Table 1 Overview of devices used in the application studiess

From: EEG-based headset sleep wearable devices

Device/study	Medical Certificate	Integrated Sensors	Wet/Dry	# of EEG electrodes	EEG Electrode Placement	Validation Results (average)	Compliance with Menghini et al¹⁴. Guidelines^a
Forehead
Dreem¹⁹	FDA-registered	EEG, 3D accelerometer, pulse oximeter	Dry	5	O1, O2, Fpz, F7, F8	κ = 0.75¹⁸	Yes
Sleep Profiler¹⁴⁹	MDSAP ISO 13485, EN ISO 13485, EC	EEG, EOG, EMG, ECG, accelerometer, PPG, nasal pressure transducer (optional), pulse oximeter (optional)	Dry	3	AF7, AF8, Fpz	κ = 0.67²¹	Yes
						κ = 0.07–0.62²²	Mostly yes (no B-A plots or CIs)
						κ = 0.63²³	Mostly yes (no CIs)
Cognionics headband¹⁵⁰	N/R	EEG	Dry	4	F3, A1, F4, A2	Accuracy = 74%²⁴	Mostly yes (no B-A plots or CIs)
Sleepgraph¹⁵¹	Japanese Medical Device Certification 231AHBZX00001000	EEG, EMG, EOG	Wet	2	(Fp1, M2) + 2 EOGs	κ = 0.80¹⁵²	Mostly yes (no CIs)
Somnowatch plus EEG¹⁵³	DIN EN ISO 13485 and 93/42/EEC	EEG, EOG, EMG, ECG, actigraphy	Wet	2 or 6	Fp2, A1⁵⁷/ F3, F4, A1, A2, AFz, FCz¹⁵⁴	κ = 0.78 (using 6 EEG electrodes)³⁴	Mostly yes (no CIs)
iSleep¹¹⁵	No certificate (prototype)	EEG	Dry	1	Fp1¹¹⁵	Accuracy = 87.5% (for sleep onset detection)¹⁵⁵	No
Sleepscope¹⁵⁶	Japanese Medical Device Certification: 225ADBZX00020000	EEG	Wet	2	Fpz, M1¹⁵⁷	κ = 0.75¹⁵⁷	Mostly yes (no CIs)
Prototype bipolar EEG (Neury)³¹	No certificate (prototype)	EEG	Wet	4	Subject specific	N/R³¹	No
Ear
Individualized 6-electrode silicone ear-EEG¹⁵⁸	No certificate (prototype)	Ear-EEG	Dry	6 per ear	ELA, ELB, ELE, ELI, ELG, ELK, ERA, ERB, ERE, ERI, ERG, ERK	κ ≈ 0.61³⁶	Partially yes (no B-A plots, CIs, discrepancy analysis, outlier handling)
						κ = 0.73³⁷	Yes
						κ ≈ 0.74³⁸	Partially yes (no B-A plots, CIs, discrepancy analysis, outlier handling)
						κ = 0.72³⁹	Partially yes (no B-A plots, CIs, discrepancy analysis)
Neck
Zmachine Insight+¹⁵⁹	FDA-cleared	EEG	Wet	2	A1, A2	κ = 0.72⁵³	Mostly yes (no CIs)
						κ = 0.85⁵²	Mostly yes (no CIs)

The symbol “≈” is used to represent the average κ value which is calculated based on the confusion matrix provided in the original study.
A1 and A2 are synonymous with M1 and M2, respectively.
B-A Bland-Altman, CI confidence interval, ECG electrocardiography, EEG electroencephalography, EMG electromyography, EOG electrooculography, PPG photoplethysmography.
^aThe guideline by Menghini et al for evaluating the performance of sleep trackers against gold-standard PSG included epoch-by-epoch analysis, standard accuracy metrics (sensitivity, specificity, F1 score, accuracy, Cohen’s Kappa), stage-specific sensitivity and specificity, Bland-Altman plots, discrepancy analysis (bias and limits of agreement), confidence intervals for accuracy metrics and biases, handling outliers, and a standardized analytical framework involving multiple sleep experts¹⁴.

Back to article page

Table 1 Overview of devices used in the application studiess

Search

Quick links