Skip to main content

Table 3 Comparison of different methods developed and evaluated on the PhenoCHF training and test sets, respectively.

From: Using text mining techniques to extract phenotypic information from the PhenoCHF corpus

  Discharge summaries Articles
  Exact Match Relaxed Match Exact Match Relaxed Match
Methods P R F P R F P R F P R F
Dictionary (MetaMap) 0.22 0.29 0.25 0.39 0.51 0.44 0.42 0.25 0.30 0.67 0.33 0.44
Rules 0.88 0.86 0.87 0.92 0.93 0.92 0.83 0.88 0.85 0.88 0.90 0.89
MEMMs 0.67 0.33 0.52 0.87 0.60 0.54 0.18 0.55 0.24 0.20 0.56 0.28
HMMs 0.90 0.63 0.74 0.90 0.65 0.76 0.30 0.55 0.39 0.32 0.58 0.41
CRFs 0.88 0.77 0.82 0.90 0.86 0.88 0.48 0.62 0.54 0.53 0.69 0.60
  1. For MEMMs, HMMs and CRFs, only the results from the model with the best performing combination of features are presented (P=precision, R=recall, F=F-score).