Cardiorespiratory Fitness Prediction in Korean Population

Development of a CRF prediction model for healthy Korean adults using health screening and exercise stress test data from Samsung Medical Center.

Period: 2020.05 – 2021.09

Affiliation: Samsung Medical Center

Summary: Developed a cardiorespiratory fitness (CRF) prediction model for the Korean population using data from health screening recipients who underwent regular checkups and exercise stress tests at Samsung Medical Center.

Key Responsibilities:

  • Designed inclusion & exclusion criteria based on literature review
  • Built analysis database by querying and merging data from health screening center DB and CDW
  • Data cleaning, EDA, and derived variable generation (lab data, questionnaire data, exercise stress test data)
  • Developed models using Linear Regression, Elastic Net, Random Forest, and Gradient Boosting Machine
  • Performed comparative study and selected Linear Regression-based formula as the final model
  • Validated estimated CRF as a significant predictor of mortality via Cox Proportional Hazard Model (dose-response relationship)

Outcome: 2 SCI publications