External validation of newly modified status epilepticus severity score for predicting mortality in patients with status epilepticus in a regional hospital in Taiwan

Calibrating Deep Learning Classifiers for Patient-Independent Electroencephalogram Seizure Forecasting

The recent scientific literature abounds in proposals of seizure forecasting methods that exploit machine learning to automatically analyze electroencephalogram (EEG) signals. Deep learning algorithms seem to achieve a particularly remarkable performance, suggesting that the implementation of clinical devices for seizure prediction might be within reach. However, most of the research evaluated the robustness of automatic forecasting methods through randomized cross-validation techniques, while clinical applications require much more stringent validation based on patient-independent testing. In this study, we show that automatic seizure forecasting can be performed, to some extent, even on independent patients who have never been seen during the training phase, thanks to the implementation of a simple calibration pipeline that can fine-tune deep learning models, even on a single epileptic event recorded from a new patient. We evaluate our calibration procedure using two datasets containing EEG signals recorded from a large cohort of epileptic subjects, demonstrating that the forecast accuracy of deep learning methods can increase on average by more than 20%, and that performance improves systematically in all independent patients. We further show that our calibration procedure works best for deep learning models, but can also be successfully applied to machine learning algorithms based on engineered signal features. Although our method still requires at least one epileptic event per patient to calibrate the forecasting model, we conclude that focusing on realistic validation methods allows to more reliably compare different machine learning approaches for seizure prediction, enabling the implementation of robust and effective forecasting systems that can be used in daily healthcare practice.

Impact of frailty, biomarkers and basic biochemical parameters on outcomes of comatose patients in status epilepticus: a single-center prospective pilot study

BACKGROUND

Status epilepticus (SE) is a severe acute condition in neurocritical care with high mortality. Searching for risk factors affecting the prognosis in SE remains a significant issue. The primary study's aim was to test the predictive values of the Clinical Frailty Scale (CFS) and the Modified 11-item Frailty Index (mFI-11), the biomarkers and basic biochemical parameters collected at ICU on the Glasgow Outcome Scale (GOS) assessed at hospital discharge (hosp), and three months later (3 M), in comatose patients with SE. The secondary aim was to focus on the association between the patient's state at admission and the duration of mechanical ventilation, the ICU, and hospital stay.

METHODS

In two years single-centre prospective pilot study enrolling 30 adult neurocritical care patients with SE classified as Convulsive SE, A.1 category according to the International League Against Epilepsy (ILAE) Task Force without an-/hypoxic encephalopathy, we evaluated predictive powers of CFS, mFI-11, admission Status Epilepticus Severity Score (STESS), serum protein S100, serum Troponin T and basic biochemical parameters on prognosticating GOS using univariate linear regression, logistic regression and Receiver Operating Characteristic (ROC) analysis.

RESULTS

Our study included 60% males, with a mean age of 57 ± 16 years (44-68) and a mean BMI of 27 ± 5.6. We found CFS, mFI-11, STESS, and age statistically associated with GOS at hospital discharge and three months later. Among the biomarkers, serum troponin T level affected GOS hosp (p = 0.027). Serum C-reactive protein significance in prognosticating GOS was found by logistic regression (hosp p = 0.008; 3 M p = 0.004), and serum calcium by linear regression (hosp p = 0.028; 3 M p = 0.015). In relation to secondary outcomes, we found associations between the length of hospital stay and each of the following: age (p = 0.03), STESS (p = 0.009), and serum troponin T (p = 0.029) parameters.

CONCLUSIONS

This pilot study found promising predictive powers of two frailty scores, namely CFS and mFI-11, which were comparable to age and STESS predictors regarding the GOS at hospital discharge and three months later in ICU patients with SE. Among biomarkers and biochemical parameters, only serum troponin T level affected GOS at hospital discharge.