Skip to main content

Table 5 Survival prediction

From: Data analytics and clinical feature ranking of medical records of patients with sepsis

method MCC F1 score accuracy TP rate TN rate
MLP 0.31±0.12 0.43±0.11 0.67±0.08 0.75±0.12 0.70±0.07
DL 0.16±0.10 0.30±0.07 0.72±0.05 0.50±0.12 0.82±0.07
NB 0.15±0.11 0.28±0.08 0.82±0.05 0.21±0.12 0.92±0.06
RF 0.15±0.10 0.28±0.08 0.64±0.05 0.58±0.16 0.65±0.06
LR 0.13±0.12 0.26±0.08 0.69±0.05 0.45±0.20 0.73±0.06
SVM (linear) 0.11±0.13 0.26±0.12 0.72±0.09 0.46±0.19 0.71±0.11
DT 0.10±0.12 0.25±0.11 0.69±0.08 0.39±0.17 0.74±0.09
SVM (kernel) 0.09±0.11 0.24±0.09 0.72±0.06 0.38±0.16 0.75±0.08
k-NN 0.08±0.13 0.22±0.09 0.69±0.11 0.38±0.24 0.73±0.15
method PR AUC ROC AUC PPV NPV  
MLP 0.29±0.07 0.86±0.13 0.94±0.07 0.19±0.07  
DL 0.20±0.06 0.86±0.18 0.88±0.05 0.19±0.13  
NB 0.07±0.06 0.80±0.28 0.88±0.03 0.31±0.16  
RF 0.32±0.06 0.86±0.25 0.91±0.04 0.19±0.07  
LR 0.24±0.05 0.86±0.24 0.90±0.04 0.20±0.08  
SVM (linear) 0.27±0.11 0.83±0.24 0.90±0.04 0.19±0.08  
DT 0.23±0.08 0.90±0.23 0.89±0.04 0.19±0.09  
SVM (kernel) 0.20±0.07 0.80±0.24 0.89±0.04 0.19±0.07  
k-NN 0.23±0.13 0.91±0.21 0.89±0.05 0.18±0.11  
  1. Performance of the learned models with the different methods evaluated with the different metrics, expressed in the format “average value ± standard deviation”, obtained on 100 executions. DT: decision tree. k-NN: k-nearest neighbors. DL: deep neural network with 3 hidden layers and weight decay. LR: logistic regression. NB: naive Bayes. MLP: multi-layer perceptron neural network. SVM (kernel): support vector machine with kernel. SVM (linear): linear support vector machine. MCC: Matthews correlation coefficient. TP rate: true positive rate (sensitivity, recall). TN rate: true negative rate (specificity). PR: precision-recall curve. ROC: receiver operating characteristic. AUC: area under the curve. MCC: worst value –1.00 and best value +1.00. PPV: positive predictive value (precision). NPV: negative predictive value. F1 score, accuracy, TP rate, TN rate, PR AUC, ROC AUC, PPV, NPV: worst value 0.00 and best value 1.00. Imbalance of this dataset: survived patients’ class: 1’s positives, #elements = 316 (86.81%), and deceased patients’ class: 0’s negatives, #elements = 48 (13.19%)
\