Resources

Code and curated polysomnography dataset (10,784 PSGs from 8,044 patients, MGB Sleep Lab 2009-2019) used in Ye et al. 2023 (Sleep) to detect dementia and MCI from sleep EEG features.

iOS-app-based EEG education and gamified learning data: 2,270 participants rated candidate interictal epileptiform discharges (IEDs) on the DiagnosUs app, with 13,262 EEG candidates, survey responses, and learning curves.

Dataset of 13,262 candidate interictal epileptiform discharges (IEDs) from EEGs at Massachusetts General Hospital, scored by 8 fellowship-trained physicians, used in an online test to quantify reader skill in IED recognition.

Prediction of Glasgow Coma Scale scores from Unstructured Electronic Health Records using NLP

Dataset and code for developing and validating the EICANS score, a machine learning–derived EEG metric for automated detection and grading of immune effector cell-associated neurotoxicity syndrome (ICANS) in patients receiving CAR-T cell therapy.

Initial version

This project aims to automatically extract mRS scores for a post-stroke patient population from unstructured electronic health records using natural language processing

A machine learning approach to accurately identify neuroinfectious diseases from clinical notes.

This project automatically extracts NIHSS scores from unstructured electronic health records using natural language processing

Code and data for identifying patients with epilepsy using automated electronic health records.