Skip to main content

Table 3 Performance results of the top three models with optimized thresholds

From: Computational determination of hERG-related cardiotoxicity of drug candidates

Algorithm Threshold Accuracy MCC Sensitivity Specificity PPVa
Logistic regression 0.57 0.814 0.307 0.557 0.844 0.292
Neural network 0.82 0.901 0.368 0.321 0.967 0.542
Ridge regression 0.64 0.864 0.332 0.448 0.912 0.371
  1. aPPV: Positive predictive value is defined as the number of true positives/(the number of true positives + the number of false positives)