Table 3 Performance metrics for SVM, RF, LSTM, Bi-LSTM, and CNN models on the test data and their average prediction time per sample
From: O2 supplementation disambiguation in clinical narratives to support retrospective COVID-19 studies
Classifier | Metric | Mean \(\varvec{\pm }\) Std Error | 95%Â Confidence Intervals | Average Prediction Time (seconds) |
|---|---|---|---|---|
SVM | P | \(0.955 \pm 0.002\) | \([0.951 - 0.959]\) | 0.258 |
R | \(0.955 \pm 0.002\) | \([0.951 - 0.959]\) | ||
F1 | \(0.955 \pm 0.002\) | \([0.951 - 0.959]\) | ||
RF | P | \(0.942 \pm 0.003\) | \([0.936 - 0.948]\) | 0.057 |
R | \(0.942 \pm 0.003\) | \([0.936 - 0.948]\) | ||
F1 | \(0.942 \pm 0.003\) | \([0.936 - 0.948]\) | ||
LSTM | P | \(0.948 \pm 0.002\) | \([0.944 - 0.952]\) | 0.501 |
R | \(0.948 \pm 0.002\) | \([0.944 - 0.952]\) | ||
F1 | \(0.948 \pm 0.002\) | \([0.944 - 0.952]\) | ||
Bi-LSTM | P | \(0.944 \pm 0.003\) | \([0.938 - 0.950]\) | 0.502 |
R | \(0.946 \pm 0.003\) | \([0.938 - 0.950]\) | ||
F1 | \(0.944 \pm 0.003\) | \([0.938 - 0.950]\) | ||
CNN | P | \(0.954 \pm 0.002\) | \([0.950 - 0.958]\) | 0.130 |
R | \(0.954 \pm 0.002\) | \([0.950 - 0.958]\) | ||
F1 | \(0.954 \pm 0.002\) | \([0.950 - 0.958]\) |