Methodology for Attention Detection based on Heart Rate Variability

This work proposes a methodology to measure attention. The proposed methodology, based on Heart Rate Variability (HRV), is composed by the following phases: pre-processing, feature extraction and data analytics. During pre-processing stage, the electrocardiogram (ECG) signal is filtered to remove noise from the signal and HRV signal from ECG is computed. In the feature extraction phase are computed the 12 features for HRV signal description based on linear methods. This 12 linear features include both features from time domain and from frequency domain. Data analytics step is responsible to analyze both the spectral power in the high-frequency (HF) band of the HRV signal and low frequency (LF) band. The proposed methodology was tested in a game playing scenario. Such scenario consists of playing game in two distinct circumstances: playing a game with background classic facilitator music, then with annoying music. The analysis of HF and LF parameters revealed a decrease in determined moments of the experience, which is aligned with a study arguing that those parameters decrease in attentional tasks. In this work variations of those parameters were correlated with the players perception of their attention.