The current issue and full text archive of this journal is available on Emerald Insight at:

https://www.emerald.com/insight/2049-6427.htm

IJIUS
11,3 Linear regression model and least
square method for experimental
identification of AMBU bag in
378 simple ventilator
Received 24 July 2021 Cong Toai Truong, Kim Hieu Huynh and Van Tu Duong
Revised 24 March 2022
Accepted 11 July 2022 National Key Laboratory of Digital Control and System Engineering (DCSELab),
Ho Chi Minh City University of Technology (HCMUT), Ho Chi Minh City, Vietnam;
Faculty of Mechanical Engineering,
Ho Chi Minh City University of Technology (HCMUT),
Ho Chi Minh City, Vietnam and
Vietnam National University Ho Chi Minh City, Ho Chi Minh City, Vietnam
Huy Hung Nguyen
Faculty of Electronics and Telecommunication, Saigon University,
Ho Chi Minh City, Vietnam
Le An Pham
Ho Chi Minh City University of Medicine and Pharmacy,
Ho Chi Minh City, Vietnam, and
Tan Tien Nguyen
National Key Laboratory of Digital Control and System Engineering (DCSELab),
Ho Chi Minh City University of Technology (HCMUT), Ho Chi Minh City, Vietnam;
Faculty of Mechanical Engineering,
Ho Chi Minh City University of Technology (HCMUT),
Ho Chi Minh City, Vietnam and
Vietnam National University Ho Chi Minh City, Ho Chi Minh City, Vietnam

Abstract
Purpose – In the COVID-19 outbreak periods, people’s life has been deranged, leading to disrupt the world.
Firstly, the number of deaths is growing and has the potential to surpass the highest level at any time. Secondly,
the pandemic broke many countries’ fortified lines of epidemic prevention and gave people a more honest view
of its seriousness. Finally, the pandemic has an impact on life, and the economy led to a shortage in medical,
including a lack of clinicians, facilities and medical equipment. One of those, a simple ventilator is a necessary
piece of medical equipment since it might be useful for a COVID-19 patient’s treatment. In some cases, the
COVID-19 patients require to be treated by modern ventilators to reduce lung damage. Therefore, the addition
of simple ventilators is a necessity to relieve high work pressure on medical bureaucracies. Some low-income
countries aim to build a simple ventilator for primary care and palliative care using locally accessible and low-
cost components. One of the simple principles for producing airflow is to squeeze an artificial manual breathing
unit (AMBU) iterative with grippers, which imitates the motion of human fingers. Unfortunately, the squeezing
angle of grippers is not proportional to the exhaust air volume from the AMBU bag. This paper aims to model
the AMBU bag by a mathematical equation that enables to implement on a simple controller to operate a bag-
valve-mask (BVM) ventilator with high accuracy performance.

International Journal of Intelligent

Unmanned Systems This research is funded by Vietnam National University Ho Chi Minh City (VNU-HCM) under Grant
Vol. 11 No. 3, 2023
pp. 378-395 number TX 2022-20b-01. The authors acknowledge the support of time and facilities from National Key
© Emerald Publishing Limited Laboratory of Digital Control and System Engineering (DCSELab), Ho Chi Minh City University of
2049-6427
DOI 10.1108/IJIUS-07-2021-0072 Technology (HCMUT), VNU-HCM for this study.
Design/methodology/approach – This paper provides a curvature function to estimate the air volume Linear
exhausting from the AMBU bag. Since the determination of the curvature function is sophisticated, the
coefficients of the curvature function are approximated by a quadratic function through the experimental regression
identification method. To obtain the high accuracy performance, a linear regression model and a least square model of simple
method are employed to investigate the characteristic of the BVM ventilator’s grippers angle with respect to the
airflow volume produced by the AMBU bag. ventilator
Findings – This paper investigates the correlation between the exhausting airflow of the AMBU bag and the
grippers angle of the BVM ventilator.
Originality/value – The experimental results validated that the regression model of the characteristic of the 379
exhausting airflow of the AMBU bag with respect to the grippers’ angle has been fitted with a coefficient over
98% within the range of 350–750 ml.
Keywords Simple ventilator, AMBU bag, COVID-19, Bag valve mask
Paper type Research paper

1. Introduction
Coronavirus (COVID-19) (WHO, 2020b) is a dangerous pandemic that is started at the end of
2018 and shows no sign of ending at the moment. The pandemic originated wildlife products
in the Wu-Han province of China (Feng et al., 2020; Kannan et al., 2020; Li et al., 2020) and
spread to most countries in the world. The COVID-19 pandemic is very complicating and
multi-staged manner. Currently, the number of deaths from the pandemic has reached
4,049,372 until July 14, 2021 (World Health Organization, 2021). Looking back on the past, it is
very difficult to imagine the COVID-19 pandemic turned upside anything down the globe
around the world (Haleem et al., 2020; Nigrovic, 2021). Medical bureaucracies have
established many “front lines” in the battle against COVID-19 that people’s lives have been
changed because of the dangers of a pandemic. Additionally, COVID-19 has lots of new
COVID-19 variants, which characteristic faster and dangerous (Area and Nieto, 2021;
Bindayna and Crinion, 2021). The health-care system has shown an overload not only in the
emergency resuscitation sector with doctors and nurses (Lapostolle et al., 2020; Ouyang, 2020)
but also medical device industry. Especially, this situation is more common in low-income
countries (Pinhu et al., 2003; Lagasse, 2020; Lurie et al., 2020) than the others that include
personal protective equipment (PPE), testing kits, sophisticated equipment like X-ray
machines (Gereffi, 2020) and especially life-saving mechanical ventilators (Ranney
et al., 2020).
COVID-19 is an infectious disease and has different transmission of atypical modes
(Wiwanitkit, 2020). Significantly, airborne transmission (Morawska et al., 2020) is higher,
evidenced by the rapid increase in the number of COVID-19 patients. Additionally, some high-
income countries have not estimated the devastating nature of the COVID-19 pandemic at the
beginning, so it was one of the reasons for the virus to spread rapidly around the world
(Forman et al., 2020; Schuchat, 2020). According to the recorded documents, COVID-19
patients have clinical symptoms based on the physical condition of each person. The most
clinical symptoms of COVID-19 patients are fever, cough, fatigue, bone pain, dyspnea,
sputum and ultimately leading to pneumonia (Alimohamadi et al., 2020) as similar symptoms
common flu. Meanwhile, many infected common are asymptomatic (Jenco, 2020; Padula,
2020); thus, it is difficult to screen and cohort isolation in a dedicated area for COVID-19
patients. Furthermore, some of the COVID-19 patients are deterioration of health faster than
others because of the COVID-19 secretion of multiple cytokines, which in turn trigger an
aberrant uncontrolled response known as “cytokine storm” (CSS) (Abdin et al., 2020). The CSS
causes severe acute respiratory distress syndrome (ARDS) and leads to pneumonia and
respiratory failure that are thought to be a critical factor in COVID-19-related mortality rates
(Sun et al., 2020; Tang et al., 2020). To reduce CSS-related mortality, the clinicians prescribe
using electronic medicines optimization system (EMOS) to help respiration and evade lung
IJIUS damage in the COVID-19 periods. But, EMOS has many hidden dangers (MacLaren et al.,
11,3 2020) and technology requirements (a well-trained clinician to operate, staffing, ECMO
planning and resource allocation) (Ramanathan et al., 2020). When a pandemic occurs, the
selection of patients using ECMO and priority patients is also a problematic hidden inequality
for doctors. Some of the critical situations, inter- and intra-hospital transport (Warren et al.,
2004) of critically ill patients require ventilator assistance (Gebremichael et al., 2000) or bag-
valve-mass ventilators (BVM) to be used for transportation. Some studies have shown that
380 the utilization of a manual AMBU can have negative consequences due to erratic breathing
(Blakeman and Branson, 2013). Thus, the addition of the BVM ventilator might be significant
merits solving the medical equipment crisis (the inter- and intra-hospital transport, weaning
from ventilators, vegetative life, patients whose lungs are not damaged but not can breathe
spontaneously, using at-home patients who cannot afford to be treated at the hospital for a
long time).
As mentioned above, the COVID-19 patients must be treated with ventilators that have
more functional requirements in the intensive care unit (ICU), but simple assist ventilators
equipped for primary, palliative care and safe are still worth. Thus, the number of modern-
ventilators required for COVID-19 treatment is a big challenge for the medical bureaucracies
(Ercole et al., 2009; Smetanin et al., 2009; Stiff et al., 2011; Kim et al., 2012), especially in
developing countries and low-income countries (Fisher and Heymann, 2020; The Lancet,
2020; WHO, 2020a). Several attempts have been made to build low-cost ventilators (Russell
and Slutsky, 1999; Fang et al., 2020; Truong et al., 2021). In particular, several prototypes of
ventilators operating based on the grippers principle have been developed to deal with
overloading COVID-19 cases. From the medical perspective, several existing studies showed
the two-handed mask-face technique proving to be better than the one-handed mask-face
technique. For emergency situations, the one-handed technique is still applied due to the lack
of AMBU compression, which can be conducted by the ventilator using the grippers.
Therefore, this topology of the ventilator is essential which has the advantages of easy
manufacturing and assembly, low-cost, easy-to-find materials and control volume
adjustability according to the angle of the grippers. However, the big challenge when
developing the BVM ventilator is to identify the mathematical model of the AMBU bag, which
enables for the development of an enhanced controller for simple to obtain high accuracy.
During the development of ventilators, the critical component of the BVM ventilators is the
AMBU bag; thus, it requires to investigate the characteristics between the squeezing angle of
the ventilator concerning the volume exiting the AMBU bag.
The characteristics of the AMBU bag are instability for a long time. Thus, it is necessary to
employ a suitable method to investigate the characteristic of the squeezing angle of the
grippers and the exhaust air volume of the adult AMBU bag. Some methods are proposed as
computational fluid dynamics (CFD) and experimental identification. Some studies use CFD
methods (Li and Nielsen, 2011) in multiplicity in building ventilation. The CFD method is
usually applied to problems that cannot be applied to experimental identification methods.
This method solves the problems as convection-dominated, turbulence-dominated, molecular
diffusion-dominated and increasingly popular applications in the ventilation field. On the
other hand, the experimental identification method is applied in some studies of other fields to
solve similar problems. The results of this method can use explain phenomena in the
experimental process, find external factors that influence that simulation cannot do
unforeseen (Wang and Liou, 1991). Moreover, this method can support many times to check
the experiment results and ensure the reliability of the results. Thus, the experimental
identification method is suitable for this research because the BVM ventilator is very easy to
proceed with the experiment. Moreover, we hope to find the characteristic of the squeezing
angle of the grippers and the exhaust air volume of the adult AMBU bag that is based on the
real experiment in the environment.
 This paper investigates the characteristics of the BVM ventilator’s grippers angle Linear
concerning the airflow volume produced by the AMBU bag. To achieve the objective, regression
the experiment was conducted iteratively by driving the gripper to squeeze the AMBU bag
with a fixed angle for each cycle under the limitation of the grippers’ angle. The exhaust air
model of simple
volume from the AMBU bag is integrated over time from the perception of an airflow sensor ventilator
as well as the squeezing angle of the grippers is measured by the DC motor encoder.
The airflow sensor has been calibrated to zero in initial status before ventilator
operation. The least-squares regression technique is used in this research to calculate and 381
assess the interpolation function’s acceptability. This technique performs calculations
including the interpolation function, the correlation coefficient, root mean square error, sum
square error and mean absolute error value. As a result, the characteristic between the
squeezing angle of the ventilator and the volume exiting the AMBU bag (for adults) is
obtained by the experimental identification method. The experimental results show that the
properties of the AMBU bag with a squeezing angle can be identified accurately.

2. Materials and methods

2.1 Overall system description
Figure 1 depicts a breathing circuit with a bag valve mask ventilation system. During the
inspiration phase, the grippers squeeze the AMBU bag providing a mixture of oxygen and
air to the patient’s lungs through a humidifier and High Efficiency Particulate Air (HEPA)
filter. Due to the elasticity of the lungs and the action of the one-way valve, the patient
spontaneously exhales through a breathing mask. The positive end-expiratory pressure
(PEEP) valve should be inserted into the breath circuit to minimize excessive lung pressure.
The PEEP valve retains a positive air pressure in the patient’s lung to prevent the alveoli from
collapsing during the expiration phase. The humidifier warms up the breathing air before
passing through a HEPA filter. The HEPA filter is used to remove hazardous materials from
the breathing air, such as dust, pet dander, other contaminants, etc. Furthermore, a relief
valve is installed as close to the patient’s lung as possible to minimize barotrauma by
preventing overpressure during the breath.

Figure 1.
The operating system
of BVM ventilator
IJIUS 2.2 Bag-valve-mask ventilator
11,3 A CAD model of the BVM ventilator is shown in Plate 1a. Plates 1b and 1c show the
experimental model of the BVM ventilator, which includes the following components: AMBU
bag, flexible tubing, PEEP valve, grippers, DC motor and controller box. The BVM ventilator
is initiated by pushing the Start/Home button on the main screen. Furthermore, the Start/
Home button is used to set the home position of the BVM ventilator, which is determined by
the sensor position feedback and sets the grippers to the home position. Following the homing
382 process, clinicians can change the parameter such as tidal volume (V), inspiratory-expiratory
ratio (I: E), respiratory rate (RR) and peak inspiratory pressure (PIP) that are shown on the
LCD of the BVM ventilator. The parameters of the BVM ventilator are displayed including
the PEEP valve value, the current pressure, inspiratory-expiratory ratio and breath per
minute in the LCD. After complete configuration, the clinicians can confirm the parameter
adjustment by pressing the Start/Home button again, allowing the machine to operate.
Then, the grippers are then squeezed at a regular cycle. In the following scenarios, the BVM
may issue a warning such as the present pressure exceeds the PEEP and PIP limit ranges,
the doctor sets the settings without pushing the Start/Home button to confirm, the grippers
squeezing angle exceeds the significant limits and reaches the safety limit switch.
At this initial position, the clamping point of the grippers touches the surface of the
AMBU bag. At this moment, α is conceptualized as the angle formed by the gripper and the
center of the AMBU bag. The initial position of α is shown in Figure 2a and is perceived by
the position sensor of the BVM ventilator in the start position. The DC motor powers
torque to the grippers through the gear transmission to squeeze the AMBU bag. Figure 2

Plate 1.
The modeling to show
the relationship
between the grippers
and AMBU bag
 Linear
regression
model of simple
ventilator

383

Figure 2.
Determination between
grippers and
AMBU bag

shows the AMBU bag, which is squeezed by the grippers, and consequently, the AMBU
bag is distorted and loses fixed shape. It is easy to see that the working trajectory of the
clamping point of the grippers is non-linearity. As a result, the exhaust air volume changes
nonlinearly, and this leads it impossible to measure the total amount of air entering the
patient’s lungs.

2.3 Theoretical basis

The AMBU bag adopted for this research is an adult unit. Thus, the specification on the
AMBU bag is referred to (Secured Medical Direction UK, 2019) and shown in Table 1.
The AMBU bag is assumed to be asymmetrical object whose volume can be determined
by applying the following formula containing the object’s curvature (Otten et al., 2014).
Z L
2
VAMBU ¼ π ½f ðxÞ2 dx (1)
−L2

where f ðxÞ is the AMBU bag curvature equation.

Referring to Table 1, substituting the resuscitator volume and the AMBU bag length into
equation(1), it yields:
Z 106
1650
½f ðxÞ2 dx ¼ (2)
−106 π

Specifications Value

Bodyweight >40 kg
Stroke volume 800 ml
Resuscitator volume ðVAMBU Þ 1,650 ml Table 1.
Dimensions (Length −L 3 Diameter −D) 212 3 131 mm Manual resuscitator of
Bag reservoir volume 2,000 ml AMBU bag – adult
IJIUS The determination of the function f ðxÞ is quite complicated; thus, inspired by Figure 3,
11,3 we visually approximate f ðxÞ as a quadratic function expressed as follows:
f ðxÞ ¼ ax2 þ bx þ c (3)

where a; b; c are the coefficients of the quadratic function f ðxÞ

The initial conditions of f ðxÞ are determined:
384 
f ðx ¼ −106Þ ¼ f ðx ¼ 106Þ ¼ 0
(4)
f ðx ¼ 0Þ ¼ 65:5

Substituting equation(4) into equation(3), a system of equation to solve for coefficients of the
quadratic function is obtained as:
8
< 11236a  106b þ c ¼ 0
11236a þ 106b þ c ¼ 0 (5)
:
c ¼ 65:5

By solving equation(5) for a; b and c, f ðxÞ in equation(2) can be determined as:

f ðxÞ ¼ −0:00583x2 þ 65:5 (6)

Equation(2) can be verified by substituting the approximated quadratic function in

equation(6):
Z L Z 106
2 π  2
VAMBU ¼ π ½f ðxÞ2 dx ¼ −0:00583x2 þ 65:5 dx ≈ 1523:87 ðmlÞ (7)
−L2 1000 −106

The resuscitator volume error of the AMBU bag according to the approximate curvature
equation is 7:6% in comparison to the resuscitator volume provided by the manufacturer in
Table 1. The AMBU bag simulation with f ðxÞ in equation(6) is depicted in Figure 4.
Simulating the AMBU bag with the curvature function in equation(6), using a
piecewise linear interpolation model to fit 202 data points, the resuscitator volume is
then calculated by numerically evaluating double integral based on the tiled method.
As expressed in Figure 4, the volume of the AMBU bag is 1523:7745 ðmlÞ, which is
equal to the validated resuscitator volume of equation(7). The squeezing action of the
ventilator is simulated by the AMBU bag volume reduction when the resuscitator
volume is arbitrary in the range x ≥ d and x ≤ − d. For instance, the AMBU bag volume
at d ¼ 25mm (equivalent to a gripper angleα ¼ 11:976o) is expressed in Figure 5.
As can be seen in Figure 5, the air volume remaining in the AMBU bag is
1242:9530 ðmlÞ; thus, the exhausted air volume ðVT Þ from the AMBU bag at α ¼ 11:98o

Figure 3.
Dimension 2D of
AMBU bag
 Linear
regression
model of simple
ventilator

385

Figure 4.
Simulation of AMBU
bag at α ¼ 0o (home
position)

Figure 5.
Simulation of AMBU
bag squeezed by
grippers’
angle α ¼ 11:98o

can be calculated by subtracting the air volume remaining in the AMBU bag by the
resuscitator volume as below:
VT ðα ¼ 11:98o Þ ¼ 1523:7745  1242:953 ¼ 280:8215 (8)

By simplifying the BVM ventilator with front elevation, gripper angle α could be calculated
from volume reduction range d. For α is arbitrary in the range ½0; 29:8996, the highest value of
α is 29:8996 corresponding to exhausted air volume VT ¼ 1523:7745, the theoretical
characteristic of α and VT can be determined in Figure 6.
The characteristic of α and VT can be visually denoted that it has non-linear properties.
However, within a specific range of the exhausted air volume, for instance, VT has a range
from 350ml to 700ml, a linear regression model can be approximated. Figure 7 shows linear
regression model of the α − VT theoretical characteristic in which VT value is ranging from
350ml to 700ml.
IJIUS
11,3

386

Figure 6.
The theoretical
characteristic of
grippers angle and
exhaust air volume
from AMBU bag

Figure 7.
The characteristic of
grippers angle and
exhaust air volume
from AMBU bag
within the exhausted
air volume range from
350 to 700 ml

With the correlation coefficient R ¼ 0:9985, the theoretical relationship between the grippers
angle and the exhaust air volume in linearity is as follows:
VT ðtÞ ¼ 18:43 3 αðtÞ  73:11 (9)

It is admitted that simulation by approximating contact area as a plane (Figure 5) can create
inconsistency compared with practice. Thus, its application could be restricted by only using
as a study of the experimental characteristic to help build up more suitable experiment
scenarios. For further development of transfer function determination or controller design, a Linear
mathematical model obtained by the system identification method should be used instead of a regression
simulation model. Additionally, a simulation implemented by the computational fluid
dynamic method could be considered as an alternative.
model of simple
ventilator
3. Experimental result
For purpose of validating the theoretical characteristic of αðtÞ and V ðtÞ in equation(9), an 387
experiment was carried out on the experimental model of the BVM ventilator. An airflow
sensor is used to measure the exhausting airflow from the AMBU bag, from that, estimate the
exhausted air volume. Besides, the grippers angle is calculated from the motor encoder.
It is noted that in practice, the exhausted air volume reached its maximum value at
around 700ml compared to the value of 1523:7745ml in the simulation result. This could be
mainly explained by the compressibility of air (when the BVM ventilating, a positive air
pressure above atmospheric pressure must be maintained in the airway to keep unstable
lung units from collapsing, this value is approximately 5 ÷ 15 cmH2 O (Bime et al., 2017)).
Furthermore, this could also be due to the elasticity of the AMBU bag, which is made from
silicone material as well as the complication of the contact area between grippers’ paddles
and the AMBU bag surface leading to a difference between simulation results and
experiment results. In the future, the mentioned difference could be resolved by
approaching a more accurate simulation and different grippers’ paddles. Besides, it can
be seen that the experimental data points shown in Figure 8 are distributed around a non-
linear curve that has a similar shape as the curve presented in Figure 6. Therefore, the non-
linearity of the characteristic of αðtÞ and V ðtÞ has been identified. To simplify the
ventilator model, a two-segment linear regression model has been applied. However,
we first apply a conventional linear regression model for the purpose of comparing the
results with a two-segment linear regression model. Figure 9 shows the linear regression
model of the experimental characteristics of α and VT .
The experimental characteristic of the two variables VT and α can be approximated as a
linear function with the fitting properties presented in Table 2.

Figure 8.
The experimental
characteristic of
grippers angle and
exhaust air volume
IJIUS
11,3

388

Figure 9.
The experimental
characteristic of
grippers angle and
exhaust air volume
with linear regression

Linear model VT ðtÞ ¼ p1 3 αðtÞ þ p2

Coefficients (95% confidence bounds) p1 ¼ 36:54 ð36:18; 36:9Þ

Table 2. p2 ¼ − 276:8 ð−283:9; − 269:7Þ
Regression properties Goodness of fit Sum square error (SSE): 343700ðml 2 Þ
of experimental Coefficient of determination (R2): 0:9838
characteristic with Adjusted R2: 0:9838
linear regression Root mean square error (RMSE): 22:94ðmlÞ

With the relatively large number of observations, the regression line has relatively good model
compatibility with a correlation coefficient R ¼ 0:9838. However, the linear model could not be
used because the intercept value is −279 ðmlÞ, which is relatively far from the actual exhausted
air volume value of 0 ðmlÞ at the home position of the grippers. Therefore, a segmented linear
regression model is proposed. This method allows interpolating the experimental characteristic
of α and VT into a piecewise linear model, which goes through the home position and still keeps
the simplicity of a linear regression model. First, we look into a linear regression model that
passes through a point M with known coordinates. Considering a basic linear regression model
with one predictor variable, the model is given as follows (Neter et al., 1983):
Yi ¼ β0 þ β1 Xi þ εi (10)

where Yi is the response variable value at the ith trial; β0 and β1 are linear function coefficients;
Xi is the predictor variable value at the ith trial; εi is a random error.
The estimated linear regression model is expressed as:
Yb ¼ b0 þ b1 X (11)

where Yb is the estimated value of the regression function at the predictor variable of X; b0 and
b1 are point estimators of β0 and β1.
 For n is the number of observations, the sum of n square deviations is denoted as: Linear
Xn
regression
Q¼ ðYi  b0  b1 Xi Þ2 (12) model of simple
i¼1
ventilator
The least-square method indices that ðb0 ; b1 Þ is the minimum point of the function Q. Assume
that the linear regression model goes through point M ðm1 ; m2 Þ with all known m1 and m2.
Differentiating equation(12) with respect to b1, the value of ðb0 ; b1 Þ can be obtained by solving 389
the following system of equations:
8
>
> m2 ¼ m1 b1 þ b0
< !
vQ X n X n Xn Xn
(13)
>
: vb1 ¼ −2 Xi ðYi  b0  b1 Xi Þ ¼ −2 Xi Yi  b0 Xi  b1 Xi ¼ 0
2
>
i¼1 i¼1 i¼1 i¼1

The pair of ðb0 ; b1 Þ of equation (13) can be determined by:

8 Xn Xn !
> X i Yi  m2 Xi
>
> b0 ¼ m2  m1 Xn i¼1
Xn i¼1
>
>
< X 2 þ m1 i¼1 Xi
i¼1 i
> Xn Xn (14)
>
> Xi Yi  m2 i¼1 Xi
>
> b 1 ¼ Xn i¼1
Xn
:
X 2 þ m1 i¼1 Xi
i¼1 i

Therefore, Q has only one minimum point ðb0 ; b1 Þ, and their value could be estimated
from equation (14). The segmented regression model is expressed as follows
(Golovchenko, 2004):
8
< b
Y ¼ b1 X þ b0 ðfor X ≤ Xbp Þ
(15)
:Y b ¼ a1 X þ a0 ðfor X > Xbp Þ

where Xbp is the grippers angle at the breakpoint; b1 ; a1 ; b0 ; a0 are the regression model
coefficients.
At the breakpoint, the regression is assumed to be continuous:
b1 Xbp þ b0 ¼ a1 Xbp þ a0 (16)
a0 ¼ b0 þ Xbp ðb1  a1 Þ (17)

By substituting equation (17) to equation (15), it yields:

8
<b
Y ¼ b1 X þ b0 ðfor X ≤ Xbp Þ
(18)
:Y b ¼ a1 X þ b0 þ Xbp ðb1  a1 Þ ðfor X > Xbp Þ

In the selection of Xpb, X is corresponding to α, Y is corresponding to VT . As demonstrated

from equations (10) to (14), it can determine only one linear regression line that goes through a
breakpoint. Therefore, for each breakpoint, it can find only one linear regression line for each
side. As a result, it can determine Xbp by increasing X from 0 to maxðXi Þ (the BVM’s operating
range), for each value of X, calculating Q according to equation (12). Eventually, Xbp is the
value of X corresponding to the minimum value of Q. The determination of Xbp is shown in
Figure 10.
IJIUS The minimum of Q is determined to be equal to 3:0425 3 105 corresponding to the value of
11,3 α ¼ 10:93o. Applying equations (14) and (18) with the breakpoint value Xbp ¼ 10:93, the
segmented linear regression model is depicted in Figure 11. Besides that, the segmented linear
model has a value of Q ¼ 3:0425 3 105, which is lower than that of 3:437 3 105, as presented
in Table 2. Therefore, the segmented model is shown to be more suitable for the BVM
ventilator than the ordinary linear regression model.
390

Figure 10.
Determination Xbp
from the lowest value
of Q

Figure 11.
The experimental
characteristic of
gripper angle and
exhaust air volume
with segmented linear
regression
The regression result of the experimental characteristics of α and VT applying segmented Linear
linear regression model for the left and right sides are expressed in Tables 3 and 4, regression
respectively.
It can be seen that with the same α value, the segmented linear regression model gives a
model of simple
lower value of VT than the theoretical linear regression. This can be explained by the ventilator
variation of the AMBU bags’ elasticity over time. In addition, it is apparent that the left side
regression model has a relatively low correlation coefficient R. The gaining of more data
points is necessary to improve the accuracy of the left-side regression model. It is quite 391
difficult to acquire data points having a low value of alpha (e.g. lower than approximately
108). This is mainly due to the pressure difference inside the BVM system, to be more specific,
the air pressure in the ventilator flexible tube and test lung is higher than the air pressure in
the AMBU bag during ventilating. These two subsystems are separated by a one-way valve
(duckbill valve). Therefore, in short, this requires the air pressure in the AMBU bag to exceed
a threshold value, then the air would circulate and generate flow rate. Furthermore, other
reasons that could be mentioned are the backlash of the gear transmission, the variation of
the AMBU bags’ elasticity over time. In the future, several approaches could be developed
such as testing with different paddles profile and changing value of PEEP, PIP threshold to
adjust the airway pressure. On the other hand, the right-side regression model has a relatively
good fit with an R coefficient of 0:9812. It is noted that the air volume delivers to the patients’
lungs per breath is required to be within the range from 350ml to 750ml (Bocklage et al., 2015),
which is inside the right-side regression model. Therefore, the regression result expressed in
Figure 11 can be applied for the designing of controllers in the future.

4. Conclusion
The most critical factor of the BVM ventilator is a correlation between the AMBU bag and the
grippers angle of the BVM ventilator. Because the performance of the BVM ventilator
depends mainly on the grippers’ controllability and the change in physical properties of the
AMBU bag, establishing the mathematical model of the relationship between grippers angle
and exhausted air volume is obtained critical to enhancing the BVM ventilators control
performance. Despite the fact that the coefficients of the simulated correlation equation

Linear model VT ðtÞ ¼ b1 3 αðtÞ þ b0

Coefficients (95% confidence bounds) b1 ¼ 10:2472 ± 1:4954

b0 ¼ 0 ± 14:4955 Table 3.
Goodness of fit SSE: 7794:1333ðml 2 Þ Regression properties
Coefficient of determination (R2): 0:4198 of experimental
Adjusted R2: 0:3974 characteristic for the
RMSE: 3:4548ðmlÞ left side

Linear model VT ðtÞ ¼ b1 3 αðtÞ þ b0

Coefficients (95% confidence bounds) b1 ¼ 37:5233 ± 0:398

b0 ¼ − 298:1276 ± 8:107 Table 4.
Goodness of fit SSE: 296456:8839ðml 2 Þ Regression properties
Coefficient of determination (R2): 0.9812 of experimental
Adjusted R2: 0:9811 characteristic for the
RMSE: 21:3071ðmlÞ right side
IJIUS between the gripper angle and the exhausted air volume differ from those of the experimental
11,3 correlation equation, the simulated correlation equation and the experimental correlation
equation have the same form within the confidence limit. Our research is useful in the future
for the development of the simple BVM ventilator with mass production.
For the following work, the mechanical of the BVM ventilator will be improved to
amend the result of experimental characteristics for the left side. The R-square will be
improved to enable that the BVM ventilator can perform accurately the range of 0–700 ml;
392 thus, it is well suited for applying various animals except for humans. Additionally,
several experimental studies will be conducted for AMBU bags (for children) to determine
a correlation equation that compares the result of sample data. Based on the contribution of
this paper, an enhanced controller for the BVM ventilator will be developed to obtain high
accuracy performance.

References
Abdin, S.M., Elgendy, S.M., Alyammahi, S.K., Alhamad, D.W. and Omar, H.A. (2020), “Tackling the
cytokine storm in COVID-19, challenges and hopes”, Life Sciences, Vol. 257 July, 118054, doi: 10.
1016/j.lfs.2020.118054.
Alimohamadi, Y., Sepandi, M., Taghdir, M. and Hosamirudsari, H. (2020), “Determine the most
common clinical symptoms in COVID-19 patients a systematic review and meta-analysis”,
Journal of Preventive Medicine and Hygiene, Vol. 61, pp. E304-E312.
Area, I. and Nieto, J.J. (2021), “Forecasting the effects of the new SARS-CoV-2 variant in Europe”, The
Scientific World Journal, Vol. 2021, p. 5553240.
Bime, C., Fiero, M., Lu, Z., Oren, E., Berry, C.E., Parthasarathy, S. and Garcia, J.G.N. (2017), “High
positive end-expiratory pressure is associated with improved survival in obese patients with
acute respiratory distress syndrome”, American Journal of Medicine, Vol. 130 No. 2, pp. 207-213,
doi: 10.1016/j.amjmed.2016.09.029.
Bindayna, K.M. and Crinion, S. (2021), “Variant analysis of SARS-CoV-2 genomes in the Middle East”,
Microbial Pathogenesis, Vol. 153, p. 104741.
Blakeman, T.C. and Branson, R.D. (2013), “Inter- and intra-hospital transport of the critically ill”,
Respiratory Care, Vol. 58 No. 6, pp. 1008-1023.
Bocklage, T., Balk, R.A. and Bailey, C.J. (2015), “Setting the tidal volume in adults receiving
mechanical ventilation: lessons learned from recent investigations”, available at: http://www.
ncbi.nlm.nih.gov/pubmed/15003161%5Cnhttp://cid.oxfordjournals.org/lookup/doi/10.1093/cid/
cir991%5Cnhttp://www.scielo.cl/pdf/udecada/v15n26/art06.pdf%5Cnhttp://www.scopus.com/
inward/record.url?eid52-s2.0-84861150233&partnerID5tZOtx3y1.
Ercole, A., Taylor, B.L., Rhodes, A. and Menon, D.K. (2009), “Modelling the impact of an influenza
A/H1N1 pandemic on critical care demand from early pathogenicity data: the case for sentinel
reporting”, Anaesthesia, Vol. 64 No. 9, pp. 937-941, doi: 10.1111/j.1365-2044.2009.06070.x.
Fang, Z., Li, A.I., Wang, H., Zhang, R., Mai, X. and Pan, T. (2020), “AmbuBox: a fast-deployable low-
cost ventilator for COVID-19 emergent care”, SLAS Technology, Vol. 25 No. 6, pp. 573-584,
doi: 10.1177/2472630320953801.
Feng, D., Liu, S., Bai, Y., Tang, Y., Han, P. and Wei, W. (2020), “Management of urology during
COVID-19 pandemic: a perspective from Sichuan Province, China”, International Journal of
Surgery, Vol. 81 May, pp. 115-121, doi: 10.1016/j.ijsu.2020.08.005.
Fisher, D. and Heymann, D. (2020), “Q&A: the novel coronavirus outbreak causing COVID-19”, BMC
Medicine, Vol. 18 No. 1, pp. 18-20, doi: 10.1186/s12916-020-01533-w.
Forman, R., Atun, R., McKee, M. and Mossialos, E. (2020), “12 Lessons learned from the management
of the coronavirus pandemic”, Health Policy, Vol. 124 No. 6, pp. 577-580, doi: 10.1016/j.healthpol.
2020.05.008.
Gebremichael, M., Borg, U., Habashi, N.M., Cottingham, C., Cunsolo, L., McCunn, M. and Reynolds, Linear
H.N. (2000), “Interhospital transport of the extremely ill patient: the mobile intensive care unit”,
Critical Care Medicine, Vol. 28 No. 1, pp. 79-85, doi: 10.1097/00003246-200001000-00013. regression
Gereffi, G. (2020), “What does the COVID-19 pandemic teach us about global value chains the case of
model of simple
medical supplies”, Journal of International Business Policy, Vol. 3, pp. 287-301. ventilator
Golovchenko, N. (2004), “Least-squares fit of a continuous piecewise linear function”.
Haleem, A., Javaid, M. and Vaishya, R. (2020), “Effects of COVID-19 pandemic in daily life”, Current 393
Medicine Research and Practice, Vol. 10 No. 2, pp. 78-79, doi: 10.1016/j.cmrp.2020.03.011.
Jenco, M. (2020), “COVID-19 less severe in children than adults: study, AAP News”, available at:
https://www.aappublications.org/news/2020/03/16/coronavirus031620 (accessed 19 July 2021).
Kannan, S., Shaik Syed Ali, P., Sheeza, A. and Hemalatha, K. (2020), “COVID-19 (Novel Coronavirus
2019) - recent trends”, European Review for Medical and Pharmacological Sciences, Vol. 24 No. 4,
pp. 2006-2011.
Kim, K.M., Cinti, S., Gay, S., Goold, S., Barnosky, A. and Lozon, M. (2012), “Triage of mechanical
ventilation for pediatric patients during a pandemic”, Disaster Med Public Health, Vol. 6 No. 2,
pp. 131-137, doi: 10.1097/01.naj.0000529715.93343.b0.
Lagasse, J. (2020), “COVID-19: ventilators are in short supply, but so are medications for ventilator
patients. j Healthcare finance news, healthcare IT news”, available at: https://www.
healthcarefinancenews.com/news/covid-19-ventilators-are-short-supply-so-are-medications-
ventilator-patients (accessed 25 April 2021).
Lapostolle, F., Goix, L., Vianu, I., Chanzy, E., De Stefano, C., Gorlicki, J., Petrovic, T. and Adnet, F.
(2020), “COVID-19 epidemic in the Seine-Saint-Denis Department of Greater Paris: one month
and three waves for a tsunami”, European Journal of Emergency Medicine, Vol. 27, pp. 274-278,
doi: 10.1097/MEJ.0000000000000723.
Li, Y. and Nielsen, P.V. (2011), “Commemorating 20 years of indoor air: CFD and ventilation research”,
Indoor Air, Vol. 21 No. 6, pp. 442-453, doi: 10.1111/j.1600-0668.2011.00723.x.
Li, X., Song, Y., Wong, G. and Cui, J. (2020), “Bat origin of a new human coronavirus: there and back
again”, Science China Life Sciences, Vol. 63 No. 3, pp. 461-462, doi: 10.1007/s11427-020-1645-7.
Lurie, N., Saville, M., Hatchett, R. and Halton, J. (2020), “Developing covid-19 vaccines at pandemic
speed”, New England Journal of Medicine, Vol. 382, pp. 1969-1973, available at: nejm.org.
MacLaren, G., Fisher, D. and Brodie, D. (2020), “Preparing for the most critically Ill patients with
COVID-19: the potential role of extracorporeal membrane oxygenation”, JAMA, Vol. 323 No. 13,
pp. 1245-1246.
Morawska, L., Tang, J.W., Bahnfleth, W., Bluyssen, P.M., Boerstra, A., Buonanno, G., Cao, J., Dancer,
S., Floto, A., Franchimon, F., Haworth, C., Hogeling, J., Isaxon, C., Jimenez, J.L., Kurnitski, J., Li,
Y., Loomans, M., Marks, G., Marr, L.C., Mazzarella, L., Melikov, A.K., Miller, S., Milton, D.K.,
Nazaroff, W., Nielsen, P.V., Noakes, C., Peccia, J., Querol, X., Sekhar, C., Seppänen, O., Tanabe,
S.I., Tellier, R., Tham, K.W., Wargocki, P., Wierzbicka, A. and Yao, M. (2020), “How can airborne
transmission of COVID-19 indoors be minimised”, Environment International, Vol. 142, pp. 1-7.
Neter, J., Wasserman, W. and Kutner, M. (1983), “Applied linear regression models”.
Nigrovic, P.A. (2021), “COVID-19 cytokine storm: what is in a name?”, Annals of the Rheumatic
Diseases, Vol. 80 No. 1, pp. 3-5, doi: 10.1056/nejmoa2021436.
Otten, D., Liao, M.M., Wolken, R., Douglas, I.S., Mishra, R., Kao, A., Barrett, W., Drasler, E., Byyny,
R.L. and Haukoos, J.S. (2014) (2014), “Comparison of bag-valve-mask hand-sealing techniques in
a simulated model”, Annals of Emergency Medicine, Vol. 63 No. 1, pp. 6-12, doi: 10.1016/j.
annemergmed.2013.07.014.
Ouyang, H. (2020), “I’m an E.R. Doctor in New York. None of us will ever Be the same”, New York
times, available at: https://www.nytimes.com/2020/04/14/magazine/coronavirus-er-doctor-diary-
new-york-city.html (accessed 19 July 2021).
IJIUS Padula, W.V. (2020), “Why only test symptomatic patients consider random screening for COVID-19”,
Applied Health Economics and Health Policy, Vol. 18, pp. 333-334.
11,3
Pinhu, L., Whitehead, T., Evans, T. and Griffiths, M. (2003), “Ventilator-associated lung injury”,
Lancet, Vol. 361 No. 9354, pp. 332-340, doi: 10.1016/S0140-6736(03)12329-X.
Ramanathan, K., Antognini, D., Combes, A., Paden, M., Zakhary, B., Ogino, M., MacLaren, G., Brodie,
D. and Shekar, K. (2020), “Planning and provision of ECMO services for severe ARDS during
the COVID-19 pandemic and other outbreaks of emerging infectious diseases”, The Lancet
394 Respiratory Medicine, Vol. 8 No. 5, pp. 518-526, doi: 10.1016/S2213-2600(20)30121-1.
Ranney, M.L., Griffeth, V. and Jha, A.K. (2020), “Critical supply shortages — the need for ventilators
and personal protective equipment during the covid-19 pandemic”, The New England Journal of
Medicine, Vol. 382, p. e41.
Russell, J. and Slutsky, A. (1999), “International consensus conferences in intensive care medicine:
ventilator-associated lung injury in ARDS”, American Journal of Respiratory and Critical Care
Medicine, Vol. 160 No. 6, pp. 2118-2124, doi: 10.1164/ajrccm.160.6.ats16060.
Schuchat, A. (2020), “Public health response to the initiation and spread of pandemic COVID-19 in the
United States, February 24–April 21, 2020”, The COVID-19 Reader, pp. 142-151, doi: 10.4324/
9781003141402-16.
Secured Medical Direction UK (2019), “Manual resuscitator bag (Ambu bag), smd-medical”, available
at: https://www.smd-medical.com//catalogue/airway-management/30-Manual-Resuscitator-
catalogue.pdf.
Smetanin, P., Stiff, D., Kumar, A., Kobak, P., Zarychanski, R. and Simonsen, F.P. (2009), “Potential intensive
care unit ventilator demand/Canada”, Journal of Infectious Diseases, The, Vol. 20 No. 4, pp. 115-123.
Stiff, D., Kumar, A., Kissoon, N., Fowler, R., Jouvet, P., Skippen, P., Smetanin, P., Kesselman, M. and
Veroukis, S. (2011), “Potential pediatric intensive care unit demand/capacity mismatch due to
novel pH1N1 in Canada”, Pediatric Critical Care Medicine, Vol. 12 No. 2, pp. 51-57, doi: 10.1097/
PCC.0b013e3181e2a4fe.
Sun, X., Wang, T., Cai, D., Hu, Z., Chen, J., Liao, H., Zhi, L., Wei, H., Zhang, Z., Qiu, Y., Wang, J. and
Wang, A. (2020), “Cytokine storm intervention in the early stages of COVID-19 pneumonia”,
Cytokine and Growth Factor Reviews, Vol. 53, pp. 38-42.
Tang, Y., Liu, J., Zhang, D., Xu, Z., Ji, J. and Wen, C. (2020), “Cytokine storm in COVID-19: the current
evidence and treatment strategies”, Frontiers in Immunology, Vol. 11 July, pp. 1-13, doi: 10.3389/
fimmu.2020.01708.
The Lancet (2020), “COVID-19: too little, too late?”, The Lancet, Vol. 395 No. 10226, p. 755, doi: 10.1016/
S0140-6736(20)30522-5.
Truong, C.T., Huynh, K.H., Nguyen, H.H., Pham, L.A. and Nguyen, T.T. (2021), “Model-free volume
and pressure cycled control of automatic bag valve mask ventilator”, AIMS Bioengineering,
Vol. 8 No. 3, pp. 192-207, doi: 10.3934/bioeng.2021017.
Wang, J.H. and Liou, C.M. (1991), “Experimental identification of mechanical joint parameters”,
Journal of Vibration and Acoustics, Transactions of the ASME, Vol. 113 No. 1, pp. 28-36, doi: 10.
1115/1.2930151.
Warren, J., Fromm, R.E. Jr, Orr, R.A., Rotello, L.C. and Horst, H.M. (2004), “Guidelines for the inter -
and intrahospital transport of critically ill patients”, Critical Care Medicine, Vol. 32 No. 1,
pp. 256-262, doi: 10.1097/01.CCM.0000104917.39204.0A.
WHO (2020a), “Critical preparedness, readiness and response actions for COVID-19: WHO/2019-nCoV/
Community_Actions/2020.3”, (March), pp. 1-3, available at: https://www.who.int/publications-
detail/critical-preparedness-readiness-and-response-actions-for-covid-19.
WHO (2020b), “Naming the coronavirus disease (COVID-19) and the virus that causes it”, pp. 1-4,
available at: https://www.who.int/emergencies/diseases/novel-coronavirus-2019/technical-
guidance/naming-the-coronavirus-disease-(covid-2019)-and-the-virus-that-causes-it.
Wiwanitkit, V., (2020), “Atypical modes of COVID-19 transmission: how likely are they?”, Linear
Epidemiology and Health, Vol. 42, p. e2020059.
regression
World Health Organization (2021), “WHO Coronavirus (COVID-19) dashboard j WHO Coronavirus
(COVID-19) dashboard with vaccination data, World Health Organization”, available at: https://
model of simple
covid19.who.int/?gclid5Cj0KCQjwubHBhCyARIsAPctr7z3ZCkrGDPptsZ5tACqLtiDeL ventilator
GeqoxiHTzezQHUVW8EtmiQUX7t3qgaAuq_EALw_wcB (accessed 19 July 2021).

395
Further reading
Kwon, A.H., Slocum, A.H. Jr, Varelmann, D., Nabzdyk, C.G.S. and MIT E-Vent Team (2020), “Rapidly
scalable mechanical ventilator for the COVID-19 pandemic”, Intensive Care Medicine, Vol. 46
No. 8, pp. 1642-1644. doi: 10.1007/s00134-020-06113-3.

Corresponding author
Van Tu Duong can be contacted at: [email protected]

For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: [email protected]