Among the common challenges in healthcare data, knowledge and access, timeliness, and quality are perhaps the most crucial to solve for enabling rapid innovation in a health system. These issues are magnified for artificial intelligence (AI) and machine learning (ML) innovations, given the data requirements to produce thoughtful, safe tools for evaluation in the clinical setting. DIHI and Duke Health have prioritized and progressed infrastructure design to overcome these barriers.
The Duke Health Data Pipeline originated as a prototype at DIHI to transform data from the EHR and other sources into curated data sets for retrospective analysis, and moreover for real-time implementation of statistical models and tools derived from that analysis. The data pipeline infrastructure provides the backbone for translating leading-edge AI/ML science into real-life solutions for care delivery. Read more about our infrastructure and its support of our projects below.
Featured Projects
Innovative predictive model to anticipate steroid-induced hyperglycemia and guide insulin regimens
Identify optimal, individualized treatment policies for insulin dosing in the context of systemic glucocorticoids.
A population health approach to advance care planning in primary care
A population-health based pathway to provide ACP at the Duke Outpatient Clinic.
ED LOS: Predictive analytics to reduce emergency department length of stay for youth with behavioral health disorders
Reducing Emergency Department length of stay through population-based predictive modeling.
Rational coronary care unit triage for stable patients with NSTEMI
Predicting the development of in-hospital complications requiring ICU care in initially stable patients with NSTEMI.