IEEE Circuits and Systems Magazine - Q3 2018 - 14

Table III.
Comparison of different PPG-based BP models and algorithms.
Dataset

Calibration

Accuracy(mmHg)

[58]

Non-linear Regression,
(ECG + PPG)

5 Case

A certain time for
each person

ME: SBP(5.4%), DBP(7.7%)

[59]

Linear Regression,
(ECG + PPG), adaptive
Kalman filter

10 subjects

Every 3.3 hours

STD: SBP(1.69), DBP(1.04)

[37]

ACF, linear regression,
(ECG + PPG)

MIMIC, 100
Cases

A certain time for
each person

ME ± STD: ABP(3.06 ± 3.69)

[34]

Iterative linear, (ECG + PPG)

MIMIC, 6
Cases

A certain time for
each person

ME ± STD: SBP(0.13 ± 3.73),
DBP(0.22 ± 4.33 mmHg)

[25]

Moens-Korteweg,
(ECG + PPG)

10 Cases

Once for every 12
hours

ME ± STD: SBP(2.4 ± 5.7)

[41]

SVR, (PPG)

MIMIC

Never

MAD ± STD: SBP(12.38 ± 16.17),
MAP(7.52 ± 9.54 mmHg),
DBP(6.34 ± 8.45 mmHg)

[42]

ANN (PPG)

MIMIC, 5000
Cases

Never

MAD ± STD: SBP(3.24 ± 4.75 mmHg),
MAP(2.16 ± 3.14), DBP(1.79 ± 2.70)

[30]

SVM, (PPG)

UQ, 32 Cases

Never

Classification(20 interval): SBP(98,81
%), DBP(98.21 %)

[43]

Random Forest, (PPG)

410 Cases

Never

MAD (#5,10,15): SBP(62.43%, 86.34%,
90.37%), DBP(79.02%, 90.24%, 93.90%)

[60]

Linear regression, (PPG)

18 subjects

A certain time for
one person

MAD ± STD: SBP(sleeping: 2.68 ± 0.39
mmHg, sleepy: 4.37 ± 1.20

[45]

Windkessel model, (PPG)

UQ, 32 Cases

Once for all

MAD ± STD: SBP(0.78 ± 13.1 mmHg),
DBP(0.59 ± 10.2)

[46]

Harmonic Balancing model,
(PPG)

5 Cases

A certain time for
one person

MAD ± STD: SBP(1.37 ± 7.61 mmHg),
DBP(−1.40 ± 6.00)

Amplitude

Algorithms

1
Pulse Height
0.9 Pulse
Pulse Rate
0.8 Amplitude Difference
0.7
0.6
0.5
0.4
0.3
0.2
Trought Depth Difference
Pulse Length
0.1
0
2
4
6
8
10
12
14
Time/s
Clean PPG

Artifact

Figure 10. Time domain features of the PPG signal and motion artifact effects [32].

extracted features. Non-parametric models try to solve
the low accuracy of parametric models.
Examples of non-parametric learning methods can be
found in [30], [41]-[43]. Reference [30] tries to classify the
SBP and DBP within different prediction ranges with 19
features extracted from the PPG signal using only a sup14

IEEE CIrCUITs AND sYsTEMs MAGAzINE

port vector machine (SVM). The estimation accuracy can
reach 98.81% and 98.21% for SBP and DBP, respectively.
Reference [41] gets its features from both the PPG and ECG
signals. Using support vector regression (SVR), it reaches
grade B for both DBP and mean blood pressure (MBP)
estimations and reaches grade C for SBP according to the
BHS protocol. In [42], an ANN-based learning process is
used to train the features from the PPG signal only. The
results can fulfill the AAMI requirement. Using only the
PPG signal, [43] implements a random forest (RF)-based
prediction model and achieves grade B for both SBP and
DBP by BHS. Even though the non-parametric BP models
exhibit better accuracy than the parametric models, they
suffer from heavy calculations.
Table III compares the different PPG-based BP models
and algorithms. The first five methods in Table III [25],
[34], [37], [58], [59] use both ECG and PPG signals with linear or nonlinear regression. These methods have higher
accuracy, but they require both ECG and PPG signals and
frequent calibration. The methods in [30], [41]-[43] do
not need frequent calibration, and only PPG information
is needed. However, they require excessive training and
ThIrD qUArTEr 2018



Table of Contents for the Digital Edition of IEEE Circuits and Systems Magazine - Q3 2018

Contents
IEEE Circuits and Systems Magazine - Q3 2018 - Cover1
IEEE Circuits and Systems Magazine - Q3 2018 - Cover2
IEEE Circuits and Systems Magazine - Q3 2018 - Contents
IEEE Circuits and Systems Magazine - Q3 2018 - 2
IEEE Circuits and Systems Magazine - Q3 2018 - 3
IEEE Circuits and Systems Magazine - Q3 2018 - 4
IEEE Circuits and Systems Magazine - Q3 2018 - 5
IEEE Circuits and Systems Magazine - Q3 2018 - 6
IEEE Circuits and Systems Magazine - Q3 2018 - 7
IEEE Circuits and Systems Magazine - Q3 2018 - 8
IEEE Circuits and Systems Magazine - Q3 2018 - 9
IEEE Circuits and Systems Magazine - Q3 2018 - 10
IEEE Circuits and Systems Magazine - Q3 2018 - 11
IEEE Circuits and Systems Magazine - Q3 2018 - 12
IEEE Circuits and Systems Magazine - Q3 2018 - 13
IEEE Circuits and Systems Magazine - Q3 2018 - 14
IEEE Circuits and Systems Magazine - Q3 2018 - 15
IEEE Circuits and Systems Magazine - Q3 2018 - 16
IEEE Circuits and Systems Magazine - Q3 2018 - 17
IEEE Circuits and Systems Magazine - Q3 2018 - 18
IEEE Circuits and Systems Magazine - Q3 2018 - 19
IEEE Circuits and Systems Magazine - Q3 2018 - 20
IEEE Circuits and Systems Magazine - Q3 2018 - 21
IEEE Circuits and Systems Magazine - Q3 2018 - 22
IEEE Circuits and Systems Magazine - Q3 2018 - 23
IEEE Circuits and Systems Magazine - Q3 2018 - 24
IEEE Circuits and Systems Magazine - Q3 2018 - 25
IEEE Circuits and Systems Magazine - Q3 2018 - 26
IEEE Circuits and Systems Magazine - Q3 2018 - 27
IEEE Circuits and Systems Magazine - Q3 2018 - 28
IEEE Circuits and Systems Magazine - Q3 2018 - 29
IEEE Circuits and Systems Magazine - Q3 2018 - 30
IEEE Circuits and Systems Magazine - Q3 2018 - 31
IEEE Circuits and Systems Magazine - Q3 2018 - 32
IEEE Circuits and Systems Magazine - Q3 2018 - 33
IEEE Circuits and Systems Magazine - Q3 2018 - 34
IEEE Circuits and Systems Magazine - Q3 2018 - 35
IEEE Circuits and Systems Magazine - Q3 2018 - 36
IEEE Circuits and Systems Magazine - Q3 2018 - 37
IEEE Circuits and Systems Magazine - Q3 2018 - 38
IEEE Circuits and Systems Magazine - Q3 2018 - 39
IEEE Circuits and Systems Magazine - Q3 2018 - 40
IEEE Circuits and Systems Magazine - Q3 2018 - 41
IEEE Circuits and Systems Magazine - Q3 2018 - 42
IEEE Circuits and Systems Magazine - Q3 2018 - 43
IEEE Circuits and Systems Magazine - Q3 2018 - 44
IEEE Circuits and Systems Magazine - Q3 2018 - 45
IEEE Circuits and Systems Magazine - Q3 2018 - 46
IEEE Circuits and Systems Magazine - Q3 2018 - 47
IEEE Circuits and Systems Magazine - Q3 2018 - 48
IEEE Circuits and Systems Magazine - Q3 2018 - Cover3
IEEE Circuits and Systems Magazine - Q3 2018 - Cover4
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2023Q3
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2023Q2
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2023Q1
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2022Q4
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2022Q3
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2022Q2
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2022Q1
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2021Q4
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2021q3
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2021q2
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2021q1
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2020q4
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2020q3
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2020q2
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2020q1
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2019q4
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2019q3
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2019q2
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2019q1
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2018q4
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2018q3
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2018q2
https://www.nxtbook.com/nxtbooks/ieee/circuitsandsystems_2018q1
https://www.nxtbookmedia.com