EEG/EMG based Architecture for the Early Detection of Slip-induced Lack of Balance

In this paper, we propose the preliminary version of a novel pre-impact fall detection (PIFD) strategy, optimized for the early recognition of balance loss during the steady walking.The technique has been implemented in a multi-sensor architecture aiming to jointly analyzes the muscular and cortical activity. The physiological signals were acquired from 10 electromyography (EMC) electrodes on the lower limbs and 13 electroencephalography (EEG) sites all along the scalp.Data from the EMGs are statistically treated and used both to identify abnormal muscular activities and to trigger the cortical activity assessment. The EEG computation branch evaluate the rate of variation of the EEG power spectrum density, named m, to describe the cortical responsiveness in live bands of interest. Then, a logical conditions network allows the system to recognize the loss of balance induced by the slippage, by considering both the evaluated muscular parameters and the cortical ones.Experimental validation on six adults (supported by the motion capture system) showed that the system reacts in a time compliant with the fall dynamic request (403.16 ms), ensuring a competitive detection accuracy (Sensitivity =93.33%, Specificity=99.82 %)