GAIT2CARE: A Database for Evaluating the Effectiveness of Two Exercise Programs in Older Adults using Inertial Gait Analysis and Functional Assessments

The GAIT2CARE database contains socio-demographic information, functional assessments, and gait data of older adults, collected before and after an 8-week multicomponent physical exercise intervention, with the goal of evaluating health status and temporal evolution. The dataset includes information from 127 participants, consisting of 85 women (67%) and 42 men (33%), aged between 70 and 93 years (82.36 ± 5,34 years). Participants were divided into two groups according to the type of exercise program followed: Group A (on-site): group-based exercise guided by a specialist at the hospital’s setting (n=63). Group B (app-guided): home-based multicomponent exercise program implemented with remote supervision via the VIVIFIL App providing video instructions, monitored adherence, and allowing chat communication with the healthcare supervisor (n=64). The study followed a pre-post design, with functional and gait assessments performed at two-time visits: at week 0 (before intervention) and at week 8 (after intervention). The level of compliance/adherence at week 8 with the exercise program is also included: null (<20%), poor (20-50%), medium (50-70%) and remarkable (>70%). Functional assessment was collected by 4-meter walking test, Time Up and Go (TUG) test, Short Physical Performance Battery (SPPB), the Fried Frailty Criteria and Falls Efficacy Scale – International (FES-I). Gait data were captured by inertial sensors (IMUs) placed on the feet during walks of 13.07±5.15 minutes duration, which have been analyzed to estimate characteristic gait parameters. Inertial data from foot-mounted IMUs (acceleration (m/s2), angular velocity (rad/s) and timestamps (s)) are included in the database in .csv files for each participant, trial (week 0 and week 8) and for each foot (right foot (RF) and left foot (LF)), to allow researchers to perform other approaches for gait analysis. The complete gait analysis is also included for each participant and trial in .csv files, including the gait parameters estimated for all individual steps. The gait parameters included are: cycle duration (CD) (s), cadence (steps/min), stride length (SL) (m), path length 3D (%SL), path length 2D (%SL), stride velocity (m/s), percentage of swing (%CD), percentage of stance (%CD), percentage of stance subphases (loading, foot-flat, and pushing) (%stance), heel strike pitch (degrees), toe-off pitch (degrees), peak angle velocity (degrees/s), turning angle (degrees), heel range of motion (RoM) (degrees), double support (%CD) and stride length normalized (SL/height). The gait analysis has been conducted following the methodology described in [1]. Of the 127 participants initially registered for the GAIT2CARE project, five participants abandoned the exercise program before it was completed and inertial data from several others were lost due to IMU recording failures, or technical/human problems. Consequently, the inertial and gait analysis files include 93 participants for whom complete and valid inertial data were available, 44 for group A (on-site) and 49 for group B (application-guided). GAIT2CARE is designed to support research on the effectiveness of exercise interventions in older adults, particularly in relation to gait (inertial analysis) and functional status. However, this dataset is also appropriate for extended research on mobility, frailty, fall risk, aging and gait analysis based on foot-mounted inertial sensors. The study was approved by the Research Ethics Committee on Medicinal Products (CEIm) of the Hospital Universitario de Albacete on June 27, 2023 (Reference code No. 2023-071) and has been prospectively registered on ClinicalTrials.gov with identifier NCT06936865 (https://clinicaltrials.gov/study/NCT06936865). [1] L. Ruiz-Ruiz, J. J. García-Domínguez and A. R. Jiménez, "A Novel Foot-Forward Segmentation Algorithm for Improving IMU-Based Gait Analysis," in IEEE Transactions on Instrumentation and Measurement, vol. 73, pp. 1-13, 2024, Art no. 4010513, doi: 10.1109/TIM.2024.3449951.