Quantitative EEG may predict weaning failure in ventilated patients on the neurological intensive care unit

Abstract

Neurocritical patients suffer from a substantial risk of extubation failure. The aim of this prospective study was to analyze if quantitative EEG (qEEG) monitoring is able to predict successful extubation in these patients. We analyzed EEG-monitoring for at least six hours before extubation in patients receiving mechanical ventilation (MV) on our neurological intensive care unit (NICU) between November 2017 and May 2019. Patients were divided in 2 groups: patients with successful extubation (SE) versus patients with complications after MV withdrawal (failed extubation; FE), including reintubation, need for non-invasive ventilation (NIV) or death. Bipolar six channel EEG was applied. Unselected raw EEG signal underwent automated artefact rejection and Short Time Fast Fourier Transformation. The following relative proportions of global EEG spectrum were analyzed: relative beta (RB), alpha (RA), theta (RT), delta (RD) as well as the alpha delta ratio (ADR). Coefficient of variation (CV) was calculated as a measure of fluctuations in the different power bands. Mann–Whitney U test and logistic regression were applied to analyze group differences. 52 patients were included (26 male, mean age 65 ± 17 years, diagnosis: 40% seizures/status epilepticus, 37% ischemia, 13% intracranial hemorrhage, 10% others). Successful extubation was possible in 40 patients (77%), reintubation was necessary in 6 patients (12%), 5 patients (10%) required NIV, one patient died. In contrast to FE patients, SE patients showed more stable EEG power values (lower CV) considering all EEG channels (RB: p < 0.0005; RA: p = 0.045; RT: p = 0.045) with RB as an independent predictor of weaning success in logistic regression (p = 0.004). The proportion of the EEG frequency bands (RB, RA RT, RD) of the entire EEG power spectrum was not significantly different between SE and FE patients. Higher fluctuations in qEEG frequency bands, reflecting greater fluctuation in alertness, during the hours before cessation of MV were associated with a higher rate of complications after extubation in this cohort. The stability of qEEG power values may represent a non-invasive, examiner-independent parameter to facilitate weaning assessment in neurocritical patients.

Vision Transformer Prediction of Delirium

“…9,10 The shape, amplitude, and oscillation speed help describe the condition and assist with diagnostics. 11,12 See Figure 1. Using EEG for delirium detection was first identified in the 1940's.…”