Design

A within-subjects, repeated measures design was used to analyse changes in postural kinematic parameters in seven upright birthing positions. The research was developed in collaboration with the public and stakeholders, ensuring it was conducted ‘with’ and ‘by’ them rather than merely ‘to,’ ‘about,’ or ‘for’ them. The stakeholders and the Patient and Public Involvement and Engagement (PPIE) group influenced the development of the study’s methodological approach and the choice of birthing positions to evaluate, focusing on those most widely used worldwide and the practicality of gathering data in a timely manner. Specifically, stakeholders contributed insights into the clinical relevance of positions commonly used in high-, middle-, and low-income countries, while the PPIE group supported protocol refinement through discussion sessions and lab-based mock-ups, offering feedback on comfort, acceptability, and feasibility. By adopting this cooperative strategy, the study was made relevant and reflective of the needs and viewpoints of the intended audience [25,26].

Participants

Healthy non-pregnant females (biological sex) of reproductive age (18–49 years) [27], who were free from any injury, pain, illness, or medical condition that would limit their ability to perform the specific positions were recruited through campus-based advertisements and social media posts. In accordance with previously published rules of thumb for pilot studies [28–30], a sample size of n = 15 participants was determined for this exploratory study.

Equipment

Data were collected synchronously using two motion capture systems: a 9-Oqus-camera marker-based motion capture system (Qualisys Medical AB, Sweden) and an 8-Micus-camera markerless motion capture system (Qualisys, Medical AB, Sweden) [31]. Data from both systems were collected simultaneously in Qualisys Track Manager v2021.2 (Qualisys, Medical AB, Sweden), ensuring the systems were synchronised and calibrated in space and time by applying the same wand calibration [32]. A target calibration residual error below 0.5 mm was set, and all calibrations achieved residuals under this threshold. Footage from the markerless system was recorded at 50 Hz and data from the marker-based system was recorded at 100 Hz, as per previous comparative studies [31–33]. Sampling frequencies vary due to the different capabilities of the two systems. Markerless systems typically use lower sampling frequencies due to the constraints of the frame rate of the RGB video cameras within the system [31]. Reducing the sampling frequency of the marker-based system to match the markerless system may affect the precision and accuracy of the marker-based data [34]. A higher frame rate in the markerless system would increase the computational load during data processing, impacting real-time application [31,34].

Data collection

Data was collected between 6^th July and 24^th October 2023. Participants attended a 45-minute data collection session in the University’s Motion Analysis Laboratory. Following participant consent, age, height, and weight were recorded. To maximise the quality of the data, all participants wore athletic style shorts and an adapted t-shirt which afforded access to the spine.

Data collection using the marker-based system requires the placement of passive retro-reflective markers using the calibrated anatomical system technique to afford tracking of segmental kinematics in 6 degrees of freedom [35]. Markers were attached using dermatological friendly double-sided adhesive tape, and were applied to the acromions, sternal notch, 7^th cervical vertebra (C7), anterior superior iliac spines (ASIS), posterior superior iliac spines (PSIS), greater trochanters, and the medial and lateral epicondyles. Clusters of four non-collinear markers were attached to the thighs and spinal clusters were applied level with the 7^th thoracic vertebra, 3^rd and 5^th lumbar vertebra [36,37]. Using this configuration of markers, the study effectively delineated anatomical segments including the Lower Thoracic (LT), Upper Lumbar (UL) and Lower Lumbar (LL) spinal segments, a thorax (defined using the anatomical locations of the acromions, sternal notch, 7^th cervical vertebra and ASIS), a pelvis (defined using the anatomical locations of the ASIS and the PSIS), and hips (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Marker set used for the marker-based motion capture system.

https://doi.org/10.1371/journal.pone.0322970.g001

The seven upright birthing positions investigated were:

1. Standing upright [Upright] (Fig 2A): Participants were instructed to stand up straight with feet shoulder-width apart. They were asked to keep their palms facing their legs and look straight ahead at a specific point indicated to them, maintaining a natural and relaxed posture without stiffening their shoulders or arms.
2. Leaning forward – support on the palms [Palms] (Fig 2B): Participants were instructed to stand beside the bed and lean forward, placing their palms on the bed at a width that felt comfortable to them, with their elbows extended. They were asked to keep their knees straight and their feet at a width that was comfortable for them.
3. Leaning forward – support on elbows [Elbows, extended knees] (Fig 2C): As above, participants were instructed to stand beside the bed and lean forward, supporting their weight on their elbows and forearms, which were placed on the bed. They were asked to cross their palms, forming a triangle. The feet were positioned as described previously, with the knees kept straight.
4. Leaning forward – support on elbows – knees bent [Elbows, bent knees] (Fig 2D): Participants were instructed to stand beside the bed and lean forward, supporting their weight on their elbows and forearms, which were placed on the bed. They were asked to cross their palms, forming a triangle. Their feet were placed at a width that was comfortable for them, and they were to bend their knees to an angle of 90–120 degrees, according to their own comfort and preference.
5. Deep squat [Squat] (Fig 2E): Participants were instructed to stand to the side of the bed. With their palms placed on the bed for support, but not holding onto the bed, participants were asked to perform a deep squat within a comfortable range.
6. Hands and knees [All-4s] (Fig 2F): Participants were instructed to climb onto the bed and assume an all-fours position, with the spine in a neutral alignment and their palms placed directly under their shoulders, ensuring proper support and balance. Additionally, they were asked to look at a specific point to keep their neck straight, ensuring that the C7 marker and the sternal notch marker were not obscured. This alignment was crucial for accurate data collection and to maintain the integrity of the marker placements.
7. Sitting on a birthing ball [B-Ball] (Fig 2G): Participants were instructed to sit on a birthing ball with their legs spread wide to ensure balance and comfort. They were asked to place their hands on their thighs in a position that did not obstruct any markers. Additionally, they were directed to focus their gaze on a specific point that was indicated to them. The size of the ball was chosen according to the participant’s height. To maintain stability, participants ensured their feet were fully planted on the floor.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Birthing positions: 2A. Upright, 2B. Palms, 2C. Elbows with knees extended, 2D Elbows with knees bent, 2E. Squat, 2F. All-4s, 2G. B-Ball.

https://doi.org/10.1371/journal.pone.0322970.g002

During protocol development, and in combination with PPIE consultation, a formula was developed to set the height of the bed according to each participant’s height. This provided a practical solution, enabling standardisation relative to each participant’s stature, whilst also ensuring comfort and consistency during data collection. Maintaining a consistent reference point relative to body height is crucial for obtaining reliable kinematic data. The following formula was used: (Participant’s height in inches) + 1 = height of physio bed in cm.

Participants maintained each of the seven birthing positions for six seconds and assumed each position three times. As each position is considered as a static hold exercise, participants were required to remain motionless, counteracting the effects of gravity throughout the duration of the task. Considering the characteristics of participants who were ‘healthy’ but not necessarily athletically trained, and the number of required repetitions per task, six seconds was deemed a suitable amount of time to ensure good data was collected without causing potential muscle fatigue amongst participants. To reduce bias and control for the potential influence of confounding factors, such as the effects of any extraneous variables that might affect the results on the outcome measures being studied, the order of the seven positions was randomised individually for each participant using the online Research Randomizer Tool (http://www.randomization.com). Participants were given a minimum of 30-second intervals between positions, with additional time provided upon request. After completing all seven birthing positions once, the sequence was repeated twice more, maintaining the same randomisation order for each cycle, thus ensuring consistency across the three rounds of data collection. The data collection session concluded once all three sequences were completed.

Data processing and analysis

The time domain of the measurements for both systems was time normalised to 101 data points [31]. As the marker-based system is considered the reference standard, the most stable two seconds of data were used for each trial, which was determined using manual event determination. Events for the markerless data were then synchronised with the events from the marker-based data.

Statistical analysis

All statistical analyses were performed using SPSS v28 (IBM Corp., NY, USA). Descriptive statistics were used to describe the characteristics of participants.

To investigate the differences between birthing positions, one-way Repeated Measures Analysis of Variance (ANOVA) tests were performed for each motion capture system separately. This was also undertaken to determine the sensitivity of the markerless system, i.e., whether statistically significant differences between birthing positions were the same. Mean joint angles (°), standard deviations (SD), main effects and effect sizes (partial eta squared, ηp²) were reported. Where a main effect was observed, Least Significant Difference post-hoc pairwise comparisons were performed. Pairwise comparisons, mean differences (MD), and 95% confidence intervals of the mean differences (CI) were reported.

As a further step, to examine the sensitivity of the markerless system, and determine the extent to which data from the markerless motion capture systems matched that from the marker-based system, linear regression models were used for each outcome measurement. As it is considered the reference standard, variables from the marker-based system were selected as dependent variables and those from the markerless system as the independent/ predictor variables. This approach was chosen to reflect our aim of assessing how well the markerless system could replicate the reference-standard outputs of the marker-based system. By defining the marker-based data as the dependent variable, we directly evaluated how closely the markerless system aligned with the reference standard. The R square (R²) was reported as a goodness-of-fit measure (0.0–1.0); greater R² values represent smaller differences between the two systems. Significance values (p value) were also presented for each linear regression model. Statistical significance was set at the p < 0.05 level throughout.

Ethical considerations and data protection

The study was approved by the University of Central Lancashire – Health Ethics Committee (HEALTH 01203). Data collection conformed to the Declaration of Helsinki [41], and all information collected was kept strictly confidential, with participant anonymity maintained throughout. Data were collected, managed, analysed, and stored in accordance with General Data Protection Regulations (GDPR). Volunteers provided written informed consent prior to participation. During the consent process, participants were briefed on data protection measures and their rights concerning their data.

Participant characteristics

Fifteen healthy non-pregnant participants were included in the study; their characteristics are described in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Participant characteristics.

https://doi.org/10.1371/journal.pone.0322970.t001

Differences between upright birthing positions: Markerless motion capture system

Hip.

At the hip, significant main effects were observed in the sagittal, coronal and transverse planes (p < 0.001; Table 2). In the sagittal plane, participants assumed the most amount of hip flexion in the “squat” position, and the most amount of hip extension in the “upright” position (Table 9). Apart from when comparing between “B-Ball” and “elbows, bent knees” (p = 0.973), all other positions produced significantly different hip flexion-extension angles compared to each other (p < 0.016; Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Mean (SD), main effects (Sig) and effect sizes (ηp²) for data from the markerless and marker-based motion capture systems.

https://doi.org/10.1371/journal.pone.0322970.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Markerless system variables with main effects for hip and pelvis positions in the sagittal plane.

https://doi.org/10.1371/journal.pone.0322970.t003

In the coronal plane, the “squat” position produced the greatest amount of hip abduction, while the “upright” position produced the least. With the exception of the comparisons between “all-4s” and “elbows, extended knees” (p = 0.777), “all-4s” and “palms” (p = 0.254), and “B-Ball” and “elbows, bent knees” (p = 0.231; Table 4), all other positions were significantly different from each other (p < 0.006).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Markerless system variables with main effects for hip positions in the coronal and transverse planes.

https://doi.org/10.1371/journal.pone.0322970.t004

In the transverse plane, the “squat” position produced the greatest external rotation, while “elbows, extended knees” produced the greatest internal rotation. Non-significant differences were observed when comparing “upright” to “palms” (p = 0.059), “palms” to “elbows, extended knees” (p = 0.074), and “B-Ball” to “all-4s” (p = 0.586; Table 4). All other positions were significantly different from each other (p < 0.018).

Trunk.

For the trunk, significant main effects were observed in the sagittal and coronal planes (p < 0.001; Table 2). In the sagittal plane, the “upright” position produced the greatest amount of trunk extension while the “squat” produced the most trunk flexion. All positions were significantly different from each other (p < 0.002), except when comparing “all-4s” and “palms” (p = 0.055), “all-4s” and “B-Ball” (p = 0.194), “elbows, bent knees” and “elbows, extended knees” (p = 0.088), “squat” and “elbows, extended knees” (p = 0.760), and “squat” and “elbows, bent knees” (p = 0.317; Table 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Markerless system variables with main effects for trunk position in the sagittal and coronal planes.

https://doi.org/10.1371/journal.pone.0322970.t005

In the coronal plane, “all-4s” produced the least side flexion whilst “elbows, extended knees” produced the most side flexion. Except when comparing between “upright” and “B-Ball” (p = 0.069), “upright” and “all-4s” (p = 0.802), “all-4s” and “squat” (p = 0.121), and “all-4s” and “B-Ball” (p = 0.646; Table 5), all other positions were significantly different from each other (p < 0.049).

Differences between upright birthing positions: Marker-based motion capture system

Hip.

At the hip, significant main effects were observed in the sagittal, coronal and transverse planes (p < 0.001; Table 2). In the sagittal plane, the squat position produced the most hip flexion and the upright position produced the most hip extension. Except when comparing between “all-4s” and “elbows, extended knees” (p = 0.207), and “all-4s” and “B-Ball” (p = 0.173; Table 6), all other positions were significantly different from each other (p < 0.007).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Marker-based system variables with main effects for hip positions in the sagittal and coronal planes.

https://doi.org/10.1371/journal.pone.0322970.t006

In the coronal plane, the “squat” position produced the greatest amount of hip abduction, and the “upright” position produced the least amount. All positions were significantly different from each other (p < 0.041), except when comparing between “all-4s” and “palms” (p = 0.505), “all-4s” and “elbows, extended knees” (p = 0.505), “B-Ball” and “elbows, bent knees” (p = 0.225), and “palms” and “elbows, extended knees” (p = 0.093; Table 6).

In the transverse plane the “squat” position produced the most external rotation and the “upright” position produced the most internal rotation. Significant differences were observed when comparing the “upright” position with “palms” (p = 0.018), “elbows, extended knees” (p = 0.014), “squat” (p < 0.001) and “B-Ball” (p = 0.012; Table 7). “Palms” was also significantly different from “elbows, bent knees” (p = 0.010) and “squat” (p = 0.002). “Elbows, extended knees” was significantly different from “elbows, bent knees” (p = 0.002) and “squat” (p = 0.004). “Elbows, bent knees” was significantly different from the “squat” (p < 0.001) and “B-Ball” (p = 0.026). The “squat” position was significantly different from “B-Ball” (p = 0.005) and “all-4s” (p < 0.001).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 7. Marker-based system variables with main effects for pelvic and hip positions in the sagittal and transverse planes.

https://doi.org/10.1371/journal.pone.0322970.t007

Trunk.

For the trunk, significant main effects were observed in the sagittal plane (p = 0.002; Table 2). The “upright” position produced the most trunk extension and “elbows, bent knees” produced the most trunk flexion. The “upright” position was significantly different from all other positions (p < 0.005) except “all-4s” (p = 0.855; Table 8). “All-4s” was significantly different from “elbows, extended knees” (p < 0.001), “elbows, bent knees” (p = 0.005), “squat” (p = 0.006) and “B-Ball” (p < 0.001).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 8. Marker-based system variables with main effects for trunk and upper lumbar relative to lower lumbar positions in the sagittal plane.

https://doi.org/10.1371/journal.pone.0322970.t008

Lower thoracic segment relative to upper lumbar segment.

Significant main effects were observed in the sagittal plane (p < 0.001; Table 2). The “upright” position was significantly different from all other positions (p < 0.026) except “all-4s” (p = 0.893; Table 10). “Palms” was significantly different from “all-4s” (p = 0.010). “Elbows, extended knees” was significantly different from “elbows, bent knees”, “squat”, and “all-4s” (p < 0.019; Table 9). “Elbows, bent knees” was significantly different from “all-4s” (p = 0.012), and “B-Ball” was significantly different from “all-4s” (p = 0.005). The “upright” position produced the most extension and the “B-Ball” position produced the most flexion (Table 10).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 9. Marker-based system variables with main effects for lower thoracic relative to upper lumbar positioning in the sagittal plane.

https://doi.org/10.1371/journal.pone.0322970.t009

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 10. Comparative summary of hip, pelvis, and trunk biomechanics across upright birthing positions.

https://doi.org/10.1371/journal.pone.0322970.t010

Differences between motion capture systems

Hip.

The linear regression models revealed that the data from the markerless motion capture system matched the data from the marker-based system in only one out of seven positions for the hip joint variables in the sagittal plane. However, the markerless system performed better in the coronal plane, matching the data in six out of seven positions (Table 11). In the sagittal plane, the markerless system corresponded with the marker-based system only in the “all-4s” position (R² 0.575, p = 0.002). In the coronal plane, the markerless system matched the marker-based system in the “upright” (R² 0.799, p < 0.001), “palms” (R² 0.454, p < 0.008), “elbows, extended knees” (R² 0.639, p < 0.001), “elbows, bent knees” (R² 0.714, p < 0.001), “squat” (R² 0.653, p < 0.015), and “B-Ball” (R² 0.630, p < 0.001) positions, with a trend towards significance in “all-4s” (R² 0.270, p < 0.057). Importantly, in the transverse plane, no statistically significant relationships were found between the markerless and marker-based systems for any of the positions (all p > 0.05), suggesting limited alignment in this plane for hip joint angles.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 11. Predictive relationships between markerless and marker-based joint angle data across birthing positions.

https://doi.org/10.1371/journal.pone.0322970.t011

Markerless motion capture sensitivity

While the markerless system showed promise in providing kinematic data in certain positions, particularly in the coronal (hips and pelvis) and transverse (pelvis) planes, it did not consistently match the sensitivity of the marker-based system across all planes and positions. Strong correlations between the two systems were observed in some positions, indicating potential for markerless technology as a viable alternative in some clinical applications. However, significant discrepancies were noted in the sagittal plane, particularly for the hip joint, where the markerless system was less reliable. Previous validation studies [31,42,43] have generally reported higher sagittal-plane agreement in dynamic gait tasks. However, the lower sagittal correspondence in our study likely reflects the static nature of the postures, and the increased self-occlusion of key landmarks from body tissues, such as the abdomen or quadriceps, which can reduce depth perception and joint visibility in certain positions. They may also reflect the system’s inherent limitations in accurately recognising joint positions during complex postures, such as ‘all-4s’ and ‘elbows extended knees.’ Additionally, beyond potential issues with occlusion, limitations in algorithm accuracy [31,42,43] may further contribute to the system’s shortcomings. These findings, while diverging from sagittal plane results in previous gait-based studies [31,42,43] underscore the need for further refinement of the markerless system to ensure precision and reliability across all planes and positions. Despite its potential, the current limitations of markerless technology suggest that it may not yet be reliable enough for assessing the biomechanics of certain static positions, particularly for the joints and postures we analysed in this study.

Biomechanical variations between positions

Based on the above, we will discuss only the variations from the marker-based system. The results highlight that the biomechanics of each birthing position have unique effects on the hip in all planes, and in the sagittal plane for the pelvis, and trunk. Notably, the “squat” position exhibited the highest levels of hip flexion in the sagittal plane, hip abduction in the coronal plane, and external rotation in the transverse plane. This combination suggests that “squat” produces greater openness of the pelvic outlet, which could facilitate the descent of the foetus and aid in labour progression [44,45]. However, a recent systematic review and meta-analysis does not show the squatting position during childbirth to be definitively beneficial [46] The review indicated that, although the squatting position may decrease the need for instrumental delivery, it is also associated with an increased risk of caesarean section [46]. There was no clear evidence of a reduction in the duration of the second stage of labour or significant improvements in maternal or foetal outcomes. As there is no strong evidence either for or against squatting, women should be encouraged to choose the position they find most comfortable during labour. Furthermore, more studies are needed to better understand childbirth positions, especially in Western countries, as the majority of studies reviewed were conducted in Asia [46]. The applicability of findings from Asian studies to Western populations may be limited due to differences in cultural practices, lifestyle, and anatomical variations. In many Asian countries, squatting is a common daily activity from an early age, which may lead to greater hip mobility and muscular adaptation to the position. In contrast, in Western populations, women may find it more difficult to adopt and sustain a deep squatting position due to differences in habitual movement patterns and musculoskeletal conditioning [47–49]. Additionally, anatomical variations in pelvic morphology between racial and ethnic groups have been documented. For instance, research indicates that White women tend to have a wider pelvic inlet, wider outlet, and shallower anteroposterior outlet compared to African-American women [50]. These structural differences could influence the ease of adopting certain birthing positions and the extent to which they impact labour outcomes. Therefore, further research is needed to explore the effectiveness and feasibility of squatting during labour in diverse populations.

In contrast to squatting, the “upright” position produced the greatest hip extension in the sagittal plane and the least amount of hip abduction in the coronal plane. While this may slightly reduce the dimensions of the pelvic outlet, it could provide a more comfortable and less physically demanding posture for some women. The “upright” position also promoted significant trunk extension in the sagittal plane, which increases the activation of the trunk extensor muscles to maintain stability by counteracting anterior loads. [51]. Additionally, this posture encourages greater spinal extension, which could reduce discomfort for women who prefer an upright position. On the other hand, positions that induce greater trunk flexion, such as “elbows, bent knees,” may serve to redistribute weight and create a more forward-leaning posture, which could be advantageous during certain stages of labour.

The significant variations observed in pelvic tilt between positions are particularly important. The “B-Ball” position, which showed the greatest anterior pelvic tilt in the sagittal plane, could offer increased space in the pelvic inlet, facilitating the early stages of labour by allowing more room for the foetal head to engage. However, anterior pelvic tilt also increases lumbar lordosis (the inward curve of the lower back), which can strain the lower back [52,53], potentially causing discomfort or misaligning the pelvis during labour. On the other hand, positions such as “all-fours” exhibited the greatest posterior pelvic tilt. This reduces the curvature of the lumbar spine, leading to a more neutral or slightly flexed lumbar position, which may help ease back pain and improve spinal alignment [53]. This posterior tilt could be particularly advantageous in later stages of labour when comfort and pain relief are priorities. These variations in pelvic tilt not only influence pelvic openness but also directly impact maternal comfort, especially in later stages of labour when comfort and pain relief are priorities. Positions like “all-fours,” which decrease lumbar lordosis, may be particularly beneficial for labour progression by creating more space in the birth canal and reducing lower back pain.

Limitations of motion capture systems in labour settings

While both the marker-based and markerless motion capture systems provide valuable biomechanical data, significant limitations exist when considering their application in real-world labouring environments. The marker-based system, which requires the placement of retro-reflective markers on the participant’s body, poses several challenges for use in a clinical setting, particularly during labour. The presence of markers can be uncomfortable for a labouring woman, as they will disturb her from lying on her back or side or adopting other positions, causing pressure, discomfort, and pain. Furthermore, markers can dislodge or fall off during movement, active positions, and handling by healthcare professionals or labour companions. In cases of interventions, emergencies and/or transfers, the time required to remove the markers would introduce unnecessary delays and complexities, which, apart from being impractical in critical situations, might also have a negative effect on safety. Additionally, the marker-based system relies on multiple cameras, cables, and processing units that would be challenging to set up in a labouring room and would add the risk of tripping hazards. Also, healthcare professionals and labour companions assist and touch the labouring woman, so their hands or bodies could inadvertently obscure the markers, making it impossible to gather reliable data in such a dynamic and cluttered environment [8,9,54–56]. The markerless motion capture system, although eliminating the need for physical markers, still faces similar practical limitations in a labouring room, including set up and ensuring unobstructed views [57,58]. Additionally, our findings indicate that the markerless system demonstrated lower sensitivity across multiple positions, including in the sagittal plane. This poses a specific challenge in clinical settings, as the sagittal plane is critical for assessing trunk flexion, pelvic tilt, and foetal descent during labour. This finding further underscores the current inadequacy of existing technologies for capturing labour biomechanics in real time. Despite these current hardware-driven limitations, establishing the baseline performance of markerless motion capture under controlled, static conditions is a critical prerequisite for future application in dynamic tasks and/or pregnant cohorts. The present study provides this foundational technical validation. Once system sensitivity and reliability are confirmed, the next necessary step is to evaluate the protocol in late-pregnant cohorts within the laboratory, where hardware limitations can be accommodated. These staged developments will ultimately support the long-term goal of applying markerless techniques during labour, when the technology becomes sufficiently unobtrusive for clinical use.

Although not part of this study, it is worth mentioning that both motion capture systems face additional challenges when considering water births. For the marker-based system, the use of water would likely cause the markers to detach from the skin, rendering the system ineffective. For both systems, the presence of water would introduce additional complexity due to refraction and surface ripples, which could incorrectly track the position of markers and/or create substantial visual noise, affecting the system’s ability to generate reliable data [59,60]. Although it is possible to combine underwater cameras with a land-based camera system into a single motion capture setup, such an extremely sophisticated arrangement would further interfere with the environment, the practice, and the labouring woman’s comfort [61,62].

From the above, it is evident that no currently available motion capture system can effectively measure the biomechanics of labour. This is due to the complexity of current equipment and methodologies, the dynamic nature of birth, and the practical constrains of the labouring environment. These limitations underscore the urgent need to develop new methods capable of accurately capturing labour biomechanics. Understanding the biomechanics of labour remains a critical need—not only because biomechanical complications are a major contributor to maternal and neonatal mortality, especially in low-income countries, [10] but also because it is unacceptable in 2025 that we still lack even a basic understanding of the mechanics involved in the onset of human life. Addressing this knowledge gap is essential for improving childbirth outcomes globally.

Study limitations and future directions

As this was the first study assessing different birthing positions using two motion capture techniques, healthy non-pregnant women were selected as the study population. If pregnant women had been included, we would have needed to assess participants in the third trimester, as close as possible to their due date, due to the significant physiological [63,64] and anatomical changes that occur during pregnancy, particularly toward the latter stages of the third trimester [65,66]. This population would be considered vulnerable due to the potential physical exertion [67], raising ethical concerns. Ensuring the safety of both the mother and foetus, and taking into account the risk-benefit ratio, is paramount [68]. Early-stage research using non-pregnant women allows for the exploration of biomechanical principles without exposing pregnant women to unnecessary risks [68].

Additionally, for an initial study providing essential groundwork, non-pregnant women offer a more controlled group for studying basic biomechanics. Their body weight, centre of gravity, and musculoskeletal function remain relatively stable compared to pregnant women, whose bodies undergo rapid changes [65,69]. This stability allows researchers to isolate and study the effects of specific birthing positions without the added variables introduced by pregnancy. This also helps establish baseline biomechanical data, testing the accuracy and sensitivity of motion capture systems before applying them to the more complex and variable biomechanics of pregnant women. Such a step is crucial for refining equipment and study methods. While using a non-pregnant population allowed for the development and validation of this protocol for the first time and provided valuable insights into general joint biomechanics across upright birthing positions, the findings should be interpreted with caution when translating them to labour scenarios. Key factors such as uterine contractions, foetal descent and rotation, and the dynamic interaction between the foetus and maternal pelvis, which significantly influence labour progression, are not represented in this study. Therefore, while this work establishes foundational biomechanical patterns and supports future methodological development, further studies in term-pregnant populations are necessary to ensure ecological validity and clinical relevance.

In particular, such studies should account for changes in spinal curvature, especially the lumbar curve, which increases by approximately 41% between 12 and 32 weeks of pregnancy [65]. They should also consider hormonal influences, such as relaxin, and their effects on pelvic mobility. However, experimental studies have not shown a direct relationship between high levels of relaxin and increased pelvic mobility or peripheral joint mobility in pregnant women [70]. Shifts in the centre of gravity should also be examined [71]. Furthermore, understanding the biomechanics of labour is crucial not only to enhance maternal and foetal well-being but also to reduce adverse outcomes and complications [72]. Future work could integrate empirical findings into computational models to simulate the influence of specific postures on pelvic dynamics, foetal descent, and maternal effort, offering an additional lens through which to understand the biomechanical optimisation of labour progression. Researchers and clinicians should collaborate to create a method that allows for this analysis and knowledge, as current techniques show significant limitations for use during childbirth in clinical settings [56]. Understanding childbirth should be a top priority on the research agenda, not only because maternal and neonatal mortality remains unacceptably high [72–74] but also because, among all the Sustainable Development Goals (SDG) for 2030, the SDG 3.1 target to reduce global maternal mortality is the only one that has failed [75]. Moreover, it is unacceptable to have such limited knowledge and understanding of the biomechanics of the beginning of human life.