Balance and postural stability has been studied for decades, and recently these efforts have incorporated small-size, lightweight, and inexpensive devices that may be worn on the body inside or outside the laboratory. Generally speaking, these inertial measurement units (IMUs) contain three-direction accelerometers, gyroscopic sensors for sensing angular velocity, and magnetometers for measuring orientation. IMUs have been widely used in balance studies of the high fall-risk populations, such as the elderly, people who have chronic disease, and people with impaired balance. Pregnant women constitute another population with high fall risk, but their balance has not been studied substantially using IMUs. Instead, studies of balance in pregnant women have mostly been performed in the laboratory environment with force plate. However, there are more complicated and dynamic tasks of daily living that are best studied outside the laboratory. If balance were to be studied in pregnant women using IMUs, the location of the sensor may be a problem;​ the IMU sensor is usually placed on a belt, near the body’s center of mass. Wearing a sensor on a waist belt, however, may cause discomfort for pregnant women. Thus, the main purpose of this dissertation is to explore an alternative way of using an IMU by placing it a case worn around the neck as a pendant and to test whether a pendant IMU is as sensitive to postural instability as one work on a belt.

In the first study, the design of the pendant case was investigated by testing three different novel pendant cases that could contain an IMU. The three designs differed in their curvature and included:​ a flat case with curvature only at the edges, a slightly curved case, and a deeply curved case. We investigated the possibility that a free-swinging pendant would magnify postural instability and potentially be more sensitive to postural instability than a sensor attached to the body. In order to test this sensitivity, we had research participants perform three standing tasks:​ double-leg stance, tandem stance, and single-leg stance. Each standing task involved two vision conditions:​ eyes open and eyes closed. We assumed that in the double-leg stance with eyes open induced the least postural instability, while the single-leg stance with eyes closed induced the most. In the exploration phase of the pilot study, we found out that the pendant sensor with the slightly curved base was most sensitive to differences in stability across the different standing tasks.

In the second study, we tested the sensitivity of the pendant sensor design identified in the first study to postural instability. We recruited 26 young healthy adults who wore the pendant sensor and an IMU placed on a waist belt during experiments. In order to induce instability, three types of standing tasks were considered:​ double-leg stance, tandem stance, and single-leg stance. Two vision conditions were also involved:​ eyes open and eyes closed in each standing task. We assumed that the tandem and single-leg standing would induce postural instability, and in the eyes closed condition the instability was more obvious. Thus, our hypothesis was that the pendant sensor would be more sensitive to postural instability. The metrics we considered for postural stability were standard deviation of the acceleration, jerkiness, and sway area. We found that the pendant sensor was comparable to a belt-mounted sensor in distinguishing between different conditions of postural instability.

In the third study, a data analysis method called recurrence quantification analysis (RQA) was implemented for the standing tasks. The RQA metrics derived from the RR plots were recurrence rate, determinism, linemax, entropy, laminarity, and trapping time. We found that the RQA method is effective for differentiating between different standing tasks with the either the pendant sensor and the belt-mounted sensor. Experimental results showed that the standard deviation of the acceleration in the superior-inferior (SI) direction is more sensitive to postural instability than other components of the acceleration. The results suggested that SI acceleration should also be explored further in future balance studies that employ with RQA-based metrics. In addition, the pendant-based sensor was found to be better for detection of differences between vision conditions in the double-leg stance, while the lower back sensor detects these differences better in the tandem and single-leg standing conditions.

In the fourth study, a simulated pregnancy belly was worn by subjects during experiments in order to simulate changes to the body’s mass distribution during pregnancy. It was assumed that wearing the pregnancy belly would induce postural instability during double-leg standing and three dynamic tasks:​ walking, ascending and descending stairs, and a lifting task. During the lifting tasks experiments, it was found that the pendant sensor moved away from the body when the subject leaned, so these trials were excluded from the final analysis. As in the previous studies, the pendant sensor’s sensitivity to imposed instability was compared to that of a beltmounted sensor. The results showed that in some cases, the pendant was able to detect simulated pregnancy condition, but in others the lower back sensor with belt worked better in detecting postural instability with simulated pregnancy.

In conclusion, this dissertation is the first study to design and test a novel application for an IMU sensor by placing the sensor inside a free-swinging pendant worn around the neck. We found that in the static standing tasks, the pendant sensor was as sensitive as a belt-worn sensor at detecting instability. In addition, measures based on RQA methodology were found to be effective in identifying conditions of postural instability derived from the pendant’s acceleration in the superior-inferior (SI) direction.. When dynamic tasks were studied, a limitation of the pendant sensor was found during the lifting task when the pendant moved away from the body during trunk lean. Future work should be directed toward refining the pendant case design, and pregnant women should be tested to evaluate sensitivity and the comfort level of the pendant design relative to a belt-worn sensor.