Identifying Sensors-based Parameters Associated with Fall Risk in Community-Dwelling Older Adults: An Investigation and Interpretation of Discriminatory Parameters

doi:10.21203/rs.3.rs-2851345/v1

Download PDF

Research Article

Identifying Sensors-based Parameters Associated with Fall Risk in Community-Dwelling Older Adults: An Investigation and Interpretation of Discriminatory Parameters

https://doi.org/10.21203/rs.3.rs-2851345/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 01 Feb, 2024

Read the published version in BMC Geriatrics →

You are reading this latest preprint version

Background

Falls pose a severe threat to the health of older adults worldwide. Determining gait and kinematic parameters that are related to an increased risk of falls is essential for developing effective intervention and fall prevention strategies. This study aimed to investigate the discriminatory parameter, which lay an important basis for developing effective clinical screening tools for identifying high-fall-risk older adults.

Methods

Forty-one individuals aged 65 years and above living in the community participated in this study. The older adults were classified as high-fall-risk and low-fall-risk individuals based on their BBS scores. The participants wore an inertial measurement unit (IMU) while conducting the Timed Up and Go (TUG) test. Simultaneously, a depth camera acquired images of the participants' movements during the experiment. After segmenting the data according to subtasks, 142 parameters were extracted from the sensor-based data. A t-test or Mann-Whitney U test was performed on the parameters for distinguishing older adults at high risk of falling. Multiple logistic regression was used to further quantify the role of different parameters in identifying high-fall-risk individuals.

Results

Fifteen participants were defined as high-fall-risk individuals, while twenty-six were defined as low-fall-risk individuals. 17 parameters were tested for significance with p-values less than 0.05. Multiple logistic regressions were constructed for the parameters with the p-values less than 0.005, including the coefficient of variation of the angular velocity in pitch (CV-Ang-Pitch) during the turn, usage of walking assistance, and the max and coefficient of variation of the angular velocity in yaw (Max-Ang-Yaw, CV-Ang-Yaw) during turn-to-sit. The results showed that the CV-Ang-Pitch during the turn was the parameter with the greatest effect on identifying high-fall-risk individuals.

Conclusions

High-fall-risk individuals typically exhibited a cautious gait, such as larger step width and shorter step length during walking. Besides, we identified some abnormal gait patterns of high-fall-risk individuals compared to low-fall-risk individuals, such as greater knee angle and a tendency to tilt the pelvis forward during turning. These findings provide insight into the mechanisms underlying the activities of high-fall-risk individuals compared to normal individuals and illustrate the role of sensors in identifying high-fall-risk individuals among community-dwelling older adults. (350 words)

fall risk

timed up and go test

older adults

inertial measurement unit

depth camera

Falls have been recognized as a major health problem for older adults with significant physical and psychological consequences [1]. Community-dwelling older adults (age ≥ 65 years) experience 0.3–1.6 falls per person annually, of which approximately 5% result in fractures or require hospitalization[2]. As reported [3], falls have become the second leading cause of injury-related deaths among older adults worldwide, and they are a major cause of both death and injury in people over 65 years of age. Existing literature shows that several inherent risk factors can contribute to different gait and balance deficits that would result in falling, such as decreased muscle strength, limited range of movement, abnormal muscle tone, lack of motor coordination, poor sensory organization, and cognitive deficit [4]. It is crucial to reduce the burden of the disease brought by falls. Studies have shown that timely identifying people at risk of falls and implementing fall prevention programs can successfully reduce fall occurrences [5].

There are some clinical measurements that were widely used to evaluate the fall risk of older adults, such as the Berg Balance Scale (BBS), Tinetti Performance Oriented Mobility Assessment (POMA), Balance Evaluation System Test (BESTest), and Functional Gait Assessment (FGA) [6, 6–8]. Among these BBS is a 14-item scale that can quantitatively assesses balance and fall risk in older community-dwelling adults [9]. However, it takes 15–20 minutes and requires expert supervision to perform the BBS test [10]. Conventional assessment instruments entail considerable time and effort, which impedes their rapid and extensive application at the communal level [11].

The advancement of information technology and sensor-based methodologies has facilitated the implementation of efficacious fall risk assessment techniques. These techniques contain sophisticated information, provide comprehensive data, and can significantly reduce the time and labour required for assessment. Among these methods, nearable (environmental perception) and wearable sensors are widely employed due to their high efficacy. Typical examples of environmental perception sensors are infrared sensors, Doppler radar, depth camera systems, and force platforms [12–15] which facilitate accurate and precise monitoring, detection, and analysis of various environmental parameters. Typical examples of wearable sensors are pressure insoles and inertial measurement units (IMUs) [16, 17] which can monitor human activities in real-time and provide kinematic data for subsequent analysis.

The integration of standard tests and sensor-based device can not only simplify the fall risk assessment process, but also assess fall risk in older adults accurately by extracting gait parameters during their movement [18, 19]. There was various existing research concerning gait parameters extracted from IMU or depth camera data of the Timed Up and Go (TUG) test assessing mobility or balance among older adults. Kim et al. conducted the IMU-instrumented TUG test to evaluate the walking and turning parameters of the patients with unilateral vestibular deafferentation before and after rehabilitation. Results indicate that the IMU-instrumented TUG test can objectively quantify the change in their turning parameters after physical therapy to distinguish individuals with UVD [18]. Dubois et al. automated the TUG test with the Microsoft Kinect v2 application. Based on the Kinect data collected from tests, several parameters relating to the gait and turn pattern, the sitting position, and the duration of each phase were extracted to build an evaluation system for the mobility and balance of individuals. The evaluation system can objectively quantify the change in parameters of different subtasks of TUG to provide discrimination of high risk for falls in older adults [20]. Previous studies have demonstrated the validity and reliability of each sensor individually, but their complementary and synergistic effects have not been thoroughly examined. The combination of IMUs and depth cameras for gait and fall risk analysis is a novel and promising approach that has received limited attention in the literature. This research gap motivates further exploration of the potential advantages of integrating these two sensors in fall risk assessment.

Also, previous research on sensor-based fall risk prediction has been restricted to the extraction of parameters from a single sensor or the absence of interpretation of the parameters’ meaning. Studies by Hsu et al. and Yu et al. [21, 22] used the IMU-extracted parameters directly for fall risk prediction without discussing their role in balance assessment. Studies by Kargar B. et al. and Kim et al. [23, 24] employed only the Kinect-extracted parameters, which could be susceptible to measurement imprecision. While IMUs can precisely record the movement of particular body parts, more IMUs are necessary to record additional parts. Conversely, depth camera can capture more key points during motion despite their lack of precision. Therefore, we combined parameters from the IMU and depth camera used in previous studies to investigate their role in fall risk assessment among community-dwelling older adults.

This study applied depth camera technology and IMU to measure the gait parameters of geriatric community members. The objective was to identify factors that discriminate between high-fall-risk and low-fall-risk individuals based on 142 parameters derived from the 3m-TUG subtask. These parameters encompassed demographic, gait, kinematic, and anatomical variables. A multiple logistic regression model was employed to evaluate the contribution of factors to fall risk. The paper aimed at discussing and interpreting the significant parameters associated with fall risk in light of the relevant literature, considering the physiology of older adults. It also illustrated the potential of depth camera technology and IMU for gait assessment in non-clinical settings.

2.1 Participants

We recruited older adults who met all the following inclusion criteria: at least 65 years old, living in a community setting, having the ability to walk independently or with walking assistance, with normal (or corrected-to-normal) vision, and having the ability to provide informed consent. We excluded older adults with abnormal vision, physical impairments, and life-threatening illnesses, as they were likely unable to complete the gait and balance assessment. The pilot study was approved by the Research Ethics Committee of the City University of Hong Kong(reference number: 3-2020-02-F). All participants provided written informed consent before the initiation of our study.

The data collection was conducted between December 2019 and January 2020, and fifty-four participants from 4 different community settings were recruited based on the inclusion and exclusion criteria. However, only forty-one participants (thirty-six women and five men) aged 68–89 years (79.07 ± 6.47 years) completed all the tests due to some participants declining the depth camera recording or withdrawing from the experiment.

2.2 Assessment and protocol

During the experiment phase, participants were required to complete two sets of tests: the BBS assessment and the 3m-TUG test.

BBS: The BBS consists of 14 items examining the balance capability of older adults, and each item is scored on a scale of 0 to 4. If the participant is unable to do the task, the score is recorded as 0, and if the participant is able to complete the task according to the criteria assigned to it, the score is 4. The maximum total score on the test is 56. The BBS was developed as a balance assessment to objectively assess a person's ability to control balance in a functional task [25]. For the prediction of falls in community-dwelling older adults, a cutoff score of 50 was proposed [26]. Participants were labeled as low-fall-risk individuals if their BBS score was > 50. Otherwise, they were labeled as high-fall-risk individuals. Therefore, we identified twenty-six low-fall-risk individuals and fifteen high-fall-risk individuals.
3m-TUG test: participants were required to stand up from a chair, walk 3 meters, turn around 180°, then walk back around and sit in the chair again [27]. The TUG is a simple and feasible tool that can be administered in any setting and at any time [28]. It involves basic mobility skills such as rising, turning, sitting, and walking. Therefore, it is recommended as a screening tool to identify older adults with balance or gait impairments [29]. Figure 1 demonstrates the general process of 3m-TUG. In our experiment, all participants finish the TUG without any walking assistance.

Figure 1. The process of 3m-TUG. A Kinect is positioned on the right side to capture the motion of human skeletal joints.

During the experiment, we used a commercial IMU sensor (9 Degrees of Freedom) and a Kinect (Microsoft Kinect for Windows v2) device to capture the participants' movement parameters while they completed the 3m-TUG.

IMUs are widely used for the analysis of gait parameters and fall risk assessment due to their accuracy, portability, and low price [30]. IMU with a sampling frequency of 5 Hz consisting mainly of an accelerometer and a gyroscope was placed on the participants' lower back area between the L3 and L5 vertebrae. Acceleration signals were derived from three axes: vertical (V), mediolateral (ML), and anterior-posterior (AP). Angular velocity of a gyroscope: yaw, pitch, and roll rotate around V, ML, and AP, respectively, as shown in Fig. 2.

Figure 2. (a) The IMU is positioned between L3 and L5 vertebrae. (b) IMU's three axes.

Kinect is a portable device with an integrated depth sensor and a micro camera. Different from the traditional camera system, the depth camera technology obtains 3D information about subjects with the application of the stereo calibration method, which can produce RGB images and dynamically capture the movement of the human body for accurate analysis of human posture [31, 32]. As Fig. 3 shows, the key-point data collected from Kinect is presented in the form of 25 human skeletal joints in the three-dimensional coordinate, which means that the motion of each key-point can be studied individually.

Figure 3. 25 human key-point detected by Kinect.

2.3 Divide the TUG into subtasks

Dividing the TUG into subtasks can provide a more detailed and nuanced assessment of mobility and functional ability in older adults. Figure 4 shows an illustrated example of the gyroscope data of a person during the TUG test. TUG was divided into five subtasks: sit-to-stand, walking forward, turn, walking back, and turn-to-sit. The five segmentations were detected with angular velocities in the yaw and pitch axes using the developed algorithm presented by previous related work [33, 34].

2.4 Parameters extraction and statistical analysis

The goal of this analysis was to identify the significant difference between high-fall-risk individuals and low-fall-risk individuals. Each subtask provides different physical function information, so the parameters were extracted based on different tasks. A walking phase is created by merging the forward and backward walking. The study examined a total of 142 parameters that have been employed in previous studies on fall risk identification [21–24]. Table 1 shows the main parameters extracted from IMU and Kinect.

8 parameters were obtained from the subtasks' temporal data: duration is the time required to complete the subtasks; the percentage of duration (POD) is the subtask duration divided by the TUG duration.
3*22 parameters were obtained from acceleration data: the maximum, minimum, median, range, root mean square (RMS), coefficient of variation (CV), maximum jerk (Max jerk), and mean jerk of the triaxial accelerometers during each subtask were calculated. Among them, jerk represents the rate of change of acceleration.
3*18 parameters were obtained from angular velocity data: the maximum, minimum, median, range, RMS, and CV of the triaxial angular velocity during each subtask were calculated.
10 parameters were obtained from key-point data: legs angle is the angle between the two legs, defined as the angle between the two vectors of the line between the joints 0 and 17 and the line between 0 and 13 (Fig. 3); knee angle is the angle between the thigh and the calf, for example, the right knee angle is the angle between the two vectors 17 and 16 of the joint and 17 and 18 of the joint (Fig. 3); step counts is the number of steps during walking; step length is the total length of walking (6m) divided by the number of steps; step duration is the duration of walking divided by the number of steps; step width is the average width between the two feet.
1 parameter was the duration of the TUG test.
3 parameters were obtained from demographic information, including gender, age, and the use of walking assistance.

For all extracted parameters, after testing the normality of the distribution of the two groups, statistics comparing high-fall-risk individuals and low-fall-risk individuals were performed using the t-test or Mann-Whitney U test with a significance level of\(\alpha =0.05\). We then selected parameters with significance p-values less than 0.005 for multiple logistic regression to further quantify the role of different parameters in identifying individuals at high risk of falling. Statistical analyses were conducted using the R-4.2.1.

Table 1

Summary of the main parameters extracted from IMU and Kinect.
	Sit-to-stand	Walking	Turn	Turn-to-sit
Temporal(8)	Duration(G), POD(G)	Duration(G), POD(G)	Duration(G), POD(G)	Duration(G), POD(G)
Acceleration(3*22) (V, ML, AP)	Maximum(K), Minimum(K), Range(K), RMS(K), Max jerk(K), Mean jerk(K)	Range(K), RMS(K), Median(K), CV(K)	Range(K), RMS(K), Median(K), CV(K)	Maximum(K), Minimum(K), Range(K), RMS(K), Median(K), CV(K), Max jerk(K), Mean jerk(K)
Angular velocity(3*18) (yaw, pitch, roll)	Maximum(K), Minimum(K), Range(K), RMS(K)	Range(K), RMS(K), Median(K), CV(K)	Range(K), RMS(K), Median(K), CV(K)	Maximum(K), Minimum(K), Range(K), RMS(K), Median(K), CV(K)
Key-point(10)		Legs angle(A), Left knee angle(A), Right knee angle(A), Step counts(G), Step length(G), Step duration(G), Step width(G)	Legs angle(A), Left knee angle(A), Right knee angle(A)
Note. Each parameter is annotated with its type: K for kinematic parameters, D for demographic parameters, G for gait parameters, and A for anatomical parameters

3.1 Demographic parameters of participants

The general parameters of the participants are presented in Table 2. It can be seen that the two groups differ significantly in the parameters 'Age' and 'usage of walking assistance' in daily life. Compared with low-fall-risk individuals, older age and use of walking assistance in daily life were observed in high-fall-risk individuals, as expected.

Table 2

Demographic parameters of the participants.
Parameter	Low-fall-risk individuals	High-fall-risk individuals	p-Values
Residents (n)	26	15	--
Age (years ± standard deviation)	77.38 ± 5.34	82.00 ± 7.36	0.041
Gender (M/F)	2/24	3/12	0.262
Usage of walking assistance (Yes/NO)	4/22	10/5	0.001

3.2 Significant parameters extracted from IMU and Kinect

Seventeen parameters showed significant differences between the two groups. Table 3 shows the p-values of the significant parameters. The p-value of the CV of the angular velocity at the pitch angle (CV-Ang-Pitch) during the turn is most significant in distinguishing the two groups. During the turn-to-sit phase, the angular velocity of the low-fall-risk individuals has a greater maximum value or a greater CV in the yaw angle (Max-Ang-Yaw, CV-Ang-Yaw). Parameters are available at all stages of the 3m-TUG to distinguish high-fall-risk individuals from low-fall-risk individuals (see Table 3). Complete results, including test results for 142 parameters, are provided in the Additional file.

Table 3

Significant parameters (p < 0.05).
Parameter	Subtask	Axis	Low-fall-risk individuals	High-fall-risk individuals	p-Values
CV [deg/s]	Turn	Pitch	2.19 ± 8.77	-3.75 ± 6.31	< 0.001
Usage of walking assistance [Yes/NO]	--	--	4/22	10/15	0.001
Max [deg/s]	Turn-to-sit	Yaw	161.28 ± 44.40	122.25 ± 33.24	0.003
CV [deg/s]	Turn-to-sit	Yaw	1.31 ± 0.31	1.09 ± 0.15	0.005
Step length [m]	Walk	--	0.37 ± 0.07	0.31 ± 0.06	0.007
Step width [m]	Walk	--	0.13 ± 0.03	0.16 ± 0.03	0.008
Median [deg/s]	Turn	Pitch	2.10 ± 6.45	-2.80 ± 5.86	0.021
Right knee angle [deg]	Walk	--	158.95 ± 5.48	162.18 ± 3.19	0.022
RMS [deg/s]	Walk	Pitch	8.16 ± 2.41	6.64 ± 1.77	0.026
Max [m/s²]	Turn-to-sit	V	1.14 ± 0.05	1.19 ± 0.07	0.026
Mean jerk [m/s²]	Turn-to-sit	AP	0.10 ± 0.03	0.08 ± 0.02	0.029
POD [%]	Walk	--	39.27 ± 6.53	43.43 ± 6.74	0.031
Right knee angle [deg]	Turn	--	157.71 ± 10.60	164.44 ± 9.05	0.034
Age [years]	--	--	77.38 ± 5.34	82.00 ± 7.36	0.041
Mean jerk [m/s²]	Sit-to-stand	ML	0.04 ± 0.02	0.05 ± 0.02	0.050
RMS [deg/s]	Turn-to-sit	Yaw	71.24 ± 24.6	57.22 ± 17.33	0.050
Max [deg/s]	Turn-to-sit	Roll	18.72 ± 11.3	26.67 ± 11.55	0.050
Note. The results of ranking for 17 significant parameters' p-values from smallest to largest for the t-test or Mann-Whitney U test. The values on each parameter were reported as mean ± standard deviation.

3.3 Logistic regression model

To examine the association between the most significant parameters and fall risk as indicated by the results of significance testing, and to assess the strength of this association, we fitted multiple logistic regression model for four parameters with p-values less than 0.005: CV-Ang-Pitch during the turn, usage of walking assistance, Max-Ang-Yaw during turn-to-sit, and CV-Ang-Yaw during turn-to-sit. The results are presented in Table 4.

Table 4

Multiple logistic regression of probability of falling.
Parameter	Subtask	Axis	Coefficient	Standardized coefficients	Odds ratio	95% Confidence Limits
CV	Turn	Pitch	-0.238	-1.996	0.788	(0.616,1.008)
assistance	--	--	1.970	0.946	7.168	(1.127,45.579)
Max	Turn-to-sit	Yaw	-0.021	-0.926	0.979	(0.952,1.007)
CV	Turn-to-sit	Yaw	-4.407	-1.233	0.012	(0.000,2.013)
constant	--	--	6.501	-1.296	665.814	--

The odds ratio reflects the effect of different levels of factors represented by the independent parameter on the odds of falls. As shown in Table 4, the odds ratio of usage of walking assistance is greater than 1, suggesting that walking assistance is a risk factor for high-fall-risk individuals. This implies that older adults who use walking assistance have 7.168 times higher odds of falling than those who do not. The standardized coefficient can be used to compare the relative importance of different parameters on fall risk. Among the four parameters used for regression, the absolute value of the standardized coefficient of the CV-Ang-Pitch during the turn is the largest, indicating that it has the strongest influence on fall risk.

4.1 General considerations

An important contribution of this study is the quantitative assessment of multiple parameters associated with fall risks among community-dwelling residents aged 65 years old and above. These parameters can be classified into four categories based on previous studies [23, 35], the first category of parameters being demographic parameters (e.g., age); the second category being kinematic parameters extracted from the IMU (e.g., maximum acceleration, the angular velocity at turn); the third category being gait parameters (e.g., step count, step length, and walking duration); and the fourth category being anatomical parameters (e.g., knee angle). Our results indicate that 17 parameters are relevant for assessing the fall risk of older adults, comprising 11 kinematic parameters, 3 gait parameters, 2 demographic parameters, and 1 anatomical parameter. Therefore, we demonstrate that electronic devices such as IMU and Kinect can facilitate the detection of gait impairment in older adults and reduce the reliance on manual evaluation of fall risk by health professionals.

Participants performed a simple TUG test, which is a widely used geriatric assessment tool to evaluate balance, gait speed, and functional ability for basic activities of daily living in older adults [36]. Each subtask of the TUG test involved different competency requirements related to balance and body control. The sit-to-stand phase entailed lifting the hips from a chair, reaching maximum dorsiflexion at the ankles, and trunk and hip extension, which required momentum transfer and stabilization of all body parts to complete [37]. The forward momentum of human walking is converted into angular momentum at the onset of the walking phase. To prevent falls, this angular momentum needs to be regulated and recovered from the trip using “elevating” and “lowering” strategies [38]. The turning phase entails the body translating and rotating in the new direction of travel, while maintaining dynamic stability [39]. From standing to sitting, trunk flexion and knee flexion are almost synchronized, requiring control of the whole body’s center of gravity to achieve anterior-posterior and vertical displacement [40]. We merged the second turning and sitting phases into the turn-to-sit phase because older adults typically perform both movements simultaneously. Therefore, this phase provides more physiological information about the elderly. We identified more significant parameters in the turn-to-sit phase compared to the other phases, with seven in total. This is followed by five significant parameters in the walking phase, where previous studies have demonstrated the highest correlation between walking and the incidence of falls [41]. Assessing the ability of older adults to perform different complex tasks during TUG may be more informative than evaluating a single balance ability task.

4.2 Focusing parameters

Using TUG duration alone to distinguish fallers from non-fallers has long been controversial. Although similar to the results of previous studies [42], the TUG duration in this experiment was longer in high-fall-risk individuals than in low-fall-risk individuals, but the p-value was nonsignificant (high-risk individual: 16.11 ± 6.36, low-risk individual: 13.53 ± 3.52, p = 0.21). A systematic review [43] noted that the TUG duration is not useful in distinguishing fallers from non-fallers in a healthy state but is more valuable in poor-healthy older adults. This is consistent with our results because our participants were older healthy residents from the community, so no significant TUG time threshold was found to identify high risk in the older adults. However, we found that the percentage of the duration (POD) of walking is a significant parameter to distinguish between the two groups (high-fall-risk individual: 43.43 ± 6.74, low-fall-risk individual: 39.27 ± 6.53, p = 0.03). Motor control is a complex interplay of cognitive and sensorimotor systems, and studies have demonstrated that the speed of information processing declines with age (i.e., time increases) by approximately 26% [44]. The difference in the POD of walking between the two groups may be associated with the different speeds of information processing for postural transitions or different walking strategies.

We also found that the demographic parameters were meaningful in distinguishing the two populations. Older adults or those who use walking assistance are more likely to be defined as high-fall-risk individuals. This result is consistent with the reality that the rate of falls among older adults increases with age [45, 46]. A correlation between age and BBS score was also mentioned in a study evaluating age and gender-related balance performance in older adults [47]. It should be noted that all participants in this study performed the TUG independently without the use of walking assistance. Another study [48] assessing the risk of falls among nursing home residents showed that although more people in the falls population used walking assistance, the p-value comparing fallers and non-fallers using whether or not walking assistance was nonsignificant. One potential reason for the inconsistency between this result and ours is that our subjects were older adults generally in good physical condition living in the community, with 34% of them using walking assistance. In contrast, in this study, it was nursing home residents with 49% walking assistance use.

Older adults are more likely to be perceived as walking carefully, hobbling, and having short strides. This walking characteristic is usually considered a sign of fear of falling in older people, which aligns with the "cautious gait" concept introduced by physiologists. The introduction of "cautious gait" [49] explains many of the movement patterns of older adults. Cautious gait was considered when gait parameters showed a normal or mildly widened base, shortened stride length, and slowed walking and turning. The results of our experiments show that high-fall-risk individuals have a more cautious gait, a greater step width and a smaller step length to maintain body balance. There is also some evidence for a careful walking strategy in older adults, such as smaller values of CV, Rms, etc., for the angular velocity of yaw rotation in the turn-to-sit phase in high-fall-risk individuals than in low-fall-risk individuals. Many older adults exhibit a timid, reserved, and slow gait as a consequence of the fear of falling [50]. This fear can have detrimental effects on the well-being of older adults, such as social withdrawal, depressive symptoms, and elevated fall risk [51].

Dierick et al. [48] conducted a fall risk assessment in nursing homes using the TUG test. They found that non-fallers exhibited a higher Max-Ang-Yaw than fallers during the first turn, suggesting that older adults with a higher fall risk had a lower turning velocity. Our results showed that a similar phenomenon was found during the turn-to-sit phase (high-risk individual: 122.25 ± 33.24, low-risk individual: 161.28 ± 44.40, p = 0.003), but not during the first turn. This finding may be attributed to the differences in the experimental populations and the methodological variation of combining the second turn and the sit phase into one turn-to-sit phase. The second turn involves more cognitive skills than the first turn, as it requires planning the process of sitting down [52]. Older adults with turn difficulties adopted multi-step turns as a compensatory strategy to reduce the complexity and coordination demands of pivoting turns, which were preferred by younger and older adults without turn difficulties [53]. This strategy resulted in a significantly lower Max-Ang-Yaw among high-fall-risk individuals. Multi-step turns involve fewer feedback mechanisms and allow for smaller or slower movements that are easier to execute.

Also, the significance test results showed that the CV-Ang-Pitch during the turn was the most significant parameter in distinguishing between the two populations, and Median-Ang-Pitch during the turn was also significant. Both parameters indicated a tendency to lean forward in the position where the IMU of the high-fall-risk individuals was located during the turn. This is related to the presence of anterior pelvic tilt in the gait of older adults [54]. A study of joint angles in older and younger people showed that older subjects had more anterior pelvic tilt during walking than younger people [55]. In older adults, anterior pelvic tilt may be caused by postural changes caused by weakness in the abdominal and hip muscles or as compensation for fear of instability [56].

Another interesting phenomenon is that high-fall-risk individuals have a greater right keen angle during the walking and turning phases, i.e., less knee flexion. A study [55] showed that knee range of motion was also reduced in older adults (55 ± 5°) compared to younger adults (59 ± 5°) during the gait cycle. Increased knee flexion during walking can produce greater eccentric work [57]. A lower knee flexion angle may indicate impaired eccentric control of the knee, resulting in more difficult balance control and greater susceptibility to falls [58].

4.3 Limitations and future work

This study has several limitations. The sampling frequency of IMU is only 5hz, which may result in a partial loss of information. A greater percentage of elderly females than males took part in the experiment in the community, which is undoubtedly connected to the proportion of older adults and the fact that elderly females are more engaged in health-related activities. Some preconception bias could result from this systematic erroneous imbalance in demographic parameters. The limited adoption of depth cameras among the elderly population is another issue that merits consideration.

The results of the study suggest that both IMU and depth camera data contain parameters that are able to significantly discriminate between individuals with high and low fall risk. This finding implies that there may be a complementary role for both IMU and depth camera data in fall risk assessment, and further research is needed to fully understand the potential benefits of combining these two techniques. Additionally, we are interested in exploring the potential role of significant parameters in predicting high-fall-risk individuals in real-world settings. To this end, we will use interpretable machine learning algorithms to verify these parameters' impact on prediction accuracy and gain further insight into the underlying mechanisms behind fall risk.

The high incidence of falls among the elderly population has a significant impact on their quality of life. It is crucial to accurately identify individuals at high risk for falls and implement timely interventions, as well as to continuously monitor balance and gait patterns. Conventional clinical assessment methods, such as the BBS, can be resource-intensive. Our study aims to address this issue by leveraging technology-based solutions that utilize low-cost sensors to accurately classify fall risk in the elderly. We integrated the TUG test with various sensor technologies, including IMU and depth camera, and extracted 142 parameters relevant to fall risk. We evaluated the discriminative potential of these parameters in assessing fall risk in community-dwelling older adults. These parameters can help to identify individuals with a high fall risk and to examine the biomechanical, physiological, or psychological factors that may underlie it. The findings of this study will provide valuable information for designing fall prevention programs for frail older adults.

Acknowledgements

We wish to acknowledge the staff and participants of the four communities in Hong Kong for their cooperation, support, permission, assistance and contribution to this work.

Funding

This work was supported by the Shen-Zhen–Hong Kong–Macao Science and Technology Project Fund [grant no. SGDX20210823103403028], the Departmental Supporting Fund [grant no. P0038546], and the Start-up Fund for RAPs under the Strategic Hiring Scheme [grant no. P0036146] at the Hong Kong Polytechnic University.

Author’s contributions

XW, JC, QZ, MC, JL, HW, LY, KLT, and YZ co-conceived the study. HW, LY, KLT, and YZ secured funding. The investigation was performed by XW, JC, QZ, MC, JL,HW, LY, KLT, and YZ supervised study conduct. Analysis was performed by XW, JC, QZ and supervised by all other authors. XW and JC drafted the manuscript, and all other authors assisted in writing. The final manuscript was approved by all authors.

Availability of data and materials

The datasets generated and analyzed during the current study are not publicly available due to privacy and ethical concerns but are available from the corresponding author upon reasonable request.

Ethics approval and consent to participate

The City University of Hong Kong Research Ethics Committee approved this study (reference number: 3-2020-02-F). Participants gave their written informed consent, before participation and after the purpose of the study was explained to them. They were informed of their right to withdraw from the study at any time of their choice and that all methods were performed in accordance with the relevant guidelines and regulations.

Consent for publication

Not applicable.

Competing Interest

The authors declare no competing interest.

Author’s information

¹School of Public Health (Shenzhen), Sun Yat-sen University, Shenzhen, China; ² School of Design, the Hong Kong Polytechnic University, Hung Hom, Hong Kong; ³Grado Department of Industrial and Systems Engineering, Virginia Polytechnic Institute and State University; Blacksburg, Virginia, USA

Sun R, Aldunate RG, Paramathayalan VR, Ratnam R, Jain S, Morrow DG, et al. Preliminary evaluation of a self-guided fall risk assessment tool for older adults. Arch Gerontol Geriatr. 2019 May;82:94–9.
Rubenstein LZ. Falls in older people: epidemiology, risk factors and strategies for prevention. Age Ageing. 2006 Sep 1;35(suppl_2):ii37–41.
Park SH. Tools for assessing fall risk in the elderly: a systematic review and meta-analysis. Aging Clin Exp Res. 2018 Jan;30(1):1–16.
Meseguer-Henarejos AB, Rubio-Aparicio M, López-Pina JA, Carles-Hernández R, Gómez-Conesa A. Characteristics that affect score reliability in the Berg Balance Scale: a meta-analytic reliability generalization study. Eur J Phys Rehabil Med [Internet]. 2019 Oct [cited 2022 Nov 30];55(5).
Shukla B, Bassement J, Vijay V, Yadav S, Hewson D. Instrumented Analysis of the Sit-to-Stand Movement for Geriatric Screening: A Systematic Review. Bioengineering. 2020 Nov 6;7(4):139.
Buisseret F, Catinus L, Grenard R, Jojczyk L, Fievez D, Barvaux V, et al. Timed Up and Go and Six-Minute Walking Tests with Wearable Inertial Sensor: One Step Further for the Prediction of the Risk of Fall in Elderly Nursing Home People. Sensors. 2020 Jun 5;20(11):3207.
Fiedorová I, Mrázková E, Zádrapová M, Tomášková H. Receiver Operating Characteristic Curve Analysis of the Somatosensory Organization Test, Berg Balance Scale, and Fall Efficacy Scale–International for Predicting Falls in Discharged Stroke Patients. Int J Environ Res Public Health. 2022 Jul 27;19(15):9181.
James K, Schwartz AW, Orkaby AR. Mobility Assessment in Older Adults. Ingelfinger JR, editor. N Engl J Med. 2021 Aug 19;385(8):e22.
Blum L, Korner-Bitensky N. Usefulness of the Berg Balance Scale in Stroke Rehabilitation: A Systematic Review. Phys Ther. 2008 May 1;88(5):559–66.
Shahzad A, Ko S, Lee S, Lee JA, Kim K. Quantitative Assessment of Balance Impairment for Fall-Risk Estimation Using Wearable Triaxial Accelerometer. IEEE Sens J. 2017 Oct;17(20):6743–51.
Hsu YC, Wang H, Zhao Y, Chen F, Tsui KL. Automatic Recognition and Analysis of Balance Activity in Community-Dwelling Older Adults: Algorithm Validation. J Med Internet Res. 2021 Dec 20;23(12):e30135.
Isomoto K, Kushida D. Fall Risk Estimation with Relaxation of the Strong Constraint on Depth Camera Location. In: 2020 59th Annual Conference of the Society of Instrument and Control Engineers of Japan (SICE) [Internet]. Chiang Mai, Thailand: IEEE; 2020 [cited 2022 Nov 30]. p. 397–402.
Lafontaine V, Lapointe P, Bouchard K, Gagnon JM, Dallaire M, Gaboury S, et al. An open vibration and pressure platform for fall prevention with a reinforcement learning agent. Pers Ubiquitous Comput. 2021 Feb;25(1):7–19.
Márquez G, Veloz A, Minonzio JG, Reyes C, Calvo E, Taramasco C. Using Low-Resolution Non-Invasive Infrared Sensors to Classify Activities and Falls in Older Adults. Sensors. 2022 Mar 17;22(6):2321.
Saho K, Hayashi S, Tsuyama M, Meng L, Masugi M. Machine Learning-Based Classification of Human Behaviors and Falls in Restroom via Dual Doppler Radar Measurements. Sensors. 2022 Feb 22;22(5):1721.
Liu J, Zhang X, Lockhart TE. Fall Risk Assessments Based on Postural and Dynamic Stability Using Inertial Measurement Unit. Saf Health Work. 2012 Sep;3(3):192–8.
Subramaniam S, Faisal AI, Deen MJ. Wearable Sensor Systems for Fall Risk Assessment: A Review. Front Digit Health. 2022 Jul 14;4:921506.
Kim KJ, Gimmon Y, Millar J, Brewer K, Serrador J, Schubert MC. The Instrumented Timed “Up & Go” Test Distinguishes Turning Characteristics in Vestibular Hypofunction. Phys Ther. 2021 Jul 1;101(7):pzab103.
Xu T, Zhou Y. Elders’ fall detection based on biomechanical features using depth camera. Int J Wavelets Multiresolution Inf Process. 2018 Mar;16(02):1840005.
Dubois A, Bihl T, Bresciani JP. Automating the Timed Up and Go Test Using a Depth Camera. Sensors. 2017 Dec 22;18(2):14.
Hsu YC, Zhao Y, Huang KH, Wu YT, Cabrera J, Sun TL, et al. A Novel Approach for Fall Risk Prediction Using the Inertial Sensor Data From the Timed-Up-and-Go Test in a Community Setting. IEEE Sens J. 2020 Aug 15;20(16):9339–50.
Yu L, Zhao Y, Wang H, Sun TL, Murphy TE, Tsui KL. Assessing elderly’s functional balance and mobility via analyzing data from waist-mounted tri-axial wearable accelerometers in timed up and go tests. BMC Med Inform Decis Mak. 2021 Dec;21(1):108.
B. AHK, Mollahosseini A, Struemph T, Pace W, Nielsen RD, Mahoor MH. Automatic Measurement of Physical Mobility in Get-Up-and-Go Test Using Kinect Sensor. In: 2014 36th Annual International Conference of the IEEE Engineering in Medicine and Biology Society [Internet]. 2014 [cited 2022 Nov 24]. p. 3492–5.
Kim T, Yu X, Xiong S. A multifactorial fall risk assessment system for older people utilizing a low-cost, markerless Microsoft Kinect [Internet]. In Review; 2022 Apr [cited 2022 Nov 24].
Berg K, Wood-Dauphine S, Williams J i., Gayton D. Measuring balance in the elderly: preliminary development of an instrument. Physiother Can. 1989 Nov;41(6):304–11.
Lusardi MM, Fritz S, Middleton A, Allison L, Wingood M, Phillips E, et al. Determining Risk of Falls in Community Dwelling Older Adults: A Systematic Review and Meta-analysis Using Posttest Probability. J Geriatr Phys Ther 2001. 2017 Jan;40(1):1–36.
Podsiadlo D, Richardson S. The Timed “Up & Go”: A Test of Basic Functional Mobility for Frail Elderly Persons. J Am Geriatr Soc. 1991 Feb;39(2):142–8.
Wu CH, Lee CH, Jiang BC, Sun TL. Multiscale Entropy Analysis of Postural Stability for Estimating Fall Risk via Domain Knowledge of Timed-Up-And-Go Accelerometer Data for Elderly People Living in a Community. Entropy. 2019 Nov;21(11):1076.
Society AG, Society G, Of AA, On Falls Prevention OSP. Guideline for the Prevention of Falls in Older Persons. J Am Geriatr Soc. 2001;49(5):664–72.
Ruiz-Ruiz L, Jimenez AR, Garcia-Villamil G, Seco F. Detecting Fall Risk and Frailty in Elders with Inertial Motion Sensors: A Survey of Significant Gait Parameters. Sensors. 2021 Oct 19;21(20):6918.
Bilesan A, Komizunai S, Tsujita T, Konno A, Graduate School of information Science and Technology, Hokkaido University Kita 14, Nishi 9, Kita-ku, Sapporo, Hokkaido 060-0814, Japan, Department of Mechanical Engineering, National Defense Academy of Japan 1-10-20 Hashirimizu, Yokosuka, Kanagawa 239-8686, Japan. Improved 3D Human Motion Capture Using Kinect Skeleton and Depth Sensor. J Robot Mechatron. 2021 Dec 20;33(6):1408–22.
Xia Z, Zhao J, Xia R. An Improved Depth-Based Camera Model in Binocular Visual System. In: 2021 3rd International Symposium on Robotics & Intelligent Manufacturing Technology (ISRIMT) [Internet]. Changzhou, China: IEEE; 2021 [cited 2022 Nov 30]. p. 278–81.
Higashi Y, Yamakoshi K, Fujimoto T, Sekine M, Tamura T. Quantitative evaluation of movement using the timed up-and-go test. IEEE Eng Med Biol Mag. 2008 Jul;27(4):38–46.
Hellmers S, Izadpanah B, Dasenbrock L, Diekmann R, Bauer J, Hein A, et al. Towards an Automated Unsupervised Mobility Assessment for Older People Based on Inertial TUG Measurements. Sensors. 2018 Oct 2;18(10):3310.
Almajid R, Goel R, Tucker C, Keshner E. Balance confidence and turning behavior as a measure of fall risk. Gait Posture. 2020 Jul;80:1–6.
Beauchet O, Fantino B, Allali G, Muir SW, Montero-Odasso M, Annweiler C. Timed up and go test and risk of falls in older adults: A systematic review. J Nutr Health Aging. 2011 Dec;15(10):933–8.
Janssen WG, Bussmann HB, Stam HJ. Determinants of the Sit-to-Stand Movement: A Review. Phys Ther. 2002 Sep 1;82(9):866–79.
Roos PE, Dingwell JB. Using dynamic walking models to identify factors that contribute to increased risk of falling in older adults. Hum Mov Sci. 2013 Oct 1;32(5):984–96.
Akram SB, Frank JS, Chenouri S. Turning behavior in healthy older adults: Is there a preference for step versus spin turns? Gait Posture. 2010 Jan 1;31(1):23–6.
Lin YT, Lee HJ. Comparison of the Lower Extremity Kinematics and Center of Mass Variations in Sit-to-Stand and Stand-to-Sit Movements of Older Fallers and Nonfallers. Arch Rehabil Res Clin Transl. 2022 Mar 1;4(1):100181.
Lehtola S, Koistinen P, Luukinen H. Falls and injurious falls late in home-dwelling life. Arch Gerontol Geriatr. 2006 Mar 1;42(2):217–24.
Buisseret F, Catinus L, Grenard R, Jojczyk L, Fievez D, Barvaux V, et al. Timed Up and Go and Six-Minute Walking Tests with Wearable Inertial Sensor: One Step Further for the Prediction of the Risk of Fall in Elderly Nursing Home People. Sensors. 2020 Jun 5;20(11):3207.
Schoene D, Wu SMS, Mikolaizak AS, Menant JC, Smith ST, Delbaere K, et al. Discriminative Ability and Predictive Validity of the Timed Up and Go Test in Identifying Older People Who Fall: Systematic Review and Meta-Analysis. J Am Geriatr Soc. 2013;61(2):202–8.
Ketcham CJ, Stelmach GE. Movement Control in the Older Adult [Internet]. Technology for Adaptive Aging. National Academies Press (US); 2004 [cited 2023 Apr 6].
Cuevas-Trisan R. Balance Problems and Fall Risks in the Elderly. Clin Geriatr Med. 2019 May;35(2):173–83.
Viswanathan A, Sudarsky L. Balance and gait problems in the elderly. In: Handbook of Clinical Neurology [Internet]. Elsevier; 2012 [cited 2022 Oct 28]. p. 623–34.
Steffen TM, Hacker TA, Mollinger L. Age- and Gender-Related Test Performance in Community-Dwelling Elderly People: Six-Minute Walk Test, Berg Balance Scale, Timed Up & Go Test, and Gait Speeds. Phys Ther. 2002 Feb 1;82(2):128–37.
Dierick F, Stoffel PL, Schütz G, Buisseret F. High Specificity of Single Inertial Sensor-Supplemented Timed Up and Go Test for Assessing Fall Risk in Elderly Nursing Home Residents. Sensors. 2022 Mar 17;22(6):2339.
Nutt JG, Marsden CD, Thompson PD. Human walking and higher-level gait disorders, particularly in the elderly. Neurology. 1993 Feb 1;43(2):268–268.
Herman T, Giladi N, Gurevich T, Hausdorff JM. Gait instability and fractal dynamics of older adults with a “cautious” gait: why do certain older adults walk fearfully? Gait Posture. 2005 Feb;21(2):178–85.
Drummond FMM, Lourenço RA, Lopes CS. Association between fear of falling and spatial and temporal parameters of gait in older adults: the FIBRA-RJ study. Eur Geriatr Med. 2022 Apr 1;13(2):407–13.
Mirelman A, Weiss A, Buchman AS, Bennett DA, Giladi N, Hausdorff JM. Association Between Performance on Timed Up and Go Subtasks and Mild Cognitive Impairment: Further Insights into the Links Between Cognitive and Motor Function. J Am Geriatr Soc. 2014 Apr;62(4):673–8.
Thigpen MT, Light KE, Creel GL, Flynn SM. Turning Difficulty Characteristics of Adults Aged 65 Years or Older. Phys Ther. 2000 Dec 1;80(12):1174–87.
Osoba MY, Rao AK, Agrawal SK, Lalwani AK. Balance and gait in the elderly: A contemporary review. Laryngoscope Investig Otolaryngol. 2019;4(1):143–53.
JudgeRoy JO, Davis B, Ounpuu S. Step Length Reductions in Advanced Age: The Role of Ankle and Hip Kinetics. J Gerontol A Biol Sci Med Sci. 1996 Nov 1;51A(6):M303–12.
Lee LW, Zavarei K, Evans J, Lelas JJ, Riley PO, Kerrigan DC. Reduced Hip Extension in the Elderly: Dynamic or Postural? Arch Phys Med Rehabil. 2005 Sep 1;86(9):1851–4.
Nagano H, Levinger P, Downie C, Hayes A, Begg R. Contribution of lower limb eccentric work and different step responses to balance recovery among older adults. Gait Posture. 2015 Sep;42(3):257–62.
Jeon W, Whitall J, Alissa N, Westlake K. Age-related differences in stepping stability following a sudden gait perturbation are associated with lower limb eccentric control of the perturbed limb. Exp Gerontol. 2022 Oct;167:111917.

No competing interests reported.

BMCAdditionalfile.xlsx

Download PDF

Journal Publication

published 01 Feb, 2024

Read the published version in BMC Geriatrics →

Editorial decision: Revision requested
06 Dec, 2023
Reviews received at journal
29 Nov, 2023
Reviewers agreed at journal
17 Nov, 2023
Reviews received at journal
17 Oct, 2023
Reviewers agreed at journal
15 Oct, 2023
Reviewers invited by journal
15 Sep, 2023
Editor assigned by journal
15 Sep, 2023
Editor invited by journal
26 Apr, 2023
Submission checks completed at journal
26 Apr, 2023
First submitted to journal
23 Apr, 2023

You are reading this latest preprint version

Identifying Sensors-based Parameters Associated with Fall Risk in Community-Dwelling Older Adults: An Investigation and Interpretation of Discriminatory Parameters

Status:

Journal Publication

Version 1

Abstract

Figures

1. Introduction

2. Methods

2.1 Participants

2.2 Assessment and protocol

2.3 Divide the TUG into subtasks

2.4 Parameters extraction and statistical analysis

3. Results

3.1 Demographic parameters of participants

3.2 Significant parameters extracted from IMU and Kinect

3.3 Logistic regression model

4. Discussion

4.1 General considerations

4.2 Focusing parameters

4.3 Limitations and future work

5. Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1