Thigh-Worn Accelerometry and Gait Event Detection

The dataset contains laboratory and outdoor gait data from n = 21 healthy adult participants. Participants completed four two-minute walking trials on a treadmill at speeds ranging from 2 to 5 km/h. They also completed a five-minute walk outdoors at their preferred speed on level ground. Vertical ground reaction force (GRF) was measured using force-sensing insoles (loadsol pro, novel GmbH, Germany). Thigh acceleration was measured using an IMU (Axivity AX6; Axivity Ltd., United Kingdom) placed on the lateral side of the dominant leg, midway between the hip and knee. The IMU was attached directly to the skin using medical adhesive tape. Insole force data and IMU data were sampled at 100 Hz. To synchronise the insole and IMU signals, participants performed a series of standardised heel drops, lifting their heels off the ground and dropping them rapidly back down again with straight legs. The peaks in both signals were subsequently used to align the two timeseries. The initial contacts (IC) and final contacts (FC) during walking were identified from the GRF using an adaptive threshold of 25 N + 10th percentile of the GRF during the measurement to account for trials with zeroing errors of the insole. The identified gait events were verified by checking for IC or FC events within 0.5 seconds of each other, as well as for alternating IC and FC events within the data. The GRF and identified gait events were then visually inspected and manually corrected where necessary.