Skip to main content

Advertisement

Table 3 Mean overall sensitivity (and 95 % uncertainty intervals) on three datasets for 35 train/test iterations

From: Naive Bayes classifiers for verbal autopsies: comparison to physician-based classification for 21,000 child and adult deaths

Study (training/testing sample size)a NBC OTM InterVA-4b Median, all three classifiers
MDS 0.57 0.50 0.43 0.50
(11,000/555)c (0.57, 0.58) (0.50,0.51) (0.40,0.45)
Agincourt 0.48 0.42 0.38 0.42
(2,300/2,300) (0.48,0.48) (0.41,0.42) (0.36,0.41)
Matlab 0.51 0.50 0.45 0.50
(1,000/1,000) (0.50,0.51) (0.50,0.51) (0.43,0.47)
Median, all three datasets 0.51 0.50 0.43 0.50
  1. NBC naïve Bayes classifier, OTM open-source Tariff Method, VA verbal autopsy, MDS Million Death Study
  2. aTraining/testing sample size, with no training required for InterVA-4
  3. bInterVA-4 was evaluated on a testing dataset of 50 randomly selected records out of 555 records, in each of the 35 iterations
  4. cSensitivity using 555/555 training/testing records from the MDS dataset were 0.55 (0.54, 0.55) and 0.49 (0.48, 0.50), respectively, for NBC and OTM
  5. Specificity achieved by all automated classifiers across all datasets ranged from 0.96 to 0.97, and the largest uncertainty interval observed was (0.96,0.97)