Measurement of Stand–Sit and Sit–Stand Transitions Using a Miniature Gyroscope and Its Application in Fall Risk Evaluation in the Elderly

Abstract

Abstract—A new method of evaluating the characteristics of postural transition (PT) and their correlation with falling risk in elderly people is described. The time of sit-to-stand and stand-to-sit transitions and their duration were measured using a miniature gyroscope attached to the chest and a portable recorder placed on the waist. Based on a simple model and the discrete wavelet transform, three parameters related to the PT were measured, namely, the average and standard deviation of transition duration and the occurrence of abnormal successive transitions (number of attempts to have a successful transition). The comparison between two groups of elderly subjects (with high and low fall-risk) showed that the computed parameters were significantly correlated with the falling risk as determined by the record of falls during the previous year, balance and gait disorders (Tinetti score), visual disorders, and cognitive and depressive disorders ( 0.01). In this study, the wavelet transform has provided a powerful technique for enhancing the pattern of PT, which was mainly concentrated into the frequency range of 0.04–0.68 Hz. The system is especially adapted for long-term ambulatory monitoring of elderly people.

Introduction

FALLS are the most common type of home accidents among elderly people and are a major threat to their health and independence. Thirty-two percent of a sample of communitydwelling persons 75 years and older, fell at least once a year. Among them, 24% sustained serious injuries. In addition, falling can dramatically change an elderly person’s self-confidence and motivation, affecting their ability to function independently. Considering the growing proportion of old people (over 75) in the populations of industrial countries, falls will be one of the major problems of this important part of the population. In 2050, 16.4% of the world population and 27.6% of theEuropean population are projected to be 65 years and above, and in 14 countries, including nine European countries, more than 10% of the total population will be 80 years or older. Most cases of falls sustained by elderly people appear to result from the cumulative effect of multiple specific disabilities. Among these, balance and gait disorders play a key role. Evaluating the risk of falling is important because it enables the provision of adapted assistance and of taking preventive measures with subjects deemed at risk of falling. The risk of falling is generally evaluated by using questionnaires with their associated problems of subjectivity and limited accuracy in recall. Risk of falls can also be evaluated by clinical and functional assessment including posture and gait, independence in daily life, cognition, and vision. However, no simple objective method is available. Nyberg and Gustafson reported that many falls in stroke patients occur during activities in which they change position (e.g., standing up, sitting down, or initiating walking). Although there are some studies about the relationship of postural transition (PT) duration (TD) with the risk of falls, the methods of investigation used have serious limitations since they consist of performing tests under the constraint of a laboratory and instruments such as a force-platform. Previous studies show the application of ambulatory system in fall risk evaluation during short time monitoring of gait using two or more sites of sensor attachment on body segments. Although such a system has technically the potential to be used for long-term monitoring, the number of sensors and their attachment reduce their applicability during daily life situations. Furthermore, using footswitches to detect the gait events has several limitations in pathological cases. Actually, a simple system using only one sensor, attached as a collar or a wristband is generally preferred in this situation.

Method

A. Subjects

Eleven community-dwelling elderly subjects (six female, five male, age 79 6 years) were studied. Exclusion criteria were cognitive disorders interfering with comprehension or execution of the study, inability to stand up from a chair without help or to pick up an object on the floor, regular use of a walking aid other than a cane, recent ( 3 months) surgery, or existence of an unstable medical condition. Written informed consent was obtained from all the subjects, and the ethics committee of the Faculty of Medicine of Geneva approved the protocol of the study. For each subject, a medical history was obtained by the study physician, including history of falls during the preceding year and medication used. A fall was defined as a sudden, unintentional change in position causing an individual to land at a lower level on an object, the floor, or the ground other than as consequence of sudden onset of paralysis, epileptic seizure, or overwhelming external force. To assess balance and gait disorders, a standardized mobility test was performed according to Tinetti. This assessment reflects the position changes and gait maneuvers used during normal daily activities. The balance and gait disorders were characterized as absent or insignificant, discrete or marked according to the number of abnormalities observed in the mobility test and scored accordingly (Table I). A fall risk score based on known risk factors for falls was then assigned to each subject and described in Table II. The risk factors considered for the fall risk score were a history of falls during the preceding year, balance and gait disorders assessed as previously described, visual, and cognitive and depressive disorders clinically evaluated by the study physician. Table III shows the characteristics of the subjects and their fall risk score. Subjects were divided into two groups: subjects at high risk of falls (high fall-risk) and subjects at low risk of falls (low fall-risk) based on whether their fall risk score was 5 or 5, respectively. During recording, each subject performed different activities involving PTs (such as SiSt, StSi) using different types of chairs (standard wooden chair, armchair, and upholstered chair) with and without the use of armrests and dynamic activities (walking). The measurement protocol is presented in Table IV. For each test, the subjects were asked to sit down and stand up three times.

Gyroscope and Measuring Device

Trunk tilt, corresponding to the angle between the vertical axis and anterior wall of the subject’s thorax, is required for successful identification of body PT. In order to estimate , a piezoelectric gyroscope (Murata, ENC-03J, 400 /s) was attached with a belt in front of the sternum. The gyroscope consists of a vibrating element coupled to a sensing element, acting as a Coriolis sensor. The Coriolis Effect is an apparent force that arises in a rotating reference frame and is proportional to the angular rate of rotation. The principle of operation of a piezoelectric gyroscope is the measurement of the Coriolis acceleration, which is generated when a rotational angular velocity is applied to the oscillating piezoelectric bimorph. The sensor’s low power consumption (4 mA), makes it suitable for ambulatory monitoring applications, where battery power is used. The signal from the gyroscopewas amplified and low-pass filtered (cutoff frequency 17 Hz) to remove electronic noise. The gyroscope and the conditioning electronics were placed in a small box (25 25 15 mm ). The sensor was calibrated by applying various known angular rotations to the sensor

C. Reference Method

A standard motion analysis system (Vicon, Oxford Metrics, U.K.) was used as the reference (gold standard). Five infrared cameras were arranged around the subject and four retro reflective markers were placed on the subject, one on the subject’s left acromion, one on the subject’s sternum and two on the belt holding the kinematic sensor, one on each side of the sensor. This marker arrangement enabled accurate three-dimensional (3-D) measurement of movement of the chest (Fig. 1) and consequently to detect the time of PT occurrence. The sampling frequency of the motion analysis system was 50 Hz. The actual trunk tilt and displacement of the gyroscope were calculated from the 3-D components of the markers

D. Postural Transition Estimation Using Wavelet Transform

Fig. 2 shows that during both SiSt and StSi transitions, there are two phases, initially a leaning forward phase followed by a leaning backward phase. In order to estimate the time at which the PT event occurs, first the trunk tilt was calculated by integrating the gyroscope signal. The time of postural transition ( ) was considered as the time corresponding to the minimum point (trough) of the calculated . The postural TD, was determined by estimating the time interval between the beginning of the leaning forward phase of the transition to the end of the leaning backward phase. The estimation consisted of detection of the peaks in the function before and after . Considering these two peaks, respectively, by and , and their corresponding time by

One of the disadvantages of integrating the gyroscope signal is the introduction of drift into the measured signal, through the integration process. In order to cancel this drift and to eliminate noise from other sources, such as movement artifact noise, the discrete wavelet transform (DWT) based on Mallat’s algorithm, was used as the filtering method. Wavelet transform was introduced by Morlet and Grossmann as a method of improving the relationship between time and frequency resolu- (a) (b) Fig. 2. (a) Trunk tilt corresponding to the angle  between subject’s anterior wall of his/her thorax trunk and vertical surfaces. (b) PT is detected from the pattern of sin() and a known threshold and its minimum peak is assumed as the t . TD is defined as the time interval from beginning of tilt down (P ) until the end of tilt back (P ) during a sit–stand (vs. stand–sit) transition. P and P are estimated by detecting the maximum peaks before and after t . tions in signal analysis, which makes this technique especially suitable for the analysis of nonstationary signals such as human motion signals. An important aspect of trunk-tilt signals is that the information of interest is often a combination of features that are well localized in the time and frequency domains. Wavelet transform allows analyzing the signal in both the time and frequency domains, thus, to handle events that can be at opposite extremes, in term of their time-frequency localization.

III. RESULTS

Fig. 6(a) shows a typical obtained from the integral of the gyroscope signal during several SiSt and StSi transitions. This figure shows a typical problem due to integration drift, which is canceled after DWT [Fig. 6(b)]. In addition, superfluous peak, which does not belong to transitions, can be observed in Fig. 6(a). These peaks are caused by movement artifacts during transition. In fact, in some cases, especially in subjects who have difficulty in rising from a chair, an oscillatory movement is superimposed on the measured signal. Since PT and TD detection is based on peak detection, the presence of these peaks can produce some errors in transition detection or TD estimation. As shown in Fig. 6(b), these peaks were reduced by the DWT while the true transitions were significantly enhanced.

Fig. 7 compares six PTs, a typical pattern of the displacement, and obtained with the Vicon system and derived from the gyroscope signal. These results show that the TDs estimated by the gyroscope correspond to the true transition period observed with the Vicon system. There is a close agreement between the two systems, the coefficient of correlation between the obtained from the gyroscope and that obtained from the Vicon system varies between 0.90 and 0.99 depending on the test performed.

IV. DISCUSSION AND CONCLUSION

In this paper, a new method for the detection and measurement of the duration of PT has been presented. Based on a simple model and an appropriate DWT, three parameters (i.e., and ) for evaluating the falling risk have been provided. The measuring system is very simple to use, since it is based on a light recorder (300 grams) and only one miniature sensor attached on the chest. The device interferes minimally with the subjects performing their usual activities. Therefore, it should allow the monitoring of the subjects in their daily environment and provides information that is probably as close as possible to the “real-world,” in contrast with other systems that require laboratory settings to perform adequately. The integration of the sensor and the recorder in the same module is suggested to provide a system especially adapted for ambulatory instrumentation. Moreover, the gyroscope can be attached anywhere on the chest (as long as we have the same angular velocity). There are many advantages to the use of gyroscopes instead of other kinematic sensors such as accelerometers. First, unlike the accelerometer, the gyroscope can be attached anywhere to a body segment as long as its axis is parallel to the mediolateral axis: the angular rotation is still the same along this segment. Tong and Granat have shown that the signals from different gyroscopes at different attachment site are almost identical. Second, the angular velocity is less noisy than acceleration since acceleration is the derivative of velocity and involves higher frequency components. Finally, there is no influence of the gravity acceleration on the recorded signal. Gyroscopes however, do have some weaknesses, the piezoelectric gyroscope is more delicate to use than an accelerometer and it is more sensitive to temperature and shock due to the mechanical fastening of the vibrating piezo beam inside the sensor’s case. In addition, powerful signal processing and filtering are necessary to cancel drift and movement artefact in the gyroscope signal. Nevertheless, this was accomplished in this study by using the wavelet transform. The use of wavelet transform was particularly important in ensuring a good time resolution in finding PT events and its duration time.

By checking the values of the estimated parameters on a daily basis and by monitoring their change over time, the device could be used as a promising tool in home health care in the elderly by providing objective figures of the mobility of both healthy elderly people and elderly people suffering from specific conditions.

References

J. Askham, E. Glucksman, P. Owens, C. Swift, A. Tinker, and G. Yu, “Home and leisure accident research: A review of research on falls among elderly people,” Age Concern Institute of Gerontology, King’s College, London, U.K., 1990.
2.M. E. Tinetti, M. Speechley, and S. F. Ginter, “Risk factors for falls among elderly persons living in the community,” N. Eng. J. Med., vol. 319, pp. 1701–1707, 1988.
3. M. V. Georges and J. Perreault, “Projections demographiques pour le Canada, les province et les Territoires 1984–2006,” Ministere des Approvisionnements et Services, Ottawa, ON, Canada, 1985.
4. D. Oliver, M. Britton, P. Seed, F. C. Martin, and A. H. Hopper, “Development and evaluation of evidence based risk assessment tool (STRATIFY) to predict which elderly inpatients will fall: Case-control and cohort studies,” BMJ, vol. 315, pp. 1049–1053, 1997.
5.E. Diczfalusy, “The demographic revolution and our common future,” Maturitas, vol. 38, pp. 5–15, 2001.
6. M. E. Tinetti, T. F. Williams, and R. Mayewski, “Fall risk index for elderly patients based on number of chronic disabilities,” Amer. J. Med., vol. 80, pp. 429–434, 1986.
7. S. R. Cummings, M. C. Nevitt, and S. Kidd, “Forgetting falls: The limited accuracy of recall of falls in the elderly,” J. Amer. Geriatr. Soc., vol. 36, pp. 613–616, 1988