Multiscale Entropy and Time-Frequency Analysis of Electrocardiogram and Blood Pressure in Patients With Spontaneous Intracranial Hemorrhage

Spontaneous intracranial hemorrhage is an absolute emergency in the field of neurosurgery, and it is also a devastating event that commonly results in major neurological disabilities or mortalities. Since disease severities and clinical courses vary in each patient, pathophysiological studies and prognostic factors are always worth research. From previous studies, we know that dysregulation of autonomic system plays an important role in intracranial hemorrhage. Hemorrhage itself is associated with sympathoexcitation, and patients who develop rebleeding or infarction complications are found to have an even higher degree of sympathetic storm. Therefore, the degree of autonomic activities seems to be a useful predictor.

Traditionally, sympathetic activities are measured by plasma catecholamine, while parasympathetic activities are hard to measure. In recent decades, the application of engineering in biological fields makes a great breakthrough. Waveform analysis of biological signals, such as electrocardiograms and arterial blood pressure, can indirectly determine autonomic activities. The variabilities of heart rate and blood pressure are subjected to frequency analysis. This generates several dominant frequency bands. High frequency bands (0.15-0.40Hz) are attributed to the effect of parasympathetic nervous system, while, the low frequency bands (0.04-0.15 Hz) are attributed to the effect of both sympathetic and parasympathetic nervous systems.

In this study, all patients with spontaneous intracranial bleedings undergo standard treatment and monitoring. This include electrocardiography, arterial blood pressure, and cerebral blood flow using transcranial Doppler sonography. For those who also have intracranial pressure monitoring, the intracranial pressure are also recorded. All these biological signals are exported for wave form analysis. We use frequency analysis, time-frequency analysis, and multiscale entropy to analyze these data. The results of analyses were also correlated to plasma catecholamine levels, proinflammatory markers, as well as the clinical variables. Our aim is to identify predictors of complications and grave outcomes from these biological signals. We also apply the results for future pathophysiological studies.