Understanding the Proteomic and Physiological Effects of Impella and ECMO Devices

Overview

Heart failure occurs when the heart is unable to function efficiently—often caused by the development of an infarct, or an area of dead heart tissue. The extremity of infarct formation correlates with heart failure severity, and patients are often placed on ECMO/impella devices to assess the state of the patient for surgery. However, 50% of patients with these devices die due to complications that result from unknown causes. Additionally, there is no accessible way to predict infarct size and thus evaluate the severity of the disease. Research evaluating whether ecpella (bailout and preemptive) devices influence infarct size is needed. Furthermore, the hemodynamic effects of ecpella are poorly understood. As a result, the goals of this project are to identify biomarkers and signaling pathways involved in impella and ECMO implantation using a proteomic platform as well as use tools in machine learning to identify determinants of infarct size based on hemodynamic data and interpretability. More specifically, we used Ingenuity Pathway Analysis (IPA) to evaluate proteomic differences as well as pathways, diseases, and tox functionalities involved in post vs pre impella, post vs pre ECMO, and post impella vs post ECMO groups. Using hemodynamic reperfusion data of preemptive and bailout samples, Knearest neighbors, Xgboost, and regression models (linear, polynomial, lasso, and ridge) were used to predict infarct size. Five algorithms (the regression models and Xgboost) achieved low error rates as well as high R2 values in terms of predicting infarct size. Through interpretability and analyzing data distribution, it was found that the preemptive ecpella group provided better cardiovascular health compared to the bailout group. Thus, this research helps improve the understanding of the proteomic effects of ECMO and impella as well as the hemodynamic differences.

Competition history

• JSHS 2022 Category not listed

Resources

• National JSHS abstract book ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.