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-HZ2/view. This App should be validated in a prospective assessment.

Smart devices/mobile phone in patients with epilepsy? A systematic review.

We systematically reviewed the existing literature on the safety of the use of smartphone, mobile phone/Internet, and Wi-Fi by people with epilepsy (PWE), according to the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. Scopus, MEDLINE, and Google Scholar from the inception to April 9, 2021 were searched. These key words were used "epilepsy" OR "seizure" AND "Mobile Phone" OR "Cell Phone" OR "Smartphone" OR "Wi-Fi" OR "Electromagnetic" OR "Radiation." The primary search yielded 7766 studies; 33 studies were related. In total, 19 manuscripts were based on animal/computational studies and 14 articles reported human investigations. Among animal studies, 10 articles suggested detrimental effects by electromagnetic fields (EMFs) on brain function/seizure activity, while nine studies negated this hypothesis. Among human studies, seven studies suggested detrimental effects by EMFs on brain function/seizure activity, while seven studies negated this hypothesis. None of the studies provided a good level of evidence. In one human study, all seven patients with epilepsy and abnormal EEG during the sham exposure, had an increase in the number of epileptic events with exposure to mobile phone radiation. In another study of the detrimental effects of smart technology device overuse among school students, an association was found between reporting seizures and the hours of smart technology device use. While high-quality evidence on the safety of the use of smartphone, mobile phone/Internet, and Wi-Fi in PWE is lacking, prudent use of these technologies, including using wired hand-free sets or other exposure-reducing measures is recommended.