Identification of mental stress granularity from long-term ECG recordings using novel complexity analysis

Mental health conditions often manifest as changes in physiological signals and are characterized by specific features within these signals. However, there is limited knowledge about the relationship between long-term physiological time series including ECG and psychopathology. In a complex time series, the waiting time distribution of crucial events has an inverse power law probability density function with a temporal complexity index μ (1 < μ < 3). The diffusion time series generated with such crucial events have a scaling index δ which is equal to μ-1 and 1/(μ-1) for 1 < μ < 2 and 2 < μ < 3, respectively. The ECG dataset was provided by the Korean Advanced Institute of Science and Technology (KAIST). ECGs were collected with the chest strap sensor (Polar H10) from 90 participants over four weeks. Participants rated their daily stress levels on a scale from -3 (not stressed) to +3 (very stressed) at multiple random times each day. Five-minute segments (the average amount of minutes of feeling the scored level of stress) of the labeled ECG time series were preprocessed, and the multiscale modified diffusion entropy analysis (MSMDEA) with time scale factors from 1 to 20, used for quantifying μ and δ of the ECGs. δ for a binarized scale of high and low-stress groups showed significant differences across the temporal scaling factors (p=0.05). The multi-time-scaled complexity scaling index δ for fine-grained full-scale scores of mental stress indicated that more than half of the labeled group pairs were significantly different (p <0.05). These results show that the lower the stress is, the closer the signal is to critical complexity at δ=1 (or μ=2), indicating a healthier physiological and psychological function. The results indicate that for crucial events closer to μ = 2, mental health may be driven by information content from the heart. Hence, ECG crucial event analysis can be an adjunct in precision psychiatry to assess mental health conditions.