A multimodal approach to capture post-stroke temporal dynamics of recovery

Objective. Several training programs have been developed in the past to restore motor functions after stroke. Their efficacy strongly relies on the possibility to assess individual levels of impairment and recovery rate. However, commonly used clinical scales rely mainly on subjective functional assessments and are not able to provide a complete description of patients' neuro-biomechanical status. Therefore, current clinical tests should be integrated with specific physiological measurements, i.e. kinematic, muscular, and brain activities, to obtain a deep understanding of patients' condition and of its evolution through time and rehabilitative intervention. Approach. We proposed a multivariate approach for motor control assessment that simultaneously measures kinematic, muscle and brain activity and combines the main physiological variables extracted from these signals using principal component analysis (PCA). We tested it in a group of six sub-acute stroke subjects evaluated extensively before and after a four-week training, using an upper-limb exoskeleton while performing a reaching task, along with brain and muscle measurements. Main results. After training, all subjects exhibited clinical improvements correlating with changes in kinematics, muscle synergies, and spinal maps. Movements were smoother and faster, while muscle synergies increased in numbers and became more similar to those of the healthy controls. These findings were coupled with changes in cortical oscillations depicted by EEG-topographies. When combining these physiological variables using PCA, we found that (i) patients' kinematic and spinal maps parameters improved continuously during the four assessments; (ii) muscle coordination augmented mainly during treatment, and (iii) brain oscillations recovered mostly pre-treatment as a consequence of short-term subacute changes. Significance. Although these are preliminary results, the proposed approach has the potential of identifying significant biomarkers for patient stratification as well as for the design of more effective rehabilitation protocols.