{
  "abstract": "Interictal epileptiform discharges (IEDs) are transient phenomena that are recorded on electroencephalography and can aid the diagnosis of epilepsy. Currently, neurophysiologists must manually identify IEDs in long EEG recordings which can be time consuming and be poorly sensitive due to the transient nature of IEDs. Automated systems based on deep neural networks such as the Persyst system promise to automatically detect IEDs. The Persyst system has previously been shown to be able to automatically detect IEDs with high sensitivity but low specificity. This study will evaluate the Persyst system’s accuracy in detection of IEDs against the reference standard of a trained neurophysiologist in a large cohort of patients from the Walton Centre, Liverpool. Validation in a large, heterogenous cohort of patients, that are separate from the datasets that were used to train Persyst is of paramount importance before the Persyst system can be used clinically and before clinicians trust the model enough to be used in clinical decision making processes.callumcook945@gmail.com",
  "authors": [
    {
      "affiliations": [
        "University of Liverpool"
      ],
      "name": "Cook Callum"
    },
    {
      "affiliations": [
        "University of Liverpool",
        "The Walton Centre"
      ],
      "name": "Michael Benedict"
    }
  ],
  "title": "133 Evaluation of the accuracy of the persyst system in automated detection of interictal epileptiform discharges",
  "uid": "9a0f4736-19a9-577a-b1fc-92ebc7a08f06"
}