<![CDATA[Early prediction of epilepsy onset is crucial for supporting timely clinical intervention and reducing seizure-related complications. However, the high dimensionality and complexity of Electroencephalogram (EEG) and clinical data remain major challenges for conventional predictive models. This study proposes an interpretable hybrid framework integrating Exploratory Factor Analysis (EFA) and Decision Tree classification for early epilepsy onset prediction using public datasets. EFA was employed to reduce 15 observed variables into five clinically meaningful latent factors, which were subsequently used as inputs for the Decision Tree model. . According to experimental data, the suggested framework demonstrated steady classification performance with an accuracy of 88.14% and specificity of 88.73%, while its efficacy in identifying epilepsy beginning cases for early screening is highlighted by its sensitivity of 80.77%.. The latent factor representing clinical neurological abnormalities was identified as the most influential predictor in the classification process. Compared with conventional black-box machine learning approaches, the proposed model provides transparent decision rules and clinically meaningful interpretation while maintaining reliable predictive capability. All things considered, the suggested architecture is a viable path for creating clinically interpretable prediction models that facilitate transparent and trustworthy early epilepsy screening.]]>
Copyrights © 2026
