A Multiclass Arousal Recognition using HRV Nonlinear Analysis and Affective Images.

Area of Science:

• Physiology
• Biomedical Engineering
• Data Science

Background:

• Autonomic nervous system (ANS) dynamics are crucial for understanding physiological responses to emotions.
• Heart Rate Variability (HRV) analysis offers insights into ANS function.
• Recognizing arousal states is important for various applications, including mental health and human-computer interaction.

Purpose of the Study:

• To develop and validate a multiclass arousal recognition system.
• To investigate the utility of a new hybrid method, Lagged Poincaré Plot symbolic analysis (LPPsymb), for analyzing complex HRV changes.
• To assess the system's performance in classifying arousal levels during affective visual stimulation.

Main Methods:

• A novel hybrid method, LPPsymb, combining Lagged Poincaré Plot (LPP) and symbolic analysis, was developed.
• HRV series from 22 healthy subjects viewing affective images were analyzed using LPPsymb and standard time/frequency domain methods.
• A quadratic discriminant classifier was trained and validated using a leave-one-subject-out procedure.

Main Results:

• The LPPsymb method effectively captured complex HRV dynamics.
• The pattern recognition algorithm achieved a 71.59% accuracy in recognizing four arousal classes.
• Nine features, reflecting heartbeat complex dynamics, yielded the best classification performance.

Conclusions:

• The proposed LPPsymb method is a promising tool for analyzing HRV during emotional stimuli.
• The developed system demonstrates effective multiclass arousal recognition.
• This approach offers a novel way to understand ANS responses to affective experiences.