Table 4 Machine learning and neural network model performance to identify patients with multiple hospitalizations for severe exacerbations of asthma and COPD at YNHH
From: Deep learning prediction of hospital readmissions for asthma and COPD
AUC | CI | Sensitivity | Specificity | Precision | Accuracy | |
|---|---|---|---|---|---|---|
Full model | ||||||
Random forest | 0.87 | 0.83–0.90 | 50% | 93% | 0.76 | 0.81 |
Naïve Bayes | 0.84 | 0.80–0.88 | 46% | 93% | 0.73 | 0.79 |
SVM | 0.88 | 0.84–0.91 | 55% | 93% | 0.76 | 0.82 |
Gradient boosted trees | 0.89 | 0.86–0.92 | 63% | 90% | 0.72 | 0.82 |
Multilayer perceptron | 0.87 | 0.83–0.90 | 79% | 79% | 0.61 | 0.79 |
Asthma | ||||||
Random forest | 0.81 | 0.73–0.89 | 18% | 100% | 1.00 | 0.83 |
Naïve Bayes | 0.88 | 0.81–0.94 | 62% | 90% | 0.64 | 0.84 |
SVM | 0.79 | 0.70–0.88 | 3% | 100% | 1.00 | 0.79 |
Gradient boosted trees | 0.80 | 0.71–0.89 | 29% | 98% | 0.83 | 0.84 |
Multilayer perceptron | 0.83 | 0.75–0.91 | 71% | 84% | 0.55 | 0.81 |
COPD | ||||||
Random forest | 0.87 | 0.83–0.91 | 58% | 90% | 0.73 | 0.79 |
Naïve Bayes | 0.83 | 0.79–0.87 | 13% | 97% | 0.67 | 0.69 |
SVM | 0.88 | 0.85–0.91 | 62% | 88% | 0.72 | 0.80 |
Gradient boosted trees | 0.88 | 0.84–0.91 | 69% | 85% | 0.69 | 0.80 |
Multilayer perceptron | 0.87 | 0.84–0.91 | 84% | 78% | 0.65 | 0.80 |