Table 5.
Sentence-level Precision (P), Recall (R) and F1-Score (F) of NLP tools, basic ensemble, and advanced ensemble methods. Numbers highlighted as blue underlined text indicate the best scores for each evaluation trial.
| Category | Method | Congestive Heart Failure (CHF) | Weight Management / Obesity (WM/O) | Kawasaki Disease - Public (KD-Public) | Kawasaki Disease - Private (KD-Private) | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| P | R | F | P | R | F | P | R | F | P | R | F | ||
| NLP Tool | cTAKES | .354 | .268 | .293 | .224 | .203 | .197 | .273 | .193 | .213 | .425 | .257 | .296 |
| MetaMap | .132 | .077 | .091 | .125 | .073 | .086 | .062 | .026 | .034 | .034 | .017 | .020 | |
| Basic Ensemble | Union | .369 | .280 | .306 | .237 | .216 | .209 | .281 | .204 | .223 | .432 | .263 | .303 |
| Intersection | .115 | .065 | .076 | .101 | .060 | .070 | .047 | .015 | .021 | .027 | .011 | .013 | |
| Advanced Ensemble | Binary Relevance (BR) | .329 | .281 | .285 | .791 | .793 | .767 | .208 | .128 | .144 | .412 | .315 | .341 |
| Multi-Label K-Nearest Neighbor (MLkNN) | .268 | .200 | .211 | .673 | .621 | .629 | .147 | .088 | .102 | .385 | .289 | .312 | |
| Instance-Based Logistic Regression for Multi- Label (IBLR-ML) | .211 | .273 | .209 | .683 | .656 | .654 | .144 | .100 | .103 | .364 | .326 | .321 | |
| Random k-Labelsets (RAkEL) | .350 | .276 | .294 | .790 | .793 | .767 | .208 | .128 | .144 | .413 | .314 | .341 | |
| Ensemble of Classifier Chains (ECC) | .325 | .278 | .282 | .778 | .760 | .743 | .208 | .128 | .144 | .413 | .319 | .343 | |