Table 2 Table showing each machine learning technique and its 5-fold Cross-Validation accuracy and its test set accuracy with each NLP classifier
From: Word2Vec inversion and traditional text classifiers for phenotyping lupus
Technique | Data form | CV Acc. | CV CI (α=0.95) | Test Acc. |
|---|---|---|---|---|
ICD-9 billing codes | N/A | 89.655 | N/A | 90.00 |
Word2Vec inversion | N/A | 89.653 | [89.281, 90.025] | 90.039 |
Neural network | BOWs | 84.138 | [80.887, 87.630] | 87.100 |
| Â | CUIs | 94.138 | [89.539, 92.358] | 92.10 |
Random forests | BOWs | 95.172 | [93.875, 94.539] | 95.250 |
| Â | CUIs | 95.345 | [94.889, 95.318] | 95.00 |
Naïve Bayes | BOWs | 85.000 | [80.141, 83.859] | 82.000 |
| Â | CUIs | 81.207 | [76.087, 79.013] | 77.55 |
Support vector machines | BOWs | 86.724 | [83.031, 86.469] | 84.750 |
| Â | CUIs | 90.862 | [90.470, 92.230] | 91.35 |