Nero: your secure workspace for team science

Radiology

Stanford researchers are building predictive models using AI and multi-modal radiology imaging data, in conjunction with radiology reports and patient’s clinical data  contained in the Electronic Health Records (EHR). STARR contains over half a petabyte of high quality radiology PACS data from Adulth and Pediatric Hospitals, spanning 5-6 million distinct imaging studies and Electronic Health Record (EHR) data for over 4 million patients.

Stanford researchers require an IRB approval to access identified data and a consultaion with the Research Informatics Center. De-identified PACS data is High Risk and researchers use Nero platform to analyze this data.

Neurology

Approximately 3.4 million people have epilepsy in the United States (CDC link), 3 million adults and 470,000 children.  People with epilepsy can suffer recurring seizures, leading to substantial health, social, and mortality disparities1. Using artificial intelligence (AI) technologies and clinical electroencephalography (EEG) data, Stanford researchers are exploring ways to improve seizure recognition and prediction. Dr. Lee-Messer, who’s clinical focus is pediatric epilepsy, is leveraging machine learning modeling and 34.5 TB of LPCH clinical EEG data in Nero to support research initiatives investigating epilieptic seizure detection and efficient training dataset labeling.  Dr. Lee-Messer’s current  projects are part of a larger goal to automate seizure detection to drastically improve diagnostic capabilities and patient care.

Biomedical Informatics

Stanford’s biomedical informatics community is building predictive models from Electronic Health Record (EHR) data from Stanford Adult and Pediatric Hospitals available in STARR (STAnford medicine Research data Repository). These researchers use natural language processing, text processing, deep learning and other sophisticated machine learning techniques. Stanford researchers require an IRB approval for the identified data and can consultation with the Research Informatics Center to access the identified data. A fully de-identified STARR-Navigator (aka STARR-Nav) database, a High Risk dataset, is now available that contains patient data in OHDSI OMOP Common Data Model. To access this data, researchers start with a Data Privacy Attestation.