Skip to main content

Table 2 Sensitivity, specificity and F-score values for comparisons between lactation, pup-free and VBBSM networks

From: Bayesian modeling suggests that IL-12 (p40), IL-13 and MCP-1 drive murine cytokine networks in vivo

  VBSSM vs Seeded lactation network VBSSM vs Seeded pup-free network VBSSM Lactation vs VBBSM Pup-free network
Specificity 0.94 0.95 0.88
Sensitivity 0.30 0.14 0.17
F-Score 0.46 0.25 0.28
True positives 13 6 5
False positives 18 14 15
False negatives 30 36 25
True negatives 263 268 108
  1. Bayesian networks model conditional independence so that accurately removing arcs from all possible connections is an important measurement for accuracy. This is reflected in the specificity, which is close to one if true negatives (TNs) are high and false positives (FPs) are low. For the purpose of internal validation (in addition to the experimental aspects), the VBSSM-based results obtained without prior knowledge were compared to those from seeded Bayesian learning. It is important to note that VBSSM results were derived under a strict confidence level check. However, we cannot expect a high agreement in true positives (TPs) for the structural comparisons performed for both the lactation and pup-free data but the high TN and specificity values are encouraging (first two data columns). The last column represents the network structural comparison within the VBSSM analysis between lactation and pup-free data