Artificial Intelligence in Subarachnoid Hemorrhage

Overview

The overall aim of this study is to, with the help of computer/data scientist and machine learning processes, analyse collected heart rate variability data in order to evaluate whether specific patterns could be found in patients developing delayed cerebral ischemia after subarachnoid hemorrhage.

Patients with aneurysmal subarachnoid haemorrhage (aSAH), develop delayed cerebral ischemia (DCI) in about 30% of the cases. DCI is associated with increased mortality, persistent neurological deficit as well as impaired quality of life. It would benefit both patients and society to decrease these neurological injuries. One clinical problem is that the diagnosis of cerebral ischemia in SAH patients often is delayed due to limitations in monitoring abilities. When detected, the neurological damage often turns out to be irreversible. Several studies have used univariate and multivariate logistic regression analysis to identify risk factors for the development of delayed cerebral ischemia (DCI) in patients with subarachnoid haemorrhage. However, these studies are based on data collected about the patients (e.g. age, gender), and the precision of these statistical models has generally been found to be low. Recently, machine learning algorithms for the prediction of DCI using a combination of clinical and image data have also been evaluated . However, prediction of DCI does not prevent DCI, to prevent DCI a monitoring system needs to be developed that can warn physicians of imminent risk of cerebral ischemia, making it possible to intervene and prevent cerebral ischemia. Variations in the autonomous nervous system, such as changes in the balance between the sympathetic and the parasympathetic nervous systems, can be detected by using heart rate variability (HRV) monitoring. HRV has been reported as a predictor of poor outcome after traumatic brain injury and stroke, including subarachnoid haemorrhage. However, HRV monitoring for detection of incipient cerebral ischemia has not been thoroughly evaluated. In a study of patients with aSAH, we collected HRV continuously in up to 10 days after admission, but just a small part of the HRV data was analysed off-line. Fifteen of 55 patients developed DCI during the acute phase, and the off-line analyse of HRV showed that the low/high-frequency ratio increased more in patients that developed DCI (Ref). This led us to try to analyse all of the collected HRV with the help of machine learning processes, and a collaboration with computer/data scientists was initiated. The overall aim of this study is to, with the help of computer/data scientist and machine learning processes, analyse collected HRV data in order to evaluate whether specific patterns could be found in patients developing DCI during the acute phase after subarachnoid hemorrhage.

Conditions

Interventions

Eligibility

Locations (1)

Outcomes