Patients are monitored continuously with electroencephalograms (EEGs) in Intensive Care Units (ICUs) worldwide. Recent studies show a large percentage of ICU patients have seizures, brain ischemia, encephalopathy, or other conditions that can be detected early on an EEG, allowing therapy to be initiated promptly.
However, continuous long-term EEG monitoring currently presents two major problems: it must be interpreted manually by physicians, delaying the delivery of results to the caregivers, and those caregivers rely on written reports from these studies, thus inhibiting the ability to view trends over time or forecast when a patient’s condition may deteriorate.
The implementation of an automated, cloud-based system for interpreting long-term ICU EEG data could speed response to changes in patient conditions and improve outcomes.
A team led by Brian Litt, MD, professor in Neurology & Bioengineering, is working to build Pennsieve, an automated, cloud-based platform for ICU EEG interpretation.
When a seizure is detected by the algorithm, a notification will be sent to the patient’s care team immediately, so action can be taken sooner.
Through a series of pilots in the Neurology ICU, our team is currently assessing the best method for communicating notifications to providers, testing algorithm accuracy, and improving provider workflow.
When fully implemented, Pennsieve will:
- Improve accuracy in overall seizure detection by the EEG Lab
- Reduce time from seizure onset to treatment
- Improve patient outcomes and decreased length of stay
- Improve provider workflow through patient prioritization
In addition, the integration of data analytics into care delivery pathways where monitoring is used has the potential to be expanded to areas outside of EEG monitoring.
Applied just in the ICU, the immediate opportunity for automated seizure detection is $1million in cost savings per year.