Table 2.

Summary of eating detection and recognition works, validated in free-living conditions.

	Sensor Type	No. Sensors	Features	Classifier	Performance Metrics
[7]	Inertial	1	Manipulation, linear acceleration, amount of wrist roll motion, regularity of wrist roll motion	Naive Bayes	Accuracy (weighted): 81%
[8]	Multimodal	2	MAV, RMS, maximum, median, entropy, zero crossings, no. peaks, average range, wavelength, no. slope sign changes, energy of the frequency spectrum, energy spectrum, entropy spectrum, fractal dimension, peak frequency, standard deviation, and other features derived from the aforementioned	ANN	Accuracy: 90%
[6]	Inertial	1	Mean, Variance, Skewness, Kurtosis, RMS	Random Forest	F1-scores: 71–76%
[9]	Inertial	2	RMS, variance, entropy, peak power, power spectral density, zero crossing, variance of zero crossing, peak frequency, number of auto-correlation peaks, prominent peaks, weak peaks, maximum auto-correlation value, first peak	Random Forest	Accuracy: 93%; F1-score: 80%