Enhancing fall risk assessment: instrumenting vision with deep learning during walks

BACKGROUND

Falls are common in a range of clinical cohorts, where routine risk assessment often comprises subjective visual observation only. Typically, observational assessment involves evaluation of an individual's gait during scripted walking protocols within a lab to identify deficits that potentially increase fall risk, but subtle deficits may not be (readily) observable. Therefore, objective approaches (e.g., inertial measurement units, IMUs) are useful for quantifying high resolution gait characteristics, enabling more informed fall risk assessment by capturing subtle deficits. However, IMU-based gait instrumentation alone is limited, failing to consider participant behaviour and details within the environment (e.g., obstacles). Video-based eye-tracking glasses may provide additional insight to fall risk, clarifying how people traverse environments based on head and eye movements. Recording head and eye movements can provide insights into how the allocation of visual attention to environmental stimuli influences successful navigation around obstacles. Yet, manual review of video data to evaluate head and eye movements is time-consuming and subjective. An automated approach is needed but none currently exists. This paper proposes a deep learning-based object detection algorithm (VARFA) to instrument vision and video data during walks, complementing instrumented gait.

METHOD

The approach automatically labels video data captured in a gait lab to assess visual attention and details of the environment. The proposed algorithm uses a YoloV8 model trained on with a novel lab-based dataset.

RESULTS

VARFA achieved excellent evaluation metrics (0.93 mAP50), identifying, and localizing static objects (e.g., obstacles in the walking path) with an average accuracy of 93%. Similarly, a U-NET based track/path segmentation model achieved good metrics (IoU 0.82), suggesting that the predicted tracks (i.e., walking paths) align closely with the actual track, with an overlap of 82%. Notably, both models achieved these metrics while processing at real-time speeds, demonstrating efficiency and effectiveness for pragmatic applications.

CONCLUSION

The instrumented approach improves the efficiency and accuracy of fall risk assessment by evaluating the visual allocation of attention (i.e., information about when and where a person is attending) during navigation, improving the breadth of instrumentation in this area. Use of VARFA to instrument vision could be used to better inform fall risk assessment by providing behaviour and context data to complement instrumented e.g., IMU data during gait tasks. That may have notable (e.g., personalized) rehabilitation implications across a wide range of clinical cohorts where poor gait and increased fall risk are common.

Different neurocognitive controls modulate obstacle avoidance through pregnancy

Understanding why falls during pregnancy occur at over 25% rate over gestation has clinical impacts on the health of pregnant individuals. Attention, proprioception, and perception of the environment are required to prevent trips and falls. This research aimed to understand how the changes to these neurocognitive processes control obstacle avoidance through gestation. Seventeen pregnant participants were tested five times in 6-week intervals. Participants walked an obstacle course (OC), and we analyzed the crossings over obstacles that were set to 10% of participants' body height. Participants also performed an attentional network test (ANT: performance of specific components of attention), an obstacle perception task (OP: ability to visually define an obstacle and translate that to a body posture), and a joint position sense task (JPS: ability to recognize and recreate a joint position from somatosensation). In the OC task, average leading and trailing foot crossing heights significantly reduced by 13% and 23% respectively, with no change in variation, between weeks 13 and 31 of pregnancy, indicating an increased risk of obstacle contact during this time. The variability in minimum leading foot distances from the obstacle was correlated with all three neurocognition tasks (ANT, OP, and JPS). Increased fall rates in the second and third trimesters of pregnancy may be driven by changes in attention, with additional contributions of joint position sense and environmental perception at various stages of gestation. The results imply that a holistic examination on an individual basis may be required to determine individual trip risk and appropriate safety modifications.

Determining fall risk change throughout pregnancy: the accuracy of postpartum survey and relationship to fall efficacy

All epidemiological studies on pregnancy fall risk to date have relied on postpartum recall. This study investigated the accuracy of postpartum recall of falls that were reported during pregnancy, including assessment of fall efficacy as a possible reason for recall inaccuracy. Twenty participants reported fall experiences weekly during pregnancy, but one participant was excluded as an outlier. A fall efficacy questionnaire was completed every six weeks during pregnancy. A postpartum survey to mimic previous studies (Dunning, Lemasters, and Bhattacharya 2010; Dunning et al. 2003) was delivered to determine recall accuracy. Postpartum recall of fall events each gestational month matches the previous study (Dunning, Lemasters, and Bhattacharya 2010). However, recall of falls is 16% underestimated and recall of all fall events is 30% overestimated in postpartum survey. There is a slight relationship between fall efficacy and true falls, but not between fall efficacy and fall recall. Our study suggests fall risk needs to be intermittently surveyed throughout pregnancy rather than assessed via postpartum survey.Practitioner summary: This study investigated the accuracy of postpartum survey of fall risk during pregnancy and the possibility of fall efficacy as a covariate. We used three corresponding surveys. We found inaccuracies in postpartum survey, not explain by fall efficacy.

Upper extremity kinematics during walking gait changes through pregnancy

BACKGROUND

Thirty percent of adults in the United States use wearable fitness devices as of 2020 [1], such as fitness watches, to monitor and track health and physical activity parameters. Physical changes during pregnancy may impact wrist worn device accuracy. The arms may be needed as compensation during walking because thorax axial rotation may be inhibited by pelvic tilt during pregnancy [2].

METHODS

To examine arm motion changes, twenty-three pregnant women (28 ± 4 y) were tested in four-week intervals ( ± 2 weeks) at 18-, 22-, 26-, 30- and 34-weeks' gestation. Kinematic data were measured during self-selected speed walking. Segment angles and angular velocities were analyzed over time. Linear regressions were used to analyze the correlations between arm motion and the other kinematic variables.

RESULTS

Arm range of motion significantly increased (p = 0.006) over gestation, but leg, thorax, and pelvis range of motions did not significantly change. Arm range of motion was correlated with pelvis (r² =0.311, p = 0.001, β = 1.724) and leg (r² = 0.285, p = 0.004, β = 1.520) range of motion and gait velocity (r² =0.566, p = 0.001, β = 39.110). Arm velocities significantly increased (p < 0.012), as did leg velocities (p < 0.022) over gestation time, but thorax and pelvis rotational velocities did not significantly change over time. Arm velocity was correlated with leg velocity in both flexion (r² =0.598, p = 0.001, β = 1.61) and extension (r² =0.568, p = 0.001, β = 1.35).

SIGNIFICANCE

Arm swing increases over the course of gestation during walking, which does not follow the exact pattern of changes seen in the legs, thorax, and pelvis. These results show that a typical gait analysis of lower body motions may miss important biomechanical changes or compensations at different points over pregnancy. Future studies should examine why these changes may occur. Studies should also be conducted to see if arm changes impact outcome parameters from fitness watches and affect their validity as an exercise tracker during pregnancy.

Assessment of physical fitness during pregnancy: validity and reliability of fitness tests, and relationship with maternal and neonatal health - a systematic review

Objectives

To systematically review studies evaluating one or more components of physical fitness (PF) in pregnant women, to answer two research questions: (1) What tests have been employed to assess PF in pregnant women? and (2) What is the validity and reliability of these tests and their relationship with maternal and neonatal health?

Design

A systematic review.

Data sources

PubMed and Web of Science.

Eligibility criteria

Original English or Spanish full-text articles in a group of healthy pregnant women which at least one component of PF was assessed (field based or laboratory tests).

Results

A total of 149 articles containing a sum of 191 fitness tests were included. Among the 191 fitness tests, 99 (ie, 52%) assessed cardiorespiratory fitness through 75 different protocols, 28 (15%) assessed muscular fitness through 16 different protocols, 14 (7%) assessed flexibility through 13 different protocols, 45 (24%) assessed balance through 40 different protocols, 2 assessed speed with the same protocol and 3 were multidimensional tests using one protocol. A total of 19 articles with 23 tests (13%) assessed either validity (n=4), reliability (n=6) or the relationship of PF with maternal and neonatal health (n=16).

Conclusion

Physical fitness has been assessed through a wide variety of protocols, mostly lacking validity and reliability data, and no consensus exists on the most suitable fitness tests to be performed during pregnancy.

PROSPERO registration number

CRD42018117554.

Changes in static balance during pregnancy and postpartum: A systematic review

BACKGROUND

Because pregnant women show a high risk of falling, some researchers examined their balance during static standing. This systematic review summarized the findings from all studies evaluating static balance in women during pregnancy and postpartum.

RESEARCH QUESTION

Do pregnant and postpartum women show differences in static balance compared to non-pregnant women, and does static balance change during pregnancy and postpartum?

METHODS

Pubmed, Embase, CINAHL, and Web of Science databases were searched systematically from inception until Feb 23, 2022. Studies were eligible for inclusion if they measured COP sway with a force plate during bipedal static standing, and compared COP outcomes between healthy pregnant or postpartum women and non-pregnant women, and/or during different stages of pregnancy and the postpartum period. Methodological quality was assessed overall with a modified version of the Downs and Black checklist, and specifically related to COP measurement by using recommendations of Ruhe et al. (2010). The protocol was registered in PROSPERO (CRD42020166302).

RESULTS

Thirteen studies were included. Because methodological approaches varied greatly between studies, results were summarized descriptively. Studies reported either greater overall and anteroposterior COP sway magnitude, velocity and variability in women from the second half of pregnancy until six months postpartum compared to non-pregnant controls, or no differences in static balance. Changes in static balance throughout pregnancy were generally not found. Finally, there was no clear consensus on the influence of pregnancy on the reliance on visual inputs for balance control, and on whether differences in balance in pregnant and postpartum women reflect poorer balance or positive adaptations to the physical changes experienced during pregnancy.

SIGNIFICANCE

Methodological heterogeneity between studies prevented us from drawing strong conclusions regarding the effect of pregnancy on static balance. Assessing the methodological quality of the studies revealed weaknesses that should be taken into account in future studies.

Self-selection of gestational lumbopelvic posture and bipedal evolution

BACKGROUND

Not all pregnant women seem to select the more curved lumbopelvic posture that their sexual dimorphic anatomy allows even though many previous researchers have assumed lumbopelvic curvature to be standard during pregnancy. This study is vital to understanding coevolution of lumbopelvic sexual dimorphism and bipedalism, and understanding some clinical implications of intervening in gestational posture changes.

RESEARCH QUESTIONS

Are there anthropometric changes that correspond with selection of lumbopelvic curvature change during pregnancy? What are the biomechanical costs and benefits of gestational lumbopelvic curvature change?

METHODS

Twenty pregnant women were tested at five different times in the 2nd and 3rd trimesters of pregnancy. Lumbopelvic posture, standing kinetics and gait kinetics were measured longitudinally. Additionally, we modeled the effects on standing and gait without lumbopelvic postural changes, but with anthropometric changes, for each individual.

RESULTS

We found greater lumbopelvic angulation to correspond with a shorter body height (6 cm difference between groups, p = 0.048) and deeper 2nd trimester abdomen (2 cm difference between groups, p = 0.013). Lumbopelvic angulation lowers support requirements (in standing and walking (6% lower support impulse, p = 0.056), but at the cost of shifting the propulsive actions to a less efficient pulling action rather than pushoff (13 % reduction in pushoff time, p = 0.001). We observed minimal effects on walking kinematics and balance control.

SIGNIFICANCE

Our findings suggest the evolutionary advantage of the female lumbopelvic unit is the adaptability it provides to adjust for the individual needs of the pregnant woman. We discuss multiple potential contributing factors that may have shaped hominin female lumbopelvic evolution and are involved in self-selecting lumbopelvic posture.

Shoulder and elbow requirements during sagittal reach as a result of changing anthropometry throughout pregnancy

During pregnancy, anthropometric and physiological changes can result in difficulty reaching for and lifting everyday objects. The aims of this study were to determine the changes in sagittal plane anterior reach space (SPARS) and shoulder/elbow strength requirements throughout pregnancy. Seventeen participants were tested through a longitudinal observational cohort study between 16 and 36 weeks gestation in four-week intervals. A 25% decrease in SPARS was observed at the L3-4 torso height. Combined with arm mass increases, shoulder and elbow moment requirements at the minimum and maximum static reach distances significantly increased. However, inverse dynamics analysis determined that mass gains in the arm alone only minimally impact dynamic shoulder moments. Additionally, torso flexion increases throughout pregnancy demonstrates that women are attempting to compensate for decreased SPARS, possibly indicating the additional perceptual importance of reach space in accommodations for pregnant workers.

Correlations between joint kinematics and dynamic balance control during gait in pregnancy

BACKGROUND

Dynamic balance control degrades during pregnancy, but it is not yet understood why. Mechanical aspects of the body should directly affect walking balance control, but we have recently published papers indicating that weight gains during pregnancy explain very little dynamic balance changes. Our goal was to determine if lower extremity joint kinematic changes are an indicator of walking balance control. This information is vital to understanding the route by which pregnancy increases fall risk.

METHODS

Twenty-three pregnant women were tested at five different times in the 2nd and 3rd trimesters of pregnancy. Participants performed walking trials at a self-selected pace. Motion capture was used to measure joint kinematics (discrete and coordination variables) and body center of mass motion. Changes over time were statistically analyzed. Correlations between kinematics and walking balance were modelled with hierarchical multiple regression models.

RESULTS

As pregnancy progresses, it appears that a more flexed hip posture could be driving lower extremity kinematic changes toward increased coordination between joints and increased knee and ankle motions. Walking balance changes were also detected through increased COM motion (lateral range of motion and velocity) in the lateral directions. However, there was little correlation between kinematic and balance changes (r² < 0.4). Strong correlations were only observed when all kinematics (including those that don't ubiquitously change during pregnancy) were used in the regression model (r² > 0.7).

SIGNIFICANCE

Our findings suggest that walking balance control is not altered by a common kinematic change between all pregnant women. While increased lateral center of mass motion should be expected with pregnancy, the kinematics leading to this increase may be person-specific. The cause of dynamic imbalance in each pregnant women (physiological, mechanical, and neurocognitive) may play an important role in determining the kinematic means by which lateral center of mass motion increases.

An analysis of postpartum walking balance and the correlations to anthropometry

BACKGROUND

Falls caused by balance issues during pregnancy are quite common, and these issues can continue postpartum, potentially posing a danger to both the mother and baby. While there has been research on changes to walking gait during pregnancy, walking balance in the postpartum period has yet to be examined. Therefore, the aims of this study were to examine if balance changes persist in postpartum and the contribution of anthropometry changes.

METHODS

This was done through longitudinal observational cohort study at 16 and 40 weeks gestation and at four-week intervals postpartum. Balance was measured as lateral center of mass motion during treadmill walking, and recorded with motion capture cameras following anthropometric measurements. Balance variables were statistically analyzed to observe how they changed over time. Hierarchical regression analyses determined correlations between balance and anthropometry.

RESULTS

Balance was observed to improve significantly just following birth. Additionally, there were changes that continued to indicate improvement throughout the postpartum period. Anthropometry changes were significantly, but minimally, correlated with balance changes.

SIGNIFICANCE

Many women begin to return to normal activities soon after birth. With women participating in various forms of exercise, potentially rigorous work requirements, and tasks around the home, it is important that they, their medical providers, and employers understand and consider the continued risks of imbalance.