Physiologically-Informed Gaussian Processes for Interpretable Modelling of Psycho-Physiological States

Abstract

The widespread popularity of Machine Learning (ML) models in healthcare solutions has increased the demand for their interpretability and accountability. In this paper, we introduce the Physiologically-Informed Gaussian Process (PhGP) classification model, an interpretable machine learning model founded on the Bayesian nature of Gaussian Processes (GPs). Specifically, we inject problemspecific domain knowledge of inherent physiological mechanisms underlying the psycho-physiological states as a prior distribution over the GP latent space. Thus, to estimate the hyper-parameters in PhGP, we rely on the information from raw physiological signals as well as the designed prior function encoding the physiologically-inspired modelling assumptions. Alongside this new model, we present novel interpretability metrics that highlight the most informative input regions that contribute to the GP prediction. We evaluate the ability of PhGP to provide an accurate and interpretable classification on three different datasets, including electrodermal activity (EDA) signals collected during emotional, painful, and stressful tasks. Our results demonstrate that, for all three tasks, recognition performance is improved by using the PhGP model compared to competitive methods. Moreover, PhGP is able to provide physiological sound interpretations over its predictions.

A Real-Time, Open-Source, IoT-like, Wearable Monitoring Platform

“…Recently, several studies have started to adopt multivariate approaches combining some of these bio-signals and contextual information to infer psychophysiological and physical states [1][2][3], even in contexts beyond the simple laboratory experimental setup (i.e., natural settings) [4,5]. In this scenario, biosignals have been used as important indicators for rehabilitation [6], epileptic seizure prediction [7], sleep stage scoring [8], affective computing [9], and arrhythmia detection [10].…”