 6 cm placed on the opposite wall. This instruction helped the students to maintain the sitting position in a neutral posture with an upright head. In this posture, tablet was not in use.

All observations were taken while the study participants were sitting at the table. The height of the table was 75 cm and the chairs comply with DIN EN 1729-1:2016-0 (note: this European Standard specifies functional dimensions and markings for all chairs, stools and tables for educational institutions, including non-adjustable and adjustable chairs and tables.). The height of the chair seat was 45 cm and was not adjusted to the height of the students. The seat and backrest were ergonomically shaped and made of multilayer glued beech wood. The angle between the backrest and the vertical was 15°. The backrest supports the parts of the thoracic spine. When necessary, the subjects’ hair was tied back to ensure visibility of the neck region. In addition, to capture posture under realistic conditions, the subjects were not asked to sit still during the recordings but could write on the tablet.

For the dataset collection we used Astra Orbbec Pro 3D cameras, capturing 2D RGB and depth images from the frontal view and the other the sagittal views. Please note, depth images were not used in the current study and are planed to be applied in the further work. The frontal camera was positioned to capture the entire tabletop and the upperbody with the head while sitting upright. The lateral camera was positioned to capture the arms, shoulders, and entire head in the sagittal plane. The distance between the participants and the cameras placed on a tripod was fixed at 1.5 m for the sagittal camera and 1.2 m for the frontal camera. The resolution of the lateral images taken was 1280 × 800 pixels.

Three imaging rounds were performed on each subject to eliminate potential data biases. In each round, 60 sagittal and 60 frontal images were taken for each posture class within 2 s. For the three rounds of recording, there were a total of 1440 images for each subject.

2.3. Data Analysis

2.3.1. Landmarks Annotation

In order to quantify and analyze postures in a consistent and accurate manner, we asked four independent annotators to accurately identify and draw specific key points on the test person’s body in each selected image. These points served as reference coordinates for subsequent posture angle measurements and analysis.

The open-source Computer Vision Annotation Tool (CVAT) (https://github.com/opencv/cvat (accessed on 28 July 2023) was used to annotate the required anatomical landmarks. CVAT is a web-based tool designed to facilitate the annotation and labeling of images and videos for computer vision tasks. For our purposes the annotators annotated the images by drawing the pre-defined frontal and lateral landmarks within visual data. Four landmarks, acromion, C7 spinal apophysis, ear tragus and eye canthus were set in each sagittal image (see Figure 1a). The six landmarks cantus right, cantus left, nose tip, chin point, acromion right and acromion left were set in each frontal image (see Figure 1b).

In order to evaluate each posture, a total of six images per subject were randomly chosen for analysis from the entire set of photographs. To ensure consistent analysis, each image was assigned a timestamp during storage, ensuring that only the sagittal and frontal images taken simultaneously were utilized. To accurately assess the selected images, four independent experts annotated predefined landmarks present in the images. This allowed for consistent identification of key points necessary for posture evaluation. To determine the reliability of the assigned landmarks’ values among the experts, the study calculated a Krippendorff alpha [24] coefficient, resulting in a value of $0.81$. This coefficient indicates a high level of agreement among the experts regarding the landmarks values assigned, demonstrating strong interrater reliability.

2.3.2. Measured Angles

Three well-established angles for sagittal posture analysis were used to assess the head and shoulder posture of each individual: craniovertebral angle (CVA), head tilt angle (HTA) and forward shoulder angle (FSA) [25,26,27,28,29] (see Figure 2a).

To analyze the frontal posture the three following angles were measured: head shoulder angle (HSA) (see Figure 2b adapted from [30]), lateral head tilt angle (LHTA) (see Figure 3) and trunk flexion angle (TFA) (see Figure 4) were determined.

For each half of the body, the HSA is determined to analyse the posture of the head in relation to the position of the shoulders. The HSAs are defined by two half-straight lines starting from a common point, the midpoint between the two annotated acromium points. The first line is defined by the acromium point and the midpoint between the annotated acromium points. The second line extends from the midpoint between the annotated acromium point and the corresponding cantus point. The LHTA distinguishes between left and right head tilt. The angle is defined by the perpendicular line passing through the center of the tip of the nose and the straight line passing through the center of the tip of the nose and the tip of the chin. A negative angle ${\mathsf{\gamma }}_L$ describes a head tilt to the left, where a positive angle ${\mathsf{\gamma }}_R$ implies the head tilting to the right side.

2.3.3. Angle Calculation

The corresponding angles were automatically measured by a custom script using the 2D pixel positions of the annotated landmarks in each selected image. The postural angles were calculated using the trigonometric properties of the triangles spanned by the landmark points and the corresponding reference horizontals (for example in CVA, HTA or FSA) or vertical (as for LHTA) lines.

In order to analyse the consistency and reliability of the angle measurements on the key points that are annotated by different experts, we determined Root Sum of Squares Error (RSSE) for all lateral and frontal angles per posture:

$$\mathrm{RSSE}=\sqrt{\frac{1}{r}\sum _{j=1}^r{(x_j-\widehat{x})}^2},$$

(1)

where r is a number of raters, $x_j$ is the average measurement of the specific angle in all test persons for the j-th rater, and $\widehat{x}$ is the mean measurement over all raters for the respective angle. The calculated lateral ${\mathrm{RSSE}}_{lat}$ and frontal ${\mathrm{RSSE}}_{front}$ values for the corresponding postures are depicted in

Table 2. Root Sum of Squares Errors (RSSE) [degree] for all posture classes.

Raters Error	Straight Sitting	Tablet Individually	Tablet Flat
${\mathrm{RSSE}}_{lat}$	1.82°	1.83°	2.01°
${\mathrm{RSSE}}_{front}$	0.39°	0.73°	0.94°

.

The error analysis reveals that the average deviations in annotations by different raters were highest in the lateral images, particularly for the student postures when working with the tablet placed flat on the table. In this specific case, the lateral angle deviations, measured as ${\mathrm{RSSE}}_{lat}$, reached 2.01°. On the other hand, the agreement among experts was higher in the frontal images, as indicated by smaller ${\mathrm{RSSE}}_{front}$ values, with the posture class “tablet flat” exhibiting a deviation of up to 0.94°.

3. Results

The results of measured angles in the lateral and frontal photographs are depicted in

Table 3. Angle measurements [degree] of recorded posture classes from latterla view.

Posture Class	Mean	SD	Max	Min	95% CI
Craniovertebral angle (CVA) measurements
straight sitting	45.76	8.00	61.92	22.41	45.15–46.38
tablet individually	28.25	15.39	66.66	−7.41	27.07–29.42
tablet flat	26.04	15.73	66.11	−13.84	24.82–27.26
Head tilt angle (HTA) angle measurements
straight sitting	21.84	7.63	37.79	2.46	21.25–22.42
tablet individually	−8.17	12.21	19.03	−36.64	−9.11–7.24
tablet flat	−11.76	14.08	21.71	−46.39	−13.02–10.85
Forward shoulder angle (FSA) measurements
straight sitting	118.71	27.49	175.28	13.32	116.61–120.82
tablet individually	103.58	30.92	167.30	24.77	101.21–105.95
tablet flat	100.14	31.28	166.06	28.77	97.71–102.56

Note: Maximum und minimum angle values are the limits of non-outlier range calculated using 1.5× IQR rule. 95% CI denotes 95% of the confidence interval.

,

Table 4. Correlations (r) of craniovertebral angles (CVA), head tilt angles (HTA) and forward shoulder angles (FSA) for measured postures.

	Straight Sitting			Tablet Individually			Tablet Flat
Variable	CVA	HTA	FSA	CVA	HTA	FSA	CVA	HTA	FSA
CVA	1.00	−0.24	0.00	1.00	0.52	0.20	1.00	0.62	0.19
HTA	−0.24	1.00	0.23	0.52	1.00	0.23	0.62	1.00	0.16
FSA	0.00	0.23	1.00	0.20	0.23	1.00	0.19	0.16	1.00

and

Table 5. Angle measurements [degree] of recorded posture classes from frontal view.

Posture Class	Mean	SD	Max	Min	95% CI
Left head shoulder angle (HSAl) measurements
straight sitting	69.11	7.05	88.03	46.51	68.61–69.68
tablet individually	45.69	24.95	100.28	−5.17	43.82–47.65
tablet flat	40.43	30.75	103.67	−36.96	38.30–43.01
Right head shoulder angle (HSAr) measurements
straight sitting	73.81	6.86	89.94	55.25	73.25–74.29
tablet individually	60.59	44.90	112.98	22.63	57.10–64.01
tablet flat	55.46	41.36	112.14	16.06	52.84–59.07
Left trunk flexion angle (TFAl) measurements
straight sitting	2.52	1.66	6.55	0.06	2.35–2.70
tablet individually	3.71	3.27	9.59	0.09	3.38–4.03
tablet flat	4.13	3.49	10.46	0.04	3.81–4.46
Right trunk flexion angle (TFAr) measurements
straight sitting	2.33	1.72	7.05	0.08	2.14–2.52
tablet individually	2.53	2.07	8.53	0.04	2.27–2.79
tablet flat	2.89	2.19	8.42	0.07	2.60–3.20
Lateral head tilt angle (LHTA) measurements
straight sitting	−1.54	5.11	11.06	−15.43	−1.93–1.15
tablet individually	−11.21	21.75	29.72	−48.16	−12.88–9.53
tablet flat	−9.93	21.40	35.95	−56.72	−11.58–8.29

Note: Maximum und minimum angle values are the limits of non-outlier range calculated using 1.5× IQR rule. 95 % CI denotes 95% of the confidence interval.

. In this section, we first describe our finding on the angles measured in the sagittal plane. Subsequently, the measurements of the posture angles in the frontal plane are reported. The maximum and minimum values that we report are the values of the non-outlier range, which was determined using Tukey’s fences method [31] for detecting the outliers. In our case, we define the non-outlier limits to be $1.5\times$ of interquartile range (IQR).

3.1. General Characteristics of Lateral Posture Angles

The results of the craniovertebral angle (CVA) measurements show that the mean value when using the tablet was notably different from the mean value when sitting upright (see Table 3). However, there was a much smaller difference of about 2.2° between postures when the tablet was held individually or placed flat on the table during tablet use.

Upon analyzing the interquartile range (IQR) of CVA values across different recorded posture classes, we found that for the straight sitting posture, approximately 50% of observations fell within the range of 40.72° to 51.66°. In the case of using a tablet, the IQR of CVA measurements varied depending on the posture. When the tablet was placed flat on the table, the range was found to be between 16.12° and 36.13°. However, when the tablet was used individually, the range extended from 20.34° to 38.89°.

According to Brunton et al. [32], smaller CVA values indicate a greater degree of head flexion, while larger angles are more closely associated with the “ideal” sagittal head-neck alignment. The cervical spine with a CVA of less than 20° defined as having a high degree of flexion. To investigate the distribution of the small CVA values in the recorded population, we analyzed the observations of CVA values less than 20° (see Figure 5) among the defined posture classes. Among students adopting tablet usage postures, specifically the “tablet individually” and “tablet flat” postures, over one-fifth of all observations indicated a CVA measurement smaller than 20°, where 0.1% of the observations adopted CVA smaller than 20° in the straight sitting posture.

During the examination of head tilting angles (HTA) as shown in Table 3, the average HTA value was found to be 21.84° for the “straight sitting” class. However, for the “tablet individually” and “tablet flat” posture classes, the average angles deviated into the negative range, with −8.17° and −11.74° respectively. Strong differences were detected in the maximum and minimum head tilt measurements between straight sitting posture and postures that involve tablet usage, with a particularly noticeable effect in postures that involve tablet usage.

In our analysis of forward shoulder angles (see Table 3) across all three posture classes, we observed that the average and extreme values were relatively similar. However, the standard deviation of these angles was the highest among all the reported measurements, averaging around 30°.

3.2. Correlation of Lateral Posture Angles

By computing the correlations between the defined posture angles, we aimed to analyse the potential functional relationship between different body regions. The correlations between individual variables were analyzed between measured lateral angles for all recorded posters (see Table 4). The correlation between lateral angles were calculated using Spearman correlation coefficient r [33] to describe the strength of the relation.

It can be seen in Table 4, craniovertebral and head tilt angle measurements indicated a strong positive correlation of $r=0.62$ while using a tablet placed flat on the table, and $r=0.52$ when using the tablet in freely chosen holding configuration. On the contrary, a negative value association with $r=-0.24$ was detected by the correlation test for the HTA and CVA angles in the “straight sitting” posture. A weak correlation of $r=0.2$ was found between forward shoulder and head tilt angles for the “tablet individually” posture class.

3.3. General Characteristics of Frontal Posture Angles

Three angles, head shoulder angle (HSA), trunk flexion angle (TFA) and lateral head tilt angle (LHTA), were selected to analyze the posture in the frontal plane. Mean, standard deviation, minimum, and maximum values of the measured angles are reported in Table 5.

By examining the measurements related to the lateral flexion of the head towards the shoulders (HSA), we can observe distinct patterns in the recorded data. In the neutral sitting posture, the average values for the left (HSA_l) and right (HSA_r) head shoulder angles were only slightly different. However, significantly larger average values were noted for measurements taken in postures involving tablet use. Notably, the standard deviation (SD) values for these angles in the posture classes recorded during tablet use were particularly high. The highest SD value of 44.90° was observed for the head flexed towards the right shoulder when the students were working with the tablet placed individually on the table.

While investigating the bending of the body trunk towards either the left or right side (denoted as TFA), we observed relatively minor average lateral movements of the trunk across all recorded postures. However, notable differences were observed in the maximum rotations, particularly in left trunk bending in the posture classes involving tablet usage.

Out of 56 students observed, 53% maintained left trunk flexion, whereas 47% bent the body to the right side during the straight sitting posture. While holding the table individually, the portion of the volunteers that bent their body to the left side increased up to 61% and up to 68% while using a tablet placed flat on the table. The lateral trunk flexion was maintained 39% of all students while holding the tablet device in a personally preferred manner and 32% of the participants while using the tablet flat on the table.

During the investigation of the head rotations (denoted as LHTA) in the frontal plane (see Figure 3), we found out that the students on average rotated their head to the left side while working with the tablet. This measurements were reflected in the clear negative average angles for “tablet individually” and for “tablet flat” postures.

We measured the proportion of the students that tilted their heads to the left and the right side for each observed posture. 70% of all students maintained left head tilt in the straight sitting posture, whereas 30% bent their head to the right. Using the tablet individually, 79% of the involved participants moved their head to the left and 21% to the right shoulder correspondingly. In the posture where the tablet was placed flat on the table 78% of the students tilted their head to the left, and 22% adopted the right head tilt.

3.4. Correlation of Lateral and Frontal Posture Angles

The results of the correlation between the lateral and the frontal angles are presented in

Table 6. Correlations (r) of lateral (craniovertebral angle (CVA), head tilt angle (HTA), forward shoulder angle (FSA)) and frontal (head shoulder angle (HSA), trunk flexion angle (TFA), lateral head tilt angle (LHTA)) angles for measured postures.

	Straight Sitting			Tablet Individually			Tablet Flat
Variable	CVA	HTA	FSA	CVA	HTA	FSA	CVA	HTA	FSA
HSAl	0.26	−0.07	−0.25	0.49	0.64	0.15	0.60	0.73	0.14
HSAr	0.15	−0.04	0.13	0.27	−0.008	−0.12	0.41	0.26	0.02
TFAl	0.15	−0.13	−0.20	−0.20	−0.35	−0.2	−0.15	−0.12	−0.25
TFAr	−0.03	0.002	−0.06	−0.07	0.12	−0.06	0.05	0.25	−0.10
LHTA	0.14	−0.02	−0.25	0.17	0.41	0.08	0.17	0.35	−0.05

. The frontal angle HSA that measure the head bending towards the shoulders showed strong positive correlation with the lateral craniovertebral and head tilt angles in the “tablet individually” and “tablet flat” postures. The measurements of the left trunk flexion angle (TFA_l) indicated weak relationship to each of the lateral angles in all posture classes. The lateral head tilt angle (LHTA) taken in frontal view corresponded with head tilt angle (HTA) observed in lateral images in the postures involving tablet use.

4. Discussion

In order to gain a basic knowledge of the posture situation during tablet use, the current study reports and analyzes of students’ posture while using a tablet in two different user configurations and the “straight sitting” posture by taking RGB images of the sagittal and frontal planes.

4.1. Craniovertebral Angle (CVA) Analysis

The results of the lateral posture angle CVA (see Table 3) showed that the maximum and minimum angle values strongly differentiate in the posture classes with tablet use. While the maximum angle values showed positive values of over 66°, the minimum values indicated native angle values of less than −20°. A negative CVA means that the measurement point of the tragus is below the horizontal line passing through the measurement point of the 7th cervical vertebra. Therefore, it can be concluded that the head of the students with a negative CVA is extremely tilted forward and positioned very forward in relation to the neck and shoulders.

Comparing the average value of CVA in upright sitting from the current study with those from previous studies, it is in the same range. Watson et al. [34] reported a range between 44.3° and 50.6° for asymptomatic subjects. In the current study, the average CVA of 45.76° (see Table 3) of the “straight sitting” posture is within this range.

Small CV angles were found in this study for the postures “tablet individual” (28.25°) and “tablet flat” (26.04°). Comparing this angles with the angles when using computer, we saw that the presented CV angles were obviously much smaller. Kang’s study [35] showed an average CV angle of 48.9° when working on a computer. Much larger neck flexion angle (NFA) of 46.3°, which corresponds to a CV angle of 43.7°, was recorded by Ning [36] during working with a mobile device placed flat on a table that was adjusted to user’s elbow height. From these different angle values it can be concluded that the students recruited in our study adopted a poor posture when using the tablets. This statement is supported by Choi’s classification [37] of the CV angle into “good” posture with a CVA > 48.7°, “fair” posture with a CVA between 43.8° and 48.7° and “bad” posture < 43.8°.

Through the small CV angles the head is shifted further forward, which is associated with a larger gravitational moment as the head´s center of mass moves away from the corresponding center of rotation, in this case located in the nearby the intervertebral discs [10]. The intervertebral discs potentially receives a higher torque, which must be compensated by the back muscles to keep the head in the “forward” position [38,39].

In the context of our scenario, this indicates that students who adopt postures with small lateral angles are particularly susceptible to placing excessive strain on their spinal structures, which is underpinned by the findings of Drake et al. [40], Barbir et al. [41], Chan et al. [42] and Farfan et al. [43].

4.2. Head Tilt Angle (HTA) Analysis

While small and negative CVA may suggest a forward head posture, it is not the only factor to consider. The reported results showed that the values of the HTA measurements (see Table 3) were also smaller in the body positions when using a tablet than the corresponding values in the posture “straight sitting”. For the HTA mean posture classes “straight sitting”, “tablet individually” and “tablet flat” the angle values were 21.84°, −11.76° and −8.17°. Chen [44] found that participants using a smartphone maintained a head flexion angle (HFA) of 99.9° during standing, 93.5° during unsupported sitting, and 95.1° during supported sitting, which correspond with HTA of 9.9°, 3.5°, and 5.1°, respectively. The poor head posture during tablet use becomes even more apparent when comparing HTA during tablet use with HTA during computer use. Participants in Kang’s study [35], who used computers for more than 6 h per day had an HTA of 20.6° and participants who rarely used computers had an HTA of 18.9°, which indicates a more upright posture. In other words, the posture when using the tablet is much more flexed and therefore poor.

The above mentioned statement is further supported by a correlation analysis which resulted in a strong positive correlation between CVA and HTA values in the posture class “tablet flat” ($r=0.62$) and posture class “tablet individually” ($r=0.52$). The presented correlations indicate a functional connection of the head with the cervical spine, which implies that the flexion of the head or the neck only cannot be considered in isolation when analyzing posture. A similar correlation of CVA and HTA measurements was also reported in the research done by Szczygieł et al. [45].

4.3. Forward Shoulder Angle (FSA) Analysis

Although the values reported in this study for the FSA (see Table 3) in the “straight sitting” position were in the same range as the angles published by Ormos et al. [11], the FSA values reported in our measurements have shown much more variability with $SD=27.49$ comparing the standard deviation of $SD=2.74$ reported by Ormos. While assessing the shoulder positions for the calculation of angular characteristics, we found out that in some cases, the values of the FSA angles were affected by the elevated and slightly prolonged shoulders in the body positions where the forearms were placed on the table (see Figure 6b). Suppose the table is too high, as, in our example, the shoulders were raised and involuntarily pushed upwards due to the height of the table.

It can be concluded, that posture also depends on the standardised height of the school table. Therefore, the effects of different school table heights on posture should be investigated in the future. In our future work, we might also investigate the association between the participants’ body height and postural angles that involve shoulder position.

4.4. Analysis of Frontal Angles

By analyzing the angle measurements from frontal images, we found that the angle information from the 2D images was not always accurate enough to describe the actual body position in 3D space. It turned out that in some observations students twisted their bodies in ways that made it impossible to make statements about defined angles. Figure 6a shows that although the necessary annotation points could be determined for calculating the LHTA, the angle value of approx. 70° does not give a clear statement about the head tilt due to the strong twisting of the body and the associated shifts of the annotation points within the three spatial planes. However, while the magnitude of the LHTA angles (see Table 3) could not be used in the comprehensive posture analysis, we found very helpful the “sign” (positive or negative) of the determined angle values, which indicates the direction of the head twist.

Therefore, in a follow-up project, we would like to carry out a 3D reconstruction of the depth images into a solid surface upper body model. To do this, we will use the additional depth images taken of each child by the camera’s depth sensor. With the help of 3D reconstructions, it should be possible to accurately determine the tilt angles in complex torsions of the upper body.

In spite of this limitation, the presented posture measurement results emphasize that a posture is mostly a combination of rotations around different axes. Only by examining frontal and lateral views together, it becomes more apparent that postural adjustments involve complex movements in multiple directions. It is precisely this combination of simultaneous rotations about different body axes that is the focus of attention as the cause of acceleration of a herniated disc. This statement is supported by the findings of Drake et al. [40] and Schmidt et al. [46]. They reported that axial rotation in combination with compression and flexion does not increase the proportion of vulnerable discs, but facilitates the herniation process in those discs that are at risk.

The results of the frontal posture measurements also indicate that the body posture deviates from an upright posture while using the tablet. Furthermore, we observed a strong relationship between lateral angles, which include head and neck flexion, described by CVA and HTA, and frontal angles describing the posture of head to shoulder region during tablet use. When students tilt their necks and heads, they accompany this movement with a rotation of the head to the side. In these cases, the spine’s alignment is likely strongly affected. This abnormal spinal alignment was also emphasized by Straker et al. [10], who noted that the larger spinal asymmetry associated with tablet use might also increase the risk of developing musculoskeletal symptoms.

4.5. Growing Awareness of Posture Considerations during Tablet Usage: Implications for Students and Researchers

We would also like to address that during development, children develop postural habits that they tend to maintain for the rest of their life [47]. Therefore, in order to prevent degenerative postural changes, particularly in the neck, shoulders, and upper back regions and to avoid the upper crossed syndrome [48], it is necessary to increase awareness and attention to good posture habits among tablet users. Only awareness of posture can help to change postural habits [49]. As a suggestion, e.g. a feedback system via the tablet that detects and analyzes body posture and provides appropriate feedback to the user could assist the user in inducing a change in consciousness. Moreover, it would be important to incorporate a change between almost static postures and active breaks, such as stretching or gymnastics, into a school lesson.

In addition, in the absence of evidence-based guidelines for the prevention of back pain in young people [50], the results of our study can be used to develop back education programs for young people and to modify school furniture and workplace design to promote healthy posture when using smart wearable devices.

Finally, it is important to emphasize that our goal was to capture and report on students’ postures in the real-life conditions present at school, rather than in a controlled laboratory setting. Some factors, such as adjusting the height of the table and chair to the subject’s height, were not taken into account, especially since the school has non-adjustable chairs and tables. Whether or how height affects posture should also be considered in future studies. In addition, follow-up studies may include pelvic evaluation to investigate clinical relevance, which was not done in the current study. Furthermore, in future work of the posture analysis, we aim to consider whether individuals have vision impairments in order to eliminate the possibility that they tilt their heads closer to the tablet due to visual difficulties.

As a last discussion point, we suggest the establishment of a comprehensive database encompassing postural measurements of individuals across various age groups. This database would serve the purpose of defining and constructing standard movement patterns, while also enabling the identification of indicative factors for inefficient or non-physiological utilization of the human body.

5. Conclusions

The presented study aimed to analyze upright and flexed postures during tablet use in the school by using sagittal and frontal photographs of the subjects in different body postures. The measurements results showed that the head, neck and shoulders positions adopted by the participants while using a tablet strongly affected all lateral and frontal angles. Based on the reported lateral angle values, we could see that the participants maintained a non-physiological posture while using a tablet, as the students’ heads were tilted extremely forward. This statement is reflected in the extremely small CVA values for the posture classes “tablet individually” (28.25°) and “tablet flat” (26.04°), compared to upright sitting where the average CVA was 45.7°, indicating a poor posture during tablet usage [37]. Further, the head tilt angle (HTA) measurements revealed that students’ head positions were more flexed during tablet use than in straight sitting. The forward shoulder angle (FSA) values showed more variability when using tablets compared to upright sitting, potentially influenced by the height of the school tables.

Analyzing frontal angles proved challenging, as students often twisted their bodies in ways that made it difficult to determine the angle from 2D images accurately. No valid statement about defined angles could be made in these specific cases. However, the direction of the head twist was still valuable information that could be used in the assessment of the posture during tablet use.

It’s important to note that while research in this field is advancing, the relationship between head and neck posture during the tablet use and possible spinal degeneration is still not fully investigated. By measuring the posture in lateral and frontal planes, we could see, the head and neck rotations were not limited to a single axis but occurred simultaneously around different axes. Further research is required to fully understand these processes and their impact on spinal health. To address this issue in the future, precise investigation of loads on the spine may be achieved through the use of biomechanical simulation models which can simulate the complex association between posture and spinal alignment. Therefore, our goal in the future is to extend a presented research by analysing the forces that act on the spine during the tablet use [51].