Multiple gait parameters derived from iPod accelerometry predict age-related gait changes

- Trunk accelerations were obtained in 59 healthy adults (aged 18-75) with an iPod.
- A combination of multiple gait variables indicate natural age-related changes.
- Changes in gait over the adult life span are quantified using the iPod.
- Variables associated with age have a good accuracy to classify young and old adults.
- Quantified changes provide normative data to interpreted pathological gait changes.

IntroductionNormative data of how natural aging affects gait can serve as a frame of reference for changes in gait dynamics due to pathologies. Therefore, the present study aims (1) to identify gait variables sensitive to age-related changes in gait over the adult life span using the iPod and (2) to assess if these variables accurately distinguish young (aged 18-45) from healthy older (aged 46-75) adults.MethodsTrunk accelerations were recorded with an iPod Touch in 59 healthy adults during three minutes of overground walking. Gait variables included gait speed and accelerometry-based gait variables (stride, amplitude, frequency, and trajectory-related variables) in the anterior-posterior (AP) and medio-lateral (ML) directions. Multivariate partial least square analysis (PLS) identified variables sensitive to age-related differences in gait. To classify young and old adults, a PLS-discriminant analysis (PLS-DA) was used to test the accuracy of these variables.ResultsThe PLS model explained 42% of variation in age. Influential variables were: mean stride time, phase variability index, root mean square, stride variability, AP sample entropy and ML maximal Lyaponov exponent. PLS-DA classified 83% of the participants correctly with a sensitivity of 83% and specificity of 71%.DiscussionContrary to the frequently reported high gait variability observed in old adults with frailty and fall history, the present study showed that younger compared with older healthy adults had a more variable, less predictable and more symmetrical gait pattern. A model based on a combination of variables reflecting gait dynamics, could distinguish healthy younger adults from older adults.