Abstract:Infantile epileptic spasms (IESS) are a common and refractory childhood epilepsy syndrome. Adrenocorticotropic hormone (ACTH) is one of the most effective drugs for the treatment of IESS disorder, but not effective in avoiding side effects. Therefore, improving the prognostic management of IESS by objectively predicting ACTH response is essential. To address this, this study divides patients into four groups: Normal, Mild, Moderate, and Severe, based on the effect of ACTH administration in IESS patients. The power spectral density (PSD) and Permutation entropy (PEN) of interictal EEG in IESS patients before drug administration are extracted, and the overall EEG variability between the four groups of IESS patients, as well as the variability between different brain regions, are analyzed using paired samples t-test. The PSD and PEN of five EEG brain rhythms ( $$\delta $$ , $$\theta $$ , $$\alpha $$ , $$\beta $$ , and $$\gamma $$ ) are taken as input, and support vector machine is applied to construct an ACTH drug efficacy prediction model, which achieves 96 $$\%$$ accuracy. This study contributes to the early prediction of ACTH treatment efficacy and can help to develop appropriate treatment.

Interictal EEG Based Prediction of ACTH Efficacy in Infantile Epileptic Spasms