Background: The diagnostic yield of whole-exome sequencing (WES) varies from 30%–50% among patients with mild to severe neurodevelopmental delay (NDD)/intellectual disability (ID). Routine retrospective reanalysis of undiagnosed patients has increased the total diagnostic yield by 10–15%. Here, we performed proband-only WES of 1065 patients with NDD/ID and applied a prospective, daily reanalysis automated pipeline to patients without clinically significant variants to facilitate diagnoses.Methods: The study included 1065 consecutive patients from 1056 nonconsanguineous unrelated families from 10 multimedical centers in South Korea between April 2018 and August 2021. WES data were analyzed daily using automatically updated databases with variant classification and symptom similarity scoring systems.Results: At the initial analysis, 402 patients from 1056 unrelated families (38.0%, 402/1,056 families) had a positive genetic diagnosis. Daily prospective, automated reanalysis resulted in the identification of 34 additional diagnostic variants in 31 patients (3%), which increased our molecular diagnostic yield to 41% (433/1056 families). Among these 31 patients, 26 were diagnosed with 23 different diseases that were newly discovered after 2019. The time interval between the first analysis and the molecular diagnosis by reanalysis was 1.2 ± 0.9 years, which was shorter in the patients enrolled during the latter part of the study period.Conclusion: Daily updated databases and reanalysis systems enhance the diagnostic performance in patients with NDD/ID, contributing to the rapid diagnosis of undiagnosed patients by applying the latest molecular genetic information.

The early prediction of epileptic seizures is important to provide appropriate treatment because it can notify clinicians in advance. Various EEG-based machine learning techniques have been used for automatic seizure classification based on subject-specific paradigms. However, because subject-specific models tend to perform poorly on new patient data, a generalized model with a cross-patient paradigm is necessary for building a robust seizure diagnosis system. In this study, we proposed a generalized model that combines one-dimensional convolutional layers (1D CNN), gated recurrent unit (GRU) layers, and attention mechanisms to classify preictal and interictal phases. When we trained this model with ten minutes of preictal data, the average accuracy over eight patients was 82.86%, with 80% sensitivity and 85.5% precision, outperforming other state-of-the-art models. In addition, we proposed a novel application of attention mechanisms for channel selection. The personalized model using three channels with the highest attention score from the generalized model performed better than when using the smallest attention score. Based on these results, we proposed a model for generalized seizure predictors and a seizure-monitoring system with a minimized number of EEG channels.