Towards a clinically informed, data-driven definition of elderly onset epilepsy.

Area of Science:

• Neurology
• Epileptology
• Geriatric Medicine

Background:

• Elderly onset epilepsy (EOE) is a distinct patient group with unique disease characteristics.
• Lack of a standardized definition hinders research and understanding of EOE.
• This study addresses the need for a clear definition and understanding of EOE attributes.

Purpose of the Study:

• To establish an empirical, clinically informed definition for elderly onset epilepsy.
• To identify key clinical variables associated with the age of epilepsy onset.
• To enable targeted interventions for elderly epilepsy patients.

Main Methods:

• Prospective cohort database analysis.
• Linear support vector machine learning to predict age of seizure onset.
• Receiver-operating characteristic (ROC) analysis to evaluate predictive accuracy.

Main Results:

• Identified 149 patients (6%) with seizure onset at age 65 or older among 2,449 total patients.
• Fourteen clinical variables demonstrated predictive value (absolute value ≥ 0.25).
• Optimal age threshold for EOE prediction was identified between 65-70 years, with maximized ROC curve area.

Conclusions:

• Proposes an age threshold of 65-70 years for defining elderly onset epilepsy.
• Machine learning identified threshold-specific features, aiding in understanding EOE.
• Validated definition will facilitate future research and targeted interventions for EOE.