Table 2 Comparative classification accuracies results. Balanced classification accuracies obtained by four models, i.e. 1D CNN, 2D CNN, 2D CNN + LSTM, and the REST platform, for each of the investigated tasks, i.e., apnoea, desaturation, and artefact classification.

From: Artificial intelligence based platform for the automatic and simultaneous explainable detection of apnoea, oxygen desaturation, and artefacts in paediatric polygraphy exams (REST)

 

1D CNN

2D CNN

2D CNN + LSTM

REST

Apnoea

83.11%

(1.52%)

81.47%

(1.38%)

75.02%

(1.48%)

92.50%

(1.10%)

Desaturation

77.71%

(1.25%)

73.21%

(1.37%)

76.35%

(1.25%)

98.30%

(0.43%)

Artefact

96.15%

(0.35%)

95.88%

(0.35%)

96.41%

(0.43%)

97.59%

(0.28%)

  1. The values represent mean and standard deviation, in brackets, over 100 runs. Bold values indicate the best performance per class,obtained by the proposed model.
  2. Significant values are in bold.
Back to article page