Annika Kumar is currently pursuing her MS in the biomedical engineering department at Duke University. She completed her undergraduate studies at the University of Washington, where she majored in bioengineering with the data science option. Throughout her undergraduate studies, Annika developed a strong foundation in computational bioengineering by taking classes that utilized tools like MATLAB and Python to model biological systems. She also gained knowledge in statistics and introduction to machine learning. These courses ignited her passion for computational bioengineering and led her to pursue two summer internships.
During her internships, Annika honed her Python and RStudio skills and gained practical experience working with large datasets and machine learning models. At the Institute for Systems Biology (ISB), she worked with dense data clouds of genomic data, focusing on calculating polygenic risk scores (PRS) for approximately 1,700 individuals. She used Python programs to extract genome data from the Arivale dataset and compute the PRS, and utilized RStudio for visualizing the distributions of risk scores. Additionally, she conducted simulations using Python to assess the impact of data quantity on the PRS, and visualized the results in RStudio.
Her second internship was at the FDA National Center for Toxicological Research (NCTR), where she conducted a study to evaluate the influence of training dataset sample size on the performance of models predicting Ames mutagenicity test results. She worked with quantitative structure-activity relationship (QSAR) models and trained models using five different machine learning algorithms (Logistic Regression (LR), KNN, SVM, Random Forest, and XGBoost), evaluating performance with Matthews correlation coefficients (MCC). This internship provided her with comprehensive experience in conducting a machine learning study, from data cleaning and study design to model training, parameter optimization, and performance evaluation.
Since starting her MS at Duke, Annika has contributed to research in the Big Ideas Lab, focusing on using smartwatch data and machine learning to develop digital biomarkers for conditions like diabetes, COVID-19, and flu. Her work with the Hoffman lab involves applying machine learning models and simulations to analyze image data and study protein interactions. Annika is passionate about pursuing a research career in machine learning for modeling biological systems.
