A scalable step count-based predictor of biological age: development and validation of MoveIt! Age in community-dwelling adults and geriatric rehabilitation inpatients.

Abstract

Measuring biological age typically requires invasive and costly procedures. To address this, the MoveIt! Age Score was developed: a simple, scalable, and interpretable aging clock that predicts biological age using only wearable-derived steps data. MoveIt! Age was trained on steps data from the United States National Health and Nutrition Examination Survey (NHANES), using chronological age, maximum step count, and step count variability to predict PhenoAge, a blood biochemistry biological age score. MoveIt! Age performance was evaluated in two independent cohorts: Mitochondria and Muscle Health in Elderly (MitoHealth; N = 55; healthy young adults or older adults from the Netherlands) and Restoring Health of Acutely Unwell Adults (RESORT; N = 145; geriatric rehabilitation inpatients from Australia). In RESORT, MoveIt! Age was assessed and compared to SenoClock-BloodAge and PhenoAge (hematological aging clocks). Delta age was the predicted biological age minus chronological age. In the NHANES testing dataset, MoveIt! Age demonstrated high predictive accuracy of chronological age (r = 0.97, RMSE = 5.4 years) and was more significantly associated with mortality than PhenoAge. In MitoHealth, delta MoveIt! Age showed differences between young adults and older adults who were normal, healthy, or health-impaired, with MoveIt! Age more significantly associated with muscle NAD