Skip to Main Content

Andrew Taylor, MD, MHS

Associate Professor of Emergency Medicine and of Bioinformatics & Data Science; Director of Artificial Intelligence and Data Science, Emergency Medicine

Research Summary

Dr. Taylor’s chief interest is in applying data science to solve problems in emergency medicine. His lab is involved in a diverse set of projects including: using machine learning, particularly deep learning, for image recognition and predictive analytics; cluster analysis for novel group/phenotype discovery; decision theory for optimal therapeutic pathways; and EHR-driven, outcomes-based research. In addition, he has interests in the computational reproducibility of research and the gamification of the peer-review process.

Extensive Research Description

Richard Andrew Taylor M.D. is Assistant Professor of Emergency Medicine and Director of Clinical Informatics and Analytics. His work focuses on applying data science to various aspects of emergency care. Prior work has included developing high performance prediction algorithms for urinary tract infections, sepsis severity, and hospital admissions; cost-effective analyses for diagnostic imaging, and research in point-of-care ultrasound outcomes. He is currently the PI on several grants supporting the development of better learning systems in healthcare and is a co-investigator on a PCORTF grant creating better data infrastructure for opioid used disorder. He has methodologic expertise in machine learning, databases, and the secondary use of electronic health record (EHR) data for research.

Current areas of research:

Machine learning/Deep learning for predictive analytics– Emergency medicine is a unique and exciting field for the application of predictive analytics. Providers must make numerous decisions (admission/discharge; ordering tests, medications, etc.) in a chaotic environment within a compressed time-frame that can lead to a variety of cognitive errors. Our lab is focused on augmenting this decision process and lessening the cognitive burden of providers through integration of machine learning tools into clinical work-flows. To accomplish this task, we use a variety of methods including deep learning.

Data Mining/Unsupervised Learning– Adoption of EHRs has led to an explosion of secondary data available for research. We use of variety of data science tools to mine EHR emergency medicine data, find novel relationships, and gain better insight into care processes. Our current research is focused on finding low-dimensional representations of ED encounters and using cluster analysis for phenotype discovery.

Discovery of optimal pathways of care through the use of decision analysis– Our work in this area is primarily focused on establishing appropriate testing thresholds and cost-effective clinical pathways for emergency conditions including: aortic dissection, renal colic, trauma, and head injury.

EHR-driven, outcomes-based research– Current work in this area focuses on causal analysis of difficult to randomize interventions in emergency research using observational EHR data. For example, we are interested in examining the effect of point-of-care ultrasound on mortality and other patient-centered outcomes.


Research Interests

Artificial Intelligence; Decision Theory; Medical Informatics; Natural Language Processing; Neural Networks, Computer; Data Mining; Deep Learning; Data Science

Public Health Interests

Health Care Quality, Efficiency; Health Informatics

Selected Publications