RESUMEN
Assessing the motor impairments of individuals with neurological disorders holds significant importance in clinical practice. Currently, these clinical assessments are time-intensive and depend on qualitative scales administered by trained healthcare professionals at the clinic. These evaluations provide only coarse snapshots of a person's abilities, failing to track quantitatively the detail and minutiae of recovery over time. To overcome these limitations, we introduce a novel machine learning approach that can be administered anywhere including home. It leverages a spatial-temporal graph convolutional network (STGCN) to extract motion characteristics from pose data obtained from monocular video captured by portable devices like smartphones and tablets. We propose an end-to-end model, achieving an accuracy rate of approximately 76.6% in assessing children with Cerebral Palsy (CP) using the Gross Motor Function Classification System (GMFCS). This represents a 5% improvement in accuracy compared to the current state-of-the-art techniques and demonstrates strong agreement with professional assessments, as indicated by the weighted Cohen's Kappa ( κlw = 0.733 ). In addition, we introduce the use of metric learning through triplet loss and self-supervised training to better handle situations with a limited number of training samples and enable confidence estimation. Setting a confidence threshold at 0.95 , we attain an impressive estimation accuracy of 88% . Notably, our method can be efficiently implemented on a wide range of mobile devices, providing real-time or near real-time results.
Asunto(s)
Parálisis Cerebral , Aprendizaje Automático , Humanos , Parálisis Cerebral/fisiopatología , Parálisis Cerebral/rehabilitación , Niño , Masculino , Femenino , Algoritmos , Redes Neurales de la Computación , Teléfono Inteligente , Adolescente , Trastornos Neurológicos de la Marcha/fisiopatología , Trastornos Neurológicos de la Marcha/rehabilitación , Trastornos Neurológicos de la Marcha/diagnóstico , Grabación en Video , Análisis de la Marcha/métodosRESUMEN
Multivariable ankle mechanical impedance was estimated in four stroke survivors, in coupled dorsi- plantarflexion and inversion-eversion. We applied external torque perturbation with an ankle robot and used multi-input, multi-output stochastic system identification methods to estimate impedance, in both paretic and nonparetic limbs. Subjects were instructed to remain at rest throughout the four trials performed on each leg. Impedance projected onto the directions of maximum and minimum stiffness was fit to a 2nd order linear model, including inertia, viscosity and stiffness. For most trials, stiffness and damping in dorsi-plantarflexion are increased on the paretic side. However, for two subjects, overall impedance is not increased in the absence of sustained involuntary tonic contraction, registering values comparable to the non-paretic side. Thus, we speculate that the intrinsic properties of the paretic ankle remained unaffected at the evaluated pose. Spasticity (hyperflexive stretch reflex) would have systematically increased stiffness and damping, even in the absence of involuntary contraction. Hence, we speculate that these two subjects did not exhibit spasticity, while the remaining two subjects did, since impedance was increased, with no involuntary tonic muscle contraction. Regarding inversion-eversion, impedance in this direction remained unaffected by stroke. We evaluated two volunteers before and after the application of botulinum toxin. Surprisingly, ankle stiffness was not reduced, but anisotropy was normalized.
Asunto(s)
Tobillo , Accidente Cerebrovascular , Articulación del Tobillo , Impedancia Eléctrica , Humanos , Reflejo de EstiramientoRESUMEN
Locomotion involves complex neural networks responsible for automatic and volitional actions. During locomotion, motor strategies can rapidly compensate for any obstruction or perturbation that could interfere with forward progression. In this pilot study, we examined the contribution of interlimb pathways for evoking muscle activation patterns in the contralateral limb when a unilateral perturbation was applied and in the case where body weight was externally supported. In particular, the latency of neuromuscular responses was measured, while the stimulus to afferent feedback was limited. The pilot experiment was conducted with six healthy young subjects. It employed the MIT-Skywalker (beta-prototype), a novel device intended for gait therapy. Subjects were asked to walk on the split-belt treadmill, while a fast unilateral perturbation was applied mid-stance by unexpectedly lowering one side of the split-treadmill walking surfaces. Subject's weight was externally supported via the body-weight support system consisting of an underneath bicycle seat and the torso was stabilized via a loosely fitted chest harness. Both the weight support and the chest harness limited the afferent feedback. The unilateral perturbations evoked changes in the electromyographic activity of the non-perturbed contralateral leg. The latency of all muscle responses exceeded 100ms, which precludes the conjecture that spinal cord alone is responsible for the perturbation response. It suggests the role of supraspinal or midbrain level pathways at the inter-leg coordination during gait.