Fig 1.
Preprocessing steps and general CNN-LSTM neural network architecture.
Table 1.
Performance results of each individual model on the local validation scenario.
Table 2.
Performance results of the individual local models on the external validation scenario.
Table 3.
Performance comparison between individual models and the ensemble approach in the local and external validation scenarios.
Table 4.
Global performance comparison by aggregating results across all datasets.
Table 5.
Indices of human inter-rater agreement reported in the literature compared with the performance achieved by our proposed deep-learning approach.
Table 6.
Indices of automatic scoring agreement reported in the literature in comparison with the results achieved by the proposed deep-learning approach.