INERTIAL SENSOR DYNAMICS, SELECTION AND APPLICATIONS FOR EPILEPTIC SEIZURE DETECTION

In this work, an optimal placement strategy for wearable inertial sensor placement on a human arm is presented with an ultimate goal of using a sensor cluster for epileptic seizure detection. The present study shows that the placement of sensors plays a significant role in achieving high sensing resolution. Employing the commercial package Motion Genesis (TM), a procedure is first developed to simulate the dynamic response when human body is under typical epileptic seizure episodes. The method uses Kane's method to obtain equations of motion and the resulting equations of motion are numerically solved using Matlab (TM) when the system is subjected to a prescribed seizure input. Simulation results give insights into influences of the neurological disorder on sensing in quantitative terms. Based on certain assumptions on the predicted quantitative measures of the response, optimal detection performance based on sensor location and number of sensors is proposed with particular emphasis on sensor resolution and noise. The optimization is carried out employing the genetic algorithm module of Mat lab global optimization toolbox to find the optimal placement of sensors to achieve least sensing noise in calculating the angular acceleration. The results are also verified with those predicted via simulated annealing. The predictions have also been validated via suitable sensitivity analysis to evaluate the efficacy of the method to uncertainties in the biomechanical, geometry as well as sensor noise parameters. The proposed optimal placement predictions are envisaged to be instrumental for the implementation of a wearable inertial sensor cluster.