Abstract
My technical capstone project and STS research paper are connected via the shared interest of how medical technologies shape healthcare decisions. The former focuses on investigating whether certain therapeutics may be used to encourage regeneration in heart tissue. The latter examines how the estimated glomerular filtration rate, eGFR, incorporates a race coefficient that has downstream effects on patient eligibility and access (Diao et al., 2021). While such projects are within different medical contexts, they both address how technical tools influence what clinicians and researchers are able to measure, interpret, and act upon. Rather than being neutral artifacts, this supports the narrative that medical technologies directly shape the judgment, priorities, and outcomes within the U.S. healthcare system. My capstone project, Mechanistic Machine Learning of Cardiac Regeneration, focuses on improving how researchers study the adult heart’s limited ability after a myocardial infarction (MI), commonly known as a heart attack. Adult cardiac cells lose the ability to rapidly divide due to their permanent exiting of the cell cycle, making it difficult for adult cardiac tissue to recover from damage (Lam & Sadek, 2018). My team further developed CycloTran, an image analysis pipeline that uses machine learning to examine microscope images of tissues and identify signs of cell growth and division. This computational method offers quantitative patterns within tissue samples and organizes them into different stages of cell division. We first tested this approach on a sample data set of gastrointestinal tissue as a proof of concept before applying to future application in cardiac tissue. The long-term objective of this project is to create a reliable method to investigate and modulate how and when cardiac cells can re-enter the cell cycle. In turn, this supports potential therapeutics that could transform irreversible damage done from a heart attack into a minor ailment in the future. My STS research paper argues that the race coefficient, eGFR, was not solely a neutral design feature. Rather, I introduce the point that the eGFR was a technological choice with social consequences once embedded in healthcare systems. Using Langdon Winner’s framework that artifacts carry politics, I examine how the race coefficient re-shapes care via calibration, classification, and stabilization (Winner, 1980). I argue that the race coefficient systematically increased the estimated kidney function for Black patients, leading to delays in disease recognition, specialist referrals, and transplant eligibility. Once incorporated into clinical practices, laboratory reporting, and electronic health records, the authority of the race coefficient stabilized as it became a routine and unquestioned factor. Concurrently working on both research topics has strengthened my understanding of the non-technical implications of biomedical engineering beyond performance. My capstone research has provided me with practical experiences on how computational systems develop and affect its stakeholders, while my STS research encouraged me to investigate the assumptions and consequences surrounding such systems. In turn, I urge future engineers to be vigilant in the decision-making process of computational artifacts in medicine that have the potential to guide human interpretation and affect lived experiences.
