Towards a diagnostic tool for neurological gait disorders in childhood combining 3D gait kinematics and deep learning

Zhengyang Lan, Mathieu Lempereur, Gwenael Gueret, Laetitia Houx, Marine Cacioppo, Christelle Pons, Johanne Mensah, Olivier Rémy-Néris, Abdeldjalil Aïssa-El-Bey, François Rousseau, Sylvain Brochard

Abstract: Highlights•Data-driven diagnostic models require no human intervention or expert knowledge.•The diagnostic tool distinguished different etiologies of pathological gait.•The diagnostic tool determined the onset time of stroke.•Diagnostic accuracy of deep learning-based models ranged from 0.77 to 0.99.•Deep learning extends the use of 3D gait analysis from evaluation to diagnosis.