Research Project Title:
Doctoring the Note: Using AI to Improve Clinical Care
abstract:As the use of artificial intelligence (AI) proliferates in healthcare, clinical notes are becoming an increasingly common data source for machine learning applications. Prior work has shown that features extracted from notes are highly predictive of patient outcomes and used notes to analyze the relationship between patients and their providers, focusing on patient mistrust and physician disbelief. In this project, we intend to leverage a large, multi-site collection of physician and nursing notes to comprehensively study racial biases in patient treatment and study the impact this has on patient outcomes. Lastly, we hope to create a set of debiased notes that will be useful in clinical practice and serve as a fairer source of input data for future machine learning applications.
I've always been interested in public health, so I'm excited to participate in this SuperUROP because it investigates whether racial biases exist in clinical notes and explores ways in which machine learning can be used to inform fairer clinical practice. Through this experience, I hope to gain a better understanding of topic models and natural language processing in general.