Research Article
Neural Network-Based Coronary Heart Disease Risk Prediction Using Feature Correlation Analysis
Table 4
Results of NNs eliminating the lowest ranked features (%).
| Without features | Accuracy |
| Without 1 (H_C) feature | 77.743 | Without 2 (H_C and H_B) features | 78.518 | Without 3 (without 2 features and cirrhosis) features | 80.644 | Without 4 (without 3 features and TD) features | 80.920 | Without 5 (without 4 features and hemoglobin) features | 81.120 | Without 6 (without 5 features and sex) features | 81.141 | Without 7 (without 6 features and CRF) feature | 81.163 | Without 8 (without 7 features and smoking) features | 81.018 | Without 9 (without 8 features and HDL) features | 80.921 | Without 10 (without 9 features and triglyceride) features | 80.222 | Without 11 (without 10 features and diabetes) features | 79.522 | Without 12 (without 11 features and BMI) features | 79.209 |
|
|