Backend is warming up... (this may take ~30 seconds on first visit)

About This Model

Thesis-grade classifier for metabolic obesity phenotyping.

Performance

  • Model: PSO-optimized LightGBM
  • Accuracy: 92.56%
  • F1-score: 0.9234
  • Cohen Kappa: 0.9041

Dataset

NHANES, n=3,899, ages 18-65, CDC USA.

Top 5 Features

  1. WHtR (16.57%)
  2. Waist Circumference
  3. BMI
  4. Triglycerides
  5. TG/HDL Ratio

Authors

Emir Ozcan (211805073) and Nazan Sonal (211805027)

Aydin Adnan Menderes University, Computer Engineering