Onset Estimation of Delayed Cerebral Ischemia by Detecting Change Points in a Time Series of Heart Rate Variability Parameters

Predicting the severity of patient illness and the risk of sudden change, as well as early detection and seamless response in the event of such an occurrence, are important life-or-death issues in the medical field. Many attempts have been made to address this issue in intensive care units/high care units
$(\text{ICUs}/\text{HCUs})$
, including risk assessment at the time of admission using prognostic scores and constructing an artificial intelligence-based rapid-change prediction system based on data collected from life support systems. We investigated the estimation of delayed cerebral ischemia (DCI) onset, a prognostic complication, based on continuous heart rate variability (HRV) monitoring for approximately 2 weeks and offline change point analysis of 22 HRV parameter time series for five patients admitted to an
$\text{ICU}/\text{RCU}$
with aneurysmal subarachnoid hemorrhage
$(\text{aSAH})$
due to ruptured cerebral arteries and who underwent surgical treatment. Common change points were detected simultaneously for several HRV parameters, and common change points were observed within
$\pm 1$
day of the estimated time of onset of DCI in four patients. Thus, continuous HRV monitoring and online analysis of HRV parameter time series suggested the possibility of the early detection of DCI, a complication that greatly affects patient prognosis.