A scalable and personal approach to gait rehabilitation beyond the clinic

• Published in: Expert systems with applications Vol. 285; p. 128090
• Main Authors: Wall, Conor, McMeekin, Peter, Hetherington, Victoria, Morris, Rosie, Vitorio, Rodrigo, Walker, Richard, Godfrey, Alan
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.08.2025
• Subjects: Parkinson’s disease; Personalised music cueing; Real-time gait assessment; Smartphone rehabilitation; Personalised music cueing; Real-time gait assessment; Parkinson’s disease; Smartphone rehabilitation
• ISSN: 0957-4174
• DOI: 10.1016/j.eswa.2025.128090

Falls among older adults highlight the importance of objective and valid gait assessments to inform a customised rehabilitation (retraining) intervention strategy like auditory cueing, a practical technique used to help patients improve their gait via auditory tones. Auditory cueing is a well-established approach to improve gait by providing external rhythmic stimuli that enables the brain to synchronise motor actions with tones to enhance timing, coordination, and stability. However, current cueing approaches are limited, lacking innovation, personalisation and scalability. The adoption of a smartphone (equipped with a plethora of technologies) is a viable option for gait assessment and intervention via cueing in any setting and at scale. Here, we firstly describe a novel application/app that uses smartphone-embedded technology for near real-time gait assessment and provides personalised auditory cueing for targeted gait retraining. Secondly, validity of the app was investigated in 60 healthy adults; 30 young healthy subjects (YHS) and 30 older healthy subjects (OHS) by examining gait characteristics derived from the smartphone-based approach with and without auditory cueing compared to gait characteristics from a gold-standard wearable reference (Opal ADPM). The smartphone app showed strong validation against the reference standard before and after cueing, with high Pearson correlation coefficients ranging from 0.862 to 0.996 and intra-class correlation coefficients from 0.865 to 0.997 across spatiotemporal gait characteristics of step time, stride time, stance time, swing time, cadence, stride length, and gait speed. The study also compared personalised auditory cueing modalities (metronome, instrumental music, and vocal music) adjusted to + 10 % of a participants baseline cadence. Vocal music (i.e., original song with vocals) cueing proved to be the most effective and well-adhered modality, achieving an average cadence increase of 8.40 % in the YHS group and 7.96 % in the OHS group, closely aligning with the target 10 % cadence goal. It also led to notable improvements in stride length, with increases of 7.82 cm (standardised response mean, SRM: 0.71) in the YHS group and 7.22 cm (SRM: 0.91) in the OHS group, and in gait speed, rising by 0.14 m/s (SRM: 1.27) and 0.11 m/s (SRM: 1.22), respectively. Furthermore, vocal music cues significantly reduced stride-time coefficient of variation by 0.53 % in the YHS group and 0.60 % in the OHS group, indicating enhanced gait stability. These results suggest a promising, scalable, and personalised strategy for gait assessment and rehabilitation using smartphones.