Table 2. Summary Measures of Algorithm Performance for Phase Classification.

Metric	SVM, Algorithm 1, Instrument Labels	RNN, Algorithm 2, Instrument Labels	CNN, Algorithm 3, Images	CNN-RNN, Algorithm 4, Images	CNN-RNN, Algorithm 5, Images and Instrument Labels
Unweighted accuracy (95% CI)	0.938 (0.937-0.939)	0.959 (0.958-0.960)	0.956 (0.954-0.957)	0.921 (0.920-0.923)	0.915 (0.913-0.916)
Frequency-weighted accuracy (95% CI)	0.935 (0.934-0.936)	0.957 (0.956-0.958)	0.955 (0.953-0.956)	0.919 (0.918-0.920)	0.913 (0.912-0.914)
Inverse variance−weighted accuracy (95% CI)	0.963 (0.962-0.965)	0.976 (0.975-0.978)	0.958 (0.957-0.960)	0.928 (0.926-0.930)	0.920 (0.918-0.922)
Unweighted AUC (95% CI)	0.737 (0.730-0.744)	0.773 (0.770-0.776)	0.712 (0.704-0.719)	0.752 (0.750-0.755)	0.737 (0.735-0.739)

Abbreviations: AUC, area under the receiver operating characteristic curve; CNN, convolutional neural network; RNN, recurrent neural network; SVM, support vector machine.