Patient safety events (PSEs), or medical errors, are major impediments to healthcare system safety. Health information technology (HIT) is expected to promote quality of care. Nonetheless, HIT also creates unintended consequences that concern patient safety consolidating a high-quality database of HIT events is essential to understanding their nature. Previous studies demonstrated the potential to use FDA Manufacturer and User Facility Device Experience (MAUDE) database to extract HIT events. In this study, we utilized classic and CNN models to extract HIT events from MAUDE. Both individual and combined models were evaluated on the test set, where the best model identified HIT events with ~90% accuracy and achieved a ~.87 f1 score. This model was capable of identifying HIT events in an HIT-exclusive database and serving as a quality and error check tool during event reporting. Moreover, the strategy of HIT event identification may scale in developing other PSE subtype-specific databases.