Advancing Personalized Hypertension Care through Big Data Science

项目总结。目前高血压（HTN）的治疗模式是反复试验的药物选择

PROJECT SUMMARY. The current hypertension (HTN) treatment paradigm of trial-and-error drug selection has remained essentially unchanged for nearly half a century. Personalizing care has been challenging because patients and clinicians too often lack adequate evidence to inform individual care decisions. But, broad electronic health record (EHR) adoption has created opportunities for using routinely-collected clinical data to inform evidence. Applying principles of causal inference, such data can be used to identify clinical factors that influence observed variation in treatment response and, in turn, incorporate these factors into statistical models for predicting future treatment response for individuals. The unifying theme of this NHLBI K01 proposal is the mentored career development of Dr. Steven M. Smith. This proposal will accelerate his transition to an independent researcher and establish the foundation for achieving his long-term goal of using routinely-collected clinical data to substantially improve the health and wellbeing of patients by personalizing care. Dr. Smith's objective with this project is to better understand real world use of antihypertensive drugs and factors that influence response to such drugs, with the goal of creating prediction models for use in clinical decision support tools to make personalized HTN management recommendations. The specific research aims include characterizing real world antihypertensive drug prescribing patterns and their determinants (Aim 1), identifying treatment effect modifiers for both effectiveness and safety of two common antihypertensive classes, angiotensin-converting enzyme inhibitors (ACE-Is) and thiazide diuretics (Aim 2) and, developing models for predicting response to ACE-Is and thiazide diuretics to maximize antihypertensive efficacy (Aim 3). This work will leverage observational research methodologies with the OneFlorida Data Trust, a statewide repository of longitudinal EHR data on >8 million Floridians. Dr. Smith's training and experience in clinical pharmacy, public/population health, and HTN care ensure the clinical relevance of the project. His previous clinical HTN research experience and background in applied biostatistics, combined with the proposed training incorporating biomedical informatics, pharmacoepidemiology, multilevel modeling, and leadership, ensure the feasibility of this proposed work and his further development. University of Florida resources and infrastructure, including the UF CTSI, the Biomedical Informatics Program, and the OneFlorida Research Consortium, provide an ideal environment for achieving the proposed objectives and Dr. Smith's long-term goals. Dr. Rhonda Cooper-DeHoff will lead a multidisciplinary mentorship team composed of experts in pharmacoepidemiology (Dr. Almut Winterstein), biostatistics (Dr. Matthew Gurka), biomedical informatics (Dr. Bill Hogan), clinical HTN (Dr. Carl Pepine), and leadership (Dr. Anne Libby). The integrated mentored research experience and training will allow Dr. Smith to compete for R01 funding and become an independent clinician-scientist using observational research methods with large-scale EHR data to improve drug therapy selection for individuals.