Table 3 Metrics of the five Machine Learning Models to predict ZIP-5 neighborhood primary risk clusters of the four oral diseases
From: Detecting high-risk neighborhoods and socioeconomic determinants for common oral diseases in Germany
Accuracy [%] | Specificity[%] | Precision[%] | Recall[%] | NPV [%] | PPV [%] | F1 Score | AUC (ROC) | |
|---|---|---|---|---|---|---|---|---|
Periodontitis | ||||||||
Logistic Regression | 86.8 | 91.9 | 99.8 | 86.7 | 13.8 | 99.8 | 0.93 | 0.97 |
Decision Tree | 87.3 | 73.0 | 99.3 | 87.7 | 12.1 | 99.3 | 0.93 | 0.80 |
Random Forest | 93.5 | 59.5 | 99.0 | 94.2 | 19.3 | 99.0 | 0.97 | 0.77 |
Support Vector Machine | 92.1 | 75.7 | 99.4 | 92.5 | 18.9 | 99.4 | 0.96 | 0.95 |
Neural Network | 88.2 | 78.4 | 99.4 | 88.5 | 13.6 | 99.4 | 0.94 | 0.95 |
Caries (severe) | ||||||||
Logistic Regression | 83.2 | 75.9 | 98.9 | 83.5 | 14.5 | 98.9 | 0.91 | 0.85 |
Decision Tree | 78.4 | 63.8 | 98.3 | 78.9 | 10.0 | 98.3 | 0.88 | 0.71 |
Random Forest | 88.0 | 62.1 | 98.5 | 89.0 | 17.1 | 98.5 | 0.93 | 0.76 |
Support Vector Machine | 89.2 | 62.1 | 98.5 | 90.2 | 18.9 | 98.5 | 0.94 | 0.86 |
Neural Network | 88.1 | 62.1 | 98.5 | 89.1 | 17.3 | 98.5 | 0.94 | 0.80 |
Irreversible Pulpitis | ||||||||
Logistic Regression | 88.6 | 96.6 | 99.8 | 88.2 | 31.7 | 99.8 | 0.94 | 0.95 |
Decision Tree | 87.8 | 93.2 | 99.6 | 87.5 | 29.8 | 99.6 | 0.93 | 0.90 |
Random Forest | 90.1 | 86.4 | 99.1 | 90.3 | 33.6 | 99.1 | 0.95 | 0.88 |
Support Vector Machine | 89.8 | 87.5 | 99.2 | 90.0 | 33.2 | 99.2 | 0.94 | 0.95 |
Neural Network | 91.5 | 70.5 | 98.2 | 92.7 | 35.4 | 98.2 | 0.95 | 0.91 |
Tooth Loss | ||||||||
Logistic Regression | 90.5 | 90.0 | 99.9 | 90.5 | 5.5 | 99.9 | 0.95 | 0.96 |
Decision Tree | 82.6 | 90.0 | 99.9 | 82.6 | 3.1 | 99.9 | 0.90 | 0.86 |
Random Forest | 86.2 | 100.0 | 99.9 | 86.1 | 4.3 | 99.9 | 0.93 | 0.93 |
Support Vector Machine | 84.8 | 100.0 | 99.9 | 84.7 | 3.9 | 99.9 | 0.92 | 0.98 |
Neural Network | 73.6 | 100.0 | 99.9 | 73.5 | 2.3 | 99.9 | 0.85 | 0.95 |