Evaluating Age-Related Motor System Changes Using Gait as a Functionally Relevant Marker

Aging causes gait and mobility decline, which often precedes cognitive impairment and dementia.

Biological "age gaps" (the difference between biological and chronological age) are traditionally measured using neuroimaging and electroencephalography, quantified gait metrics have not been used for this purpose. We hypothesize that specific gait metrics can effectively quantify biological aging and predict future cognitive impairment risk.

We analyzed gait data from the Mayo Clinic Study of Aging, a population-based cohort in Olmsted County, MN. Participants without known gait disorder, recruited between 2007 and 2025, were included.

We used these data to develop a machine learning-based age prediction model. This gait-based age prediction model was used to devise the "gait age gap," a quantified metric of gait abnormality defined as the difference between age as predicted by gait performance ("gait age") and chronological age.

We used the gait age gap to evaluate the risk of future cognitive impairment. We analyzed 19,637 walk trials from 5,794 unique participants (49% female), with a mean age of 78 (±8) years.

Gait age was primarily driven by changes in measures of stability. A larger gait age gap was associated with impaired cognitive and functional status.

In longitudinal analysis, each gait age gap year exceeding chronological age increased the risk of converting from a normal to cognitively impaired (CI) state by 7% per year. Gait age gap outperformed baseline velocity in distinguishing normal from CI groups and was associated with future risk of conversion from normal to a CI state hazard ratio = 1.07 (95% CI 1.03-1.13) for gait age gap vs hazard ratio = 0.99 (95% CI 0.98-1.00) for velocity.

The gait age gap is a functionally relevant marker of age-related motor system changes that quantifies deviations from expected gait performance for a given chronological age.

It is associated with a higher risk of future cognitive impairment among cognitively normal individuals. This study includes a relatively homogeneous cohort of predominantly Caucasian participants.

To assess generalizability and the trajectory of motoric aging across sociodemographic groups, these findings need to be replicated in diverse cohorts. Wearable sensor technologies could also make such measures more scalable.