Do Not Sleep on Traditional Machine Learning: Simple and Interpretable Techniques Are Competitive to Deep Learning for Sleep Scoring

Benchmark Model Rank Results
multimodal-sleep-stage-detection-on-sleep-edf-scCatBoost#1Accuracy: 86.4%Macro-F1: 0.802Cohen's kappa: 0.812
multimodal-sleep-stage-detection-on-sleep-edf-scLinear model#2Accuracy: 85.7%Macro-F1: 0.809Cohen's kappa: 0.806
multimodal-sleep-stage-detection-on-sleep-edf-stCatBoost#1Accuracy: 83.6%Macro-F1: 0.795Cohen's kappa: 0.765
multimodal-sleep-stage-detection-on-sleep-edf-stLinear model#2Accuracy: 82.9%Macro-F1: 0.792Cohen's kappa: 0.759
sleep-stage-detection-on-mass-ss3CatBoost#3Accuracy: 86.7%Cohen's kappa: 0.803Macro-F1: 0.817
sleep-stage-detection-on-sleep-edfCatBoost#2Accuracy: 86.6%Cohen's kappa: 0.816Macro-F1: 0.810
sleep-stage-detection-on-sleep-edfLinear model#4Accuracy: 86.3%Cohen's kappa: 0.813Macro-F1: 0.805