Skip to main content

Table 2 Performance measures: mean (standard deviation) of Bayesian networks in five different feature sets: Braden (B), Medication (M), Diagnosis (D), Braden & Diagnosis (BD), and Braden & Medication & Diagnosis (BMD)

From: Predictive models for pressure ulcers from intensive care unit electronic health records using Bayesian networks

 

Ba

Ma

Da

BDb

BMDb

SENS

0.021 (0.034)

0.002 (0.003)

0.315 (0.027)

0.455 (0.034)

0.478 (0.025)

SPEC

0.996 (0.006)

0.999 (0.001)

0.939 (0.005)

0.908 (0.006)

0.895 (0.007)

PPV

0.146 (0.201)

0.238 (0.385)

0.301 (0.022)

0.292 (0.184)

0.274 (0.015)

NPV

0.924 (0.002)

0.923 (0.001)

0.943 (0.002)

0.953 (0.003)

0.954 (0.002)

AUC

0.731 (0.018)

0.619 (0.016)

0.810 (0.012)

0.827 (0.011)

0.819 (0.011)

  1. aMDL scoring function, Tabu search and naïve Bayes prior structure
  2. bMDL/BDeu scoring functions give the same result, Tabu search, and naïve Bayes prior structure