Skip to main content

Advertisement

Table 4 Results of total analgesic consumption (Continuous + PCA) prediction

From: Decision tree-based learning to predict patient controlled analgesia consumption and readjustment

Total Analgesic Consum. Prediction (%) C4.5 bagging C4.5 AdaBoost C4.5 ANN* Random Forest Rotation Forest SVM NB
Low Consum. Sensitivity 84.3 79.2 77.4 69.8 80.1 83.1 8.0 79.0
Med Consum. Sensitivity 83.5 75.8 72.8 79.6 83.6 82.0 96.1 67.7
High Consum. Sensitivity 62.4 61.2 60.6 21.6 47.4 62.0 0.0 38.0
Low Consum. Precision 84.3 78.8 76.3 80.2 81.4 82.9 59.4 71.8
Med Consum. Precision 79.7 75.3 73.5 66.1 74.8 78.8 50.6 70.2
High Consum. Precision 78.5 66.0 62.8 56.9 80.4 76.3 0.0 46.3
Overall Accuracy 80.9 75.1 72.8 68.5 77.4 79.7 50.7 67.9
  1. *ANN consisting of an input layer of 279 input units, one hidden layer of 140 hidden units, and one output layer of 3 output units.
  2. Learning rate= 0.3; momentum rate= 0.2.
  3. SVM using a radial basis function, exp(−gamma*|u-v|2), where gamma=1/(number of attributes)=1/279.