Experience

Health Data Science Fellow

Insight Data Science, Silicon Valley, Sep 2020-current

  • Consulted with physicians in the Cutaneous Oncology Research Lab at OHSU to predict cancer outcomes in Merkel Cell Carcinoma to inform decision-making and improve clinical care.
  • Applied classification algorithms to identify patients with and without metastasis from tumor characteristics; results estimate that 35% of low-risk patients can avoid an invasive biopsy.
  • Developed a web application in Streamlit that deploys the ML model and provides the predicted probability of metastasis from new patient data. Link

Data Science Consultant/Instructor

Department of Psychology, University of Delaware, Aug - Sept 2020

  • Developed and delivered technical workshops on R programming and data analysis to improve reproducibility and efficiency of research workflows for psychology graduate students. Link

Postdoctoral Research Fellow

Department of Child and Adolescent Psychiatry, NYU Langone Health, March -Jul 2020

  • Designed surveys for a longitudinal research study on maternal health during the pandemic
  • Managed data collection from 1,000 pregnant and new mothers over a time span of 3 months.
  • Developed data analysis plans and conducted data quality control for a large-scale database that combined survey and EMR data.
  • Extracted key insights from the survey dataset to quantify the negative impact of the pandemic on perinatal healthcare, and created a data dashboard using Shiny.

PhD Student Research

Department of Psychology, Columbia University, Sept 2014 - March 2020

  • Modeled longitudinal changes in brain structure (MRI) using hierarchical models in R to quantify the non-linear effects of early adversity on brain development during childhood and adolescence. Manuscript preprint: https://psyarxiv.com/yp5h2/
  • Applied PCA with bootstrap aggregation to identify 5 reproducible principle components of adversity (e.g. maltreatment, instability) using clinical features in a cohhort of adopted youth.
  • Coded pipelines in python and bash to process 50GB of functional and structural brain imaging (MRI) data using the university computing cluster, decreasing processing time 100-fold.