Toward Foundational Model for Sleep Analysis Using a Multimodal Hybrid Self-Supervised Learning Framework

Open paper

Benchmark	Model	Rank	Results
sleep-stage-detection-on-shhs	SynthSleepNet (EEG2+EOG2+EMG1)	#1	Accuracy: 89.89%Cohen's Kappa: 0.860Macro-F1: 0.845
sleep-stage-detection-on-shhs	SynthSleepNet (EEG1+EOG1+EMG1)	#2	Accuracy: 89.28%Cohen's Kappa: 0.850Macro-F1: 0.835
sleep-stage-detection-on-shhs	SynthSleepNet (EEG1+EOG1)	#5	Accuracy: 88.31%Cohen's Kappa: 0.840Macro-F1: 0.820