Novel Algorithmic Tools for Improving Health Outcomes in Primary Care

Stanford University, Stanford CA

Investigators

Abstract

PROJECT SUMMARY: Primary care is often the last area in medicine to benefit from new technologies. Additionally, most machine learning tools are created without relevant data from primary care practices. Algorithms built in hospital settings are typically not directly applicable to primary care. These issues have been highlighted as reasons why machine learning may not drive meaningful changes for primary care. Leveraging large-scale, high-quality primary care health data is critical to create genuine solutions for improving health outcomes with algorithm development. Our innovative approach to these challenges is to create a first-of-its-kind overarching algorithmic framework for primary care. In the initial phase, we will focus on intervening on the data in order to generate counterfactual outcomes to represent a desired equilibrium. The second stage builds novel penalized regression estimators to enforce constraints for prediction. Our goal is to create reusable tools that advance the provision of health care to benefit all populations in primary care. We will accomplish this by developing generalizable methodology that follows a rigorous pipeline for algorithms. Our specific aims are to: (1) develop and test novel data intervention methods that rely on microsimulations for generating counterfactual outcomes, (2) develop and test new penalized regression approaches, (3) test the performance of the new algorithmic framework for a high-impact primary care application in chronic kidney disease and (4) create open-source computational tools, tutorial vignettes, and a synthetic data resource for reproducible research and dissemination. The proposed research will yield a statistically innovative reusable algorithmic framework unifying data intervention and penalized regression with robust testing in a chronic kidney disease study of quality of care. This primary care application will leverage rich registry data collected in usual care settings across the United States from multiple payers. Our approach centers robustness with rigorous methodological design, including comparisons to alternative existing estimators and standard practice in comprehensive simulation studies and national, real-world registry data. Addressing health outcomes in primary care—a hub of continuous, coordinated care—has the potential for substantial impact on improving public health via the health care system. The broad applicability of our framework and creation of reusable computational tools will facilitate deployment in many practical settings.