# Prognostic value of quantitative and visual electroencephalography in disorders of consciousness: a retrospective study

## TL;DR

This study compares visual and quantitative EEG methods for predicting outcomes in patients with impaired consciousness, finding that combining both with clinical factors improves accuracy.

## Contribution

The study directly compares visual and quantitative EEG for prognosis in disorders of consciousness and evaluates their combined use with clinical factors.

## Key findings

- Visual EEG had moderate predictive accuracy for survival and neurological outcomes.
- qEEG models showed comparable performance to visual EEG but with no significant difference.
- Adding clinical factors significantly improved predictive accuracy for neurological recovery.

## Abstract

Electroencephalography (EEG) is widely used to assess prognosis in patients with disorders of consciousness (DoC). Visual assessments by physicians and quantitative EEG (qEEG) are commonly used; however, only a few studies have directly compared their predictive accuracy. Therefore, in this study, we aimed to compare the prognostic value of visual EEG classification versus that of qEEG-based spectral analysis for survival and neurological outcomes in patients with impaired consciousness.

In this retrospective study, we examined 97 patients with impaired consciousness admitted to the Emergency and Critical Care Center of Fukushima Medical University Hospital between April 2018 and December 2023. Visual EEG grading was performed using a conventional grading system based on established criteria. Receiver operating characteristic (ROC) curves were used to compare predictive performance. Multivariate logistic regression models were developed incorporating qEEG and clinical prognostic factors (Scarpino score, rehabilitation status, and age). The incremental predictive value of clinical variables was assessed using DeLong’s test.

Visual EEG assessment showed moderate predictive accuracy [area under the curve (AUC) = 0.77 for survival; 0.677–0.725 for neurological outcomes]. qEEG-based models showed comparable performance to visual EEG classification, with slightly higher AUC values that were not statistically significant. The addition of clinical factors significantly improved predictive accuracy, particularly for neurological recovery (AUC improved from 0.729 to 0.936; P < 0.001).

Combining qEEG features and clinical prognostic factors provided a comprehensive approach for outcome prediction in patients with DoC. These findings support the potential of a multimodal prognostic framework integrating objective EEG metrics and physician-derived evaluations, although further prospective validation is required.