Zhang, Jianqiu2023-09-192023-09-192023-02https://hdl.handle.net/11299/257074University of Minnesota Ph.D. dissertation. February 2023. Major: Health Informatics. Advisor: Genevieve Melton-Meaux. 1 computer file (PDF); v, 79 pages.Acute kidney injury (AKI) is a clinical syndrome characterized by a rapid loss of kidney function and commonly seen in the inpatient settings (Bellomo et al., 2012). AKI could be potentially catastrophic and is associated with longer hospital stays, increased readmission rates, and long-term adverse outcomes (Hoste et al., 2018). United States Renal Data System (2022) reported an increasing trend of AKI hospitalizations over the past decade. Despite a large body of research in AKI, there are significant gaps in capturing, recognizing, and assessing the risk of AKI in real-world settings. In this study, we conducted a series of experiments to fill the gaps, with applications of modern health informatics technologies and secondary use of Electronic Health Records (EHR). Our first experiment was designed to address a fundamental part of research, which is the validity of data. We examined the phenotyping performance of various combinations of EHR elements in identifying AKI. We obtained significantly different quality measures (sensitivity:0.486, specificity:0.947, PPV:0.509, NPV:0.942 in the full cohort) of administrative coding from the previously reported ones in the U.S. (Grams et a., 2014). Additional use of clinical notes has been found to increase the AUC in phenotyping AKI, and AKI was better recognized in patients with heart failure, indicating disparities in the coding and management of AKI. Secondly, the COVID-19 pandemic has also introduced new challenges and unknown risks in the development of AKI, and we were particularly interested in the nephrotoxicity of Remdesivir (intravenous route). We analyzed the role of remdesivir and built multifactorial causal models of COVID-AKI by applying causal discovery machine learning techniques. Our models successfully recreated known causal pathways to changes in renal function and interactions with each other and examined the consistency of high-level causal relationships over a 4-day course of remdesivir. Results indicated a need for assessment of renal function on day 2 and 3 use of remdesivir. Lastly, there are significant gaps in the literature regarding the risk prediction of recurrent-AKI among AKI survivors. In this study, we built and compared machine learning models using both knowledge-based and data-driven features in predicting the risk of recurrent AKI within 1-year of discharge. Our results showed that the additional use of data-driven features statistically improved the model performances, with best AUC=0.766 by using logistic regression. Our models identified non-traditional risk predictors, such as ICD_Z94.83 (Pancreas transplant status) and PROC_87799 (infectious antigen detection), which may suggest some risk factors that require more attention for AKI management.enclinical informaticshealth informaticsThesis or Dissertation