Machine learning applications to differentiate comorbid functional seizures and epilepsy from pure functional seizures

Purpose: We have utilized different methods in machine learning (ML) to develop the best algorithm to differentiate comorbid functional seizures (FS) and epilepsy from those who have pure FS. Methods: This was a retrospective study of an electronic database of patients with seizures. All patients with a diagnosis of FS (with or without comorbid epilepsy) were studied at the outpatient epilepsy clinic at Shiraz University of Medical Sciences, Shiraz, Iran, from 2008 until 2021. We arbitrarily selected 14 features that are important in making the diagnosis of patients with seizures and also are easily obtainable during history taking. Pytorch and Scikit-learn packages were used to construct various models including random forest classifier, decision tree classifier, support vector classifier, k-nearest neighbor, and TabNet classifier. Results: Three hundred and two patients had FS (82.5%), while 64 patients had FS and comorbid epilepsy (17.5%). The "TabNet classifier" could provide the best sensitivity (90%) and specificity (74%) measures (accuracy of 76%) to help differentiate patients with FS from those with FS and comorbid epilepsy. Conclusion: These satisfactory differentiating measures suggest that the current algorithm could be used in clinical practice to help with the difficult task of distinguishing patients with FS from those with FS and comorbid epilepsy. Based on the results of the current study, we have developed an Application (SeiDx). This App is freely accessible at the following address: https://drive.google.com/file/d/1rAgBXKNPW9bmUCDioaGHHzLBQgzZ-HZ 2/view. This App should be validated in a prospective assessment.