Home     Projects     People     Publications     Places
Arthritis     Osteoporosis     Spinal Cord Injury     Stroke     Other
Previous Project     Next Project



man wearing glasses mounted with accelerometers

Reliability and Validity of Accelerometric Gait and Balance

Investigators: Eric E. Sabelman, PhD and Carol Winograd, MD

Project Staff: Ruth Yap, MS; Betty S. Troy, MS; Beatrice B. Lee, MS; Deborah E. Kenney, MS OTR; Sandy Dunn Gabrielli; and David L. Jaffe, MS

Project Category: Other - 2000

Objectives: The sense of balance declines with age due to combined vestibular, proprioceptive and visual losses, resulting in impaired mobility and increased risk of injurious falls. A wearable accelerometric instrument has been developed to record movements outside the laboratory, identify motion patterns that accompany loss of balance before a fall occurs, warn the individual of pre-fall behavior, and if necessary, signal that the wearer has fallen. Accelerometry is clinically useful as a diagnostic tool to quantify hitherto qualitative measures of balance, as a biofeedback device during therapy and as a fall-prevention aid - a balance orthosis for fall-prone elderly individuals.

Researchers, clinicians and potential industrial partners familiar with accelerometry have noted the need for rigorously testing its reliability and validity before it can become a routine clinical tool. In this phase of the project, we will establish statistical reliability and validity by comparison of accelerometry with conventional laboratory gait and balance measures using subjects having well-defined gender, age and mobility status.

Methodology: The accelerometric motion detection system consists of two small 3-axis sensors attached to both corners of eyeglass frames to measure head motion, and two sensors above each hip on a belt at the waist, along with a self-contained data acquisition package. An infrared remote control is used to command the wearable unit, so the wearer is unencumbered by cables. In a typical test, subjects perform tasks derived from qualitative balance assessment protocols, including: stand eyes open, then closed; ascend stairs, turn, then descend; rise from and sit in chair; normal walk 10 m; tandem walk 3 m; walk over obstacles 1 m apart.

Among the factors that affect consistent interpretation of accelerometric data are: (1) mechanical and electronic stability, (2) installation and calibration, (3) test-retest reliability, (4) accuracy vs. accepted techniques, (5) presentation of results derived from accelerometric data in a form comparable to currently accepted measures. Subjects are stratified by gender and age decade (60s, 70s, 80s); all have stable mobility status. Overall differences between initial and retest values and differences between accelerometric and conventional measures are compared using appropriate statistics.

Progress: Local collaborators have used the method for clinical diagnosis of veterans with peripheral neuropathy. Non-local researchers have studied fatigue in walking by the elderly and standing balance in a variety of subjects, including children with cerebral palsy . Collaborations have been established for developing special-purpose devices for identifying and preventing lateral falls likely to result in hip fracture, for measuring effects of training in Tai Chi Chuan on balance, and for combined head- and eye-tracking for vestibular research.

Subject recruitment and characterization - We have enrolled and tested on two or three occasions 120 able-bodied subjects in the spectrum between fit and deconditioned.

Test protocol - To reduce the number of subjects needed, we are simultaneously collecting hip, knee and ankle angle data using Penny & Giles goniometers strapped to the subjects' legs. Balance-impaired subjects are asked to perform a shortened series of tasks, as are subjects at test sites other than the RR&D Center. We have added other Palo Alto VA divisions (San Jose, Livermore, and Menlo Park) and senior and recreation centers as possible test sites. Although the full set of tasks takes only 30 minutes, due to set-up and removal of sensors, most subjects are in the test room for about 90 minutes.

1998 Project Description

Funding Source: VA RR&D Merit Review