To select or not to select: evaluating the effectiveness of SeLECT score in predicting the risk of post-stroke...

SeLECT score identifies patients at high risk of post-stroke epilepsy(PSE). This systematic review summarizes the effectiveness of SeLECT and its modified versions in predicting PSE risk.

A systematic search was done from inception to 21st July 2025. Studies using SeLECT score to predict the risk of PSE were included. Two authors independently screened and collected data. Due to the heterogeneity of the studies, a meta-analysis could not be done.

Of 3079 articles, 15 were included. Over 10,000 S patients from eleven countries were studied. The original SeLECT score showed high predictive accuracy for PSE in cohorts from Switzerland, Austria, Germany, and Italy. In China, combining SeLECT with IL-1β increased the area under the curve(AUC) from 0.756 to 0.933. In Turkey, adding the severity of diabetes and leukoaraiosis to SeLECT raised AUC from 0.893 to 0.955. Japanese investigators proposed a modified SeLECT score by including cortical superficial siderosis, increasing the model's accuracy(AUC = 0.84). SeLECT 2.0, SeLECT-ASyS, and SeLECT-EEG are modified versions of SeLECT, incorporating seizure type (status epilepticus vs. short seizure) and EEG findings to enhance PSE risk and mortality prediction. In multicenter cohorts, SeLECT 2.0 showed good discrimination (C-statistic = 0.78), while SeLECT-ASyS outperformed SeLECT in training (C = 0.68 vs. 0.58) and validation cohorts (C = 0.70 vs. 0.50). Among stroke patients without acute symptomatic seizures (ASyS), SeLECT-EEG improved time-to-PSE prediction (C = 0.75 vs. 0.71).

SeLECT is an easy-to-use tool that can predict PSE risk. Modification of SeLECT to include factors such as IL-1B, diabetes, leukoaraiosis, superficial siderosis, status epilepticus, ASyS, and EEG using an algorithmic approach should be investigated to increase the accuracy of predicting PSE risk.