Autonomous Closed-Loop Biofeedback

Abstract

This chapter is from the forthcoming The Oxford Handbook of Affective Computing edited by Rafael Calvo, Sidney K. D'Mello, Jonathan Gratch, and Arvid Kappas. This chapter defines the core concepts surrounding biofeedback and denotes their relations. Subsequently, a closed-loop human-machine architecture is introduced in which a biofeedback protocol is executed. This architecture is brought from theory to practice via a personalized affective music player (AMP). Regression and kernel density estimation are applied to model the physiological changes elicited by music. The AMP was validated via a real-world evaluation over the course of several weeks. Results show that our autonomous closed-loop biofeedback system can cope with noisy situations and handle large interindividual differences in the music domain. The AMP augments music listening, where its techniques enable autonomous affect guidance. Our approach provides valuable insights for affective computing and autonomous closed-loop biofeedback systems in general. Keywords: autonomous, closed-loop model, biofeedback, affect, music, personalized, physiological changes, validation