Force–Velocity Assessment of Caress-Like Stimuli Through the Electrodermal Activity Processing: Advantages of a Convex Optimization Approach

We propose the use of the convex optimization-based EDA (cvxEDA) framework to automatically characterize the force and velocity of caressing stimuli through the analysis of the electrodermal activity (EDA). CvxEDA, in fact, solves a convex optimization problem that always guarantees the globally optimal solution. We show that this approach is especially suitable for the implementation in wearable monitoring systems, being more computationally efficient than a widely used EDA processing algorithm. In addition, it ensures low-memory consumption, due to a sparse representation of the EDA phasic components. EDA recordings were gathered from 32 healthy subjects (16 females) who participated in an experiment where a fabric-based wearable haptic system conveyed them caress-like stimuli by means of two motors. Six types of stimuli (combining three levels of velocity and two of force) were randomly administered over time. Performance was evaluated in terms of execution time of the algorithm, memory usage, and statistical significance in discerning the affective stimuli along force and velocity dimensions. Experimental results revealed good performance of cvxEDA model for all of the considered metrics.