Table 2.

A summary of automated AF detection in RR interval signals.

Authors	Data	Digital Biomarkers	AI	Performance in %			Salient Features
				Acc	Sen	Spe
Ivanovic et al., 2019 [181]	Measurement data	-	DL	89.67	94.20	-	+Three class problem +Cross-validation used +Detailed model description −No signal detrending −No benchmark data
Anderson et al., 2019 [197]	MIT-BIH NSR/AFDB/Arrhythmia Database	-	CNN and RNN	-	98.98%	96.95%	+Three databases +Cross-validation used −No signal detrending
Faust et al., 2018 [182]	MIT-BIH AFDB	-	LSTM	98.51	98.32	98.67	+First deep learning detector based on RR intervals −No signal detrending −One database
Henzel et al., 2017 [198]	MIT-BIH AFDB	Linear measures	Threshold	93	90	95	+Insight into digital biomarkers −No AI −One database
Cui et al., 2017 [199]	MIT-BIH NSR/AFDB	Ensemble model	Threshold	97.78	97.04	97.96	+Insight into digital biomarkers +Two databases −No AI
Islam et al., 2016 [200]	MIT-BIH AFDB	Entropy	Threshold	96.38	96.39	96.38	+Insight into digital biomarkers −No AI −One database
García et al., 2016 [201]	MIT-BIH AFDB/Arrhythmia Database	RWE, SWT	Threshold	93.32	91.21	94.53	+Insight into digital biomarkers +Two databases −No AI
Kennedy et al., 2016 [202]	SPIT1, THEW2, MIT-BIH Arrhythmia Database/AFDB/ LTAFDB/NSR/SVA	Sample entropy, CV, RMSSD, MAD	RF	-	92.80	98.30	+Most data +Cross-validation −Traditional machine learning was used to gain insight into digital biomarkers
Petrėnas et al., 2015 [203]	MIT-BIH NSR/AFDB	Linear measures	Threshold	-	97.1	98.3	+Insight into digital biomarkers +Two databases −No AI
Andersson et al., 2014 [204]	MIT-BIH Arrhythmia Database/AFDB	Statistical measures	Threshold	-	94.6	95.8	+Insight into digital biomarkers +Two databases −No AI
Zhou et al., 2014 [205]	MIT-BIH Arrhythmia Database/AFDB/NSR	Shannon entropy	Threshold	96.05	96.72	95.07	+Insight into digital biomarkers +Two databases −No AI
Lee et al., 2013 [136]	MIT-BIH Arrhythmia Database/NSR	TVCF, Shannon entropy	Threshold	-	98.20	97.70	+Insight into digital biomarkers +Two databases −No AI
Lake and Moorman 2011 [69]	MIT-BIH AFDB	Sample entropy	Threshold	-	91	94	+Insight into digital biomarkers −No AI −One database
Lian et al., 2011 [68]	MIT-BIH Arrhythmia Database/AFDB/ NSR/NSR2	Statistical measures	Threshold	-	95.9	-	+Insight into digital biomarkers +Three databases −No AI
Lake and Moorman 2011 [69]	MIT-BIH AFDB	Sample entropy	Threshold	-	91	94	+Insight into digital biomarkers −No AI −One database
Mohebbi et al., 2011 [206]	MIT-BIH AFDB	RQA, recurrence plot	SVM	-	97.00	100.00	+Insight into digital biomarkers −One database
Yaghouby et al., 2010 [67]	MIT-BIH Arrhythmia Database	Statistical and geometrical	Genetic algorithm	99.11	-	-	+Insight into digital biomarkers −One database
Huang et al., 2010 [86]	MIT-BIH AFDB/NSR	Linear	Threshold	-	96.10	98.10	+Insight into digital biomarkers +Two databases −No AI
Dash et al., 2009 [63]	MIT-BIH AFDB	TPR, RMSSD, Shannon entropy	Threshold	-	90.20 - 94.40	91.20 - 95.10	+Insight into digital biomarkers −No AI −One database
Babaeizadeh et al., 2009 [64]	MIT-BIH AFDB	Statistical measures	DT	-	98	-	+Insight into digital biomarkers −One database
Ghodrati and Marinello, 2008 [207]	MIT-BIH Arrhythmia Database	Probability density function	Threshold	-	92	-	+Insight into digital biomarkers −No AI −One database
Ghodrati et al., 2008 [208]	Measurement data, MIT-BIH Arrhythmia Database/AFDB	Normalized absolute deviation and normalized absolute difference	Threshold	-	89	-	+Insight into digital biomarkers +Two databases −No AI
Kikillus et al., 2007 [209]	MIT-BIH NSR/AFDB	Linear, Frequency and RQA	Threshold	-	94.1	93.4	+Insight into digital biomarkers +Two databases −No AI
Logan and Healey, 2005 [210]	MIT-BIH AFDB	Statistical measures	Threshold	-	94.4	97.2	+Insight into digital biomarkers −No AI −One database
Tateno and Glass, 2001 [211]	MIT-BIH AFDB	CV & Kolmogorov–Smirnov test	Threshold	-	91.20	96.08	+Insight into digital biomarkers −No AI −One database
Artis et al., 1991 [212]	MIT-BIH AFDB	Statistical measures	ANN	-	92.86	-	+Insight into digital biomarkers −No AI −One database −No validation

¹ No further information on the database provided in the paper, ² (web page (13.04.2020): http://thew-project.org/).