NeuroBlu

NeuroBlu aggregates deidentified electronic health record (EHR) data from US mental healthcare providers using the MindLinc EHR system to enable real-world data analyses in mental health research.

Key Features:

• Data Scope: Structured data include sociodemographic characteristics, mental health service contacts, hospital admissions, ICD-9/ICD-10 diagnoses, prescribed medications, family history of mental disorders, Clinical Global Impression–Severity/Improvement (CGI-S/CGI-I), and Global Assessment of Functioning (GAF).
• Natural Language Processing (NLP): NLP tools extract additional information on mental state examinations (MSE) and social/environmental factors from unstructured EHR text.
• Data Security: The dataset is implemented within a trusted research environment (TRE) to provide secure data access and governance.
• Dataset Composition: As of July 31, 2021, the dataset contains records for 562,940 individuals, reporting 48.9% male, mean age 33.4 years, most common diagnoses of substance use disorders, major depressive disorder, and anxiety disorders, and a median follow-up duration of 7 months.

Scientific Applications:

• Epidemiological analyses: Enables generation of real-world evidence (RWE) including studies that have identified increased psychiatric hospitalization risk and reduced antidepressant treatment effectiveness among individuals with comorbid substance use disorders.
• Comparative effectiveness research: Supports evaluation of medication comparative effectiveness using longitudinal EHR data.
• Predictive modeling: Facilitates development of predictive models for treatment response using structured and NLP-derived variables.
• Clinical NLP research and visualization: Advances NLP applications to extract clinical insights from unstructured EHR data and supports development of data visualization approaches for clinical decision support and outcome evaluation.

Methodology:

Natural language processing applied to unstructured EHR text to extract MSE and social/environmental factors, combined with statistical analyses and development of predictive models; data maintained within a trusted research environment (TRE).