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
- WHtR (16.57%)
- Waist Circumference
- BMI
- Triglycerides
- TG/HDL Ratio
Authors
Emir Ozcan (211805073) and Nazan Sonal (211805027)
Aydin Adnan Menderes University, Computer Engineering