. 2017 Oct 3;17(10):2269. doi: 10.3390/s17102269

Table 1.

Related works on cushion based systems.

Author	Pressure Sensor	Accelerometer	Integrate	Features	Classification	Accuracy
Author	Pressure Sensor	Accelerometer	Unit	Features	Method	Accuracy
Yu et al. [26]	2 on the seat and	Backrest	No	N/A	SVM	N/A
Yu et al. [26]	4 on the backrest	Backrest	No	N/A	SVM	N/A
Barba et al. [27]	8 on the seat and	No	-	N/A	N/A	N/A
Barba et al. [27]	8 on the backrest	No	-	N/A	N/A	N/A
Zemp et al. [28]	16 pressure sensors	Backrest	No	N/A	SVM, MNR,	90.9%
					Boosting,
					NNs, RF
Cheng et al. [29]	4 under the	No	-	Mean, RMS,	LDA	88%
Cheng et al. [29]	chair leg	No	-	Center of weight	LDA	88%
Fu et al. [30]	4 on the seat and	No	-	N/A	HMM	N/A
Fu et al. [30]	4 on the backrest	No	-	N/A	HMM	N/A
Kumar et al. [31]	4 on the backrest	No	-	Mean and	ERT	86%
Kumar et al. [31]	4 on the backrest	No	-	variance, FFT etc.	ERT	86%
Zhu [32]	4 pressure sensors	No	-	Approximate Entropy	N/A	N/A
Ma et al. [33]	2 on the seat and	Waist	No	Mean and standard	J48	96.85%
Ma et al. [33]	1 on the backrest	Waist	No	deviation	J48	96.85%

SVM: Support Vector Machine; MNR: Multinomial Regression; NNs: Neural Networks; RF: Random Forest; HMM: Hidden Markov Model; ERT: Extremely Randomized Trees.