Table 1.

Summary of reviewed studies addressing blood glucose prediction: prediction horizon in minutes, objective population criteria, number of participants in the cohort, mean number of monitored days per patient, mean number of monitored hours per day, type of monitoring technology, existence of monitoring during the overnight period (O) and inclusion of exercise or physical activity information (E).

Prediction horizon (min)	Population	Cohort	Days	Time	O	E	Method	Ref	Year
15, 30, 45	T1Da	15	28	10 h	✓		ANNb	[39]	2010
30	T1D	12	10	24 h	✓		ANN	[33]	2010
75	Critical Care	1	16	15 h	✓		ANN	[40]	2010
75	T1D	27	5	24 h	✓		ANN	[41]	2011
30	T1D	5VPc, 1	7	24 h	✓	✓	ANN	[42]	2011
30, 45	T1D	30VP	8	24 h	✓		ANN	[43]	2012
15, 30, 60, 120	T1D	27	13	24 h	✓	✓	RFd	[44]	2012
30	T1D	20VP, 9	11, 7	24 h	✓		ANN	[45]	2012
30	T1D	10	3	24 h	✓		SVMe, RAf, ANN	[46]	2013
15, 30, 45	T1D	23	6.1	24 h	✓		RA, ANN	[47]	2013
15, 30, 60, 120	T1D	27	13	24 h	✓	✓	SVR	[48]	2013
30	T1D	20	3	24 h	✓		ANN	[49]	2015
15, 30, 45, 60	T1D	6	11	24 h	✓	✓	ANN	[50]	2015
30, 60, 120	T1D	10	6	24 h	✓	✓	ANN	[51]	2015
30	T1D	15	13	24 h	✓	✓	ANN	[52]	2015
30	T1D	5VP, 1	30	24 h	✓		SVR	[53]	2016
—	T2Dg	346	1	—			ANN	[54]	2016
5, 15, 30, 45, 60	T1D	15	13	24 h	✓	✓	Kernel	[55]	2016
60	T1D	5	90	24 h			EAh	[56]	2016
1440	T1D, T2D	8	3	24 h		✓	DTi	[57]	2016
30	T1D	3	10	24 h	✓		EA	[58]	2016
30,60	T1D	17	6	24 h	✓		RA	[59]	2017
60, 120, 150, 180	T1D	20VP	14	24 h	✓		EA	[60]	2017
0	T2D	3	23	—			NBj	[61]	2017
30, 60, 90, 120	T1D	10	10	24 h	✓		KNNk, RF, EA	[62]	2017
120	T1D	100VP	14	24 h	✓		EA	[63]	2017
30, 60, 90	T1D & T2D	106	<7	24 h	✓		RA and ANN	[64]	2017

^aT1D: type 1 diabetes.

^bANN: artificial neural network.

^cVP: virtual patient.

^dRF: random forest.

^eSVM: Support Vector Machine

^fRA: regression algorithm.

^gT2D: type 2 diabetes.

^hEA: evolutionary algorithm.

ⁱDT: decision tree.

^jNB: Naïve Bayes.

^kKNN: k-nearest neighbor.