Download as pdf or txt
Download as pdf or txt
You are on page 1of 60

KQB7002- Bioinstrumentation

Assignment:
Case Study - Commercial Testing Devices Related to Covid-19

Date of Submission: 16 December 2021


Lecturer: Profesor Ir. Dr. Fatimah Binti Ibrahim

Group 1

Group Member Student ID

Leong Yew Sum S2039447

Chai Ming Hooi S2029201

Wafa Habib S2100137


Table of Contents

Table of Contents i
List of Figures iii
List of Tables iv
SECTION I: COVID-19 Detection Systems 1
Chapter 1: Polymerase Chain Reaction (PCR) 1
1.1 Introduction to PCR 1
1.2 Basic Working Principle of PCR 1
1.3 Variations of PCR 2
1.3.1 Variation of PCR: RT-PCR 2
1.3.2 Variation of PCR: qPCR 3
1.4 Application of RT-qPCR in COVID-19 Detection 4
1.5 Comparison of Conventional PCR and RT-qPCR - Advantages and Limitations 4
1.6 Normalization Techniques of RT-qPCR 5
Chapter 2: Rapid Test Kit - Antigen (RTK-Ag) 7
2.1 Introduction to RTK 7
2.2 Basic Working Principle of RTK-Ag for COVID-19 Detection 8
2.3 Variations of RTK-Ag 10
2.3.1 Variation of RTK-Ag Test Kit - Electrochemically Based 10
2.3.2 Variation of RTK-Ag Test Kit - Utilizing Field-effect transistor (FET) 11
2.3.3 Variation of RTK-Ag Test Kit - Surface Acoustic Wave (SAW) 12
2.4 Assessing Sensitivity of RTK-Ag 13
2.5 Comparison of RT-qPCR Test and RTK-Ag Test (Malaysia) 14
Chapter 3: Saliva Self-test kit 16
3.1 Introduction to Saliva Test Kit 16
3.2 Basic Working Principle of Saliva Test Kit 18
SECTION 2: Breath-based COVID-19 Tests 22
Chapter 4: Comparison of Technologies 22
4.1 GeNose C19 - Indonesia 22
4.1.1 Introduction to GeNose C19 - Indonesia 22
4.1.2 Working Principle of GeNose 23
4.2 BreFence Go COVID-19 Breath Test System - Breahonix, Singapore 24
4.2.1 Introduction to BreFence Go COVID-19 Breath Test System - Breahonix,
Singapore 24
4.2.2 Working Principle of BreFence Go System 25
4.3 Trace Virus Detection (TVD)-2 Breathalyser - GreyScan, Australia 25
4.3.1 Introduction to Trace Virus Detection (TVD)-2 Breathalyser - GreyScan,
Australia 25
4.3.2 Working Principle of TVD 27
4.4 CoronaCheck - Exhalation Technology (ETL), United Kingdom (UK) 27
4.4.1 Introduction to CoronaCheck - Exhalation Technology (ETL), United
Kingdom (UK) 27
4.4.2 Working Principle of CoronaCheck 28
4.5 Comparison of breath-based COVID-19 point-of-care tests 31
Chapter 5: Improvement for Breath-based COVID-19 tests 34
5.1 Introduction 34
5.2 Objective 36
5.3 Proposed Design 36
5.3.1 General Working Principle 36
5.3.2 Sample Pre-processing - Additional Installation of Precooler Unit 38
5.3.3 Data Integration and Machine Learning 40
5.4 Costing 43
5.5 Advantages and Limitations of the New Design 44
5.5.1 Advantages 44
5.5.2 Limitations 45
References 46
Appendices 54
Appendix A: Request for TVD-2 Breathanalyser Information via Email to GreyScan 54
Appendix B: Request for BreFence Information via Email to Brethnoxix 54
Appendix C: Reply on CoronaCheck Information via Email from ETL 55
List of Figures

Figure 1: Block diagram of working principle for PCR 1


Figure 2: Molecular illustration of polymerase chain reaction 2
Figure 3: Schematic diagram of dye-based (left) and probe-based (right) qPCR 3
detection
Figure 4: Block diagram for COVID-19 detection using RT-qPCR 4
Figure 5: SARS-CoV-2 antigen-detecting rapid-diagnostic-tests algorithm 7
Figure 6: General illustration of a typical biosensor (block diagram) 8
Figure 7: Coronavirus antigen rapid test kit 9
Figure 8: Process flow diagram for COVID-19 detection using RTK-Ag Test Kit 9
Figure 9: Schematic illustration of designing an electrochemical immunosensor 10
for SARS-CoV-2 detection
Figure 10: Circuitry for an electrochemically-based RTK-Ag 11
Figure 11: FET sensing operation procedure for COVID-19 in schematic 12
diagram(top) and Response of COVID-19 FET toward SARS-CoV-2
antigen protein in universal transport media and related
dose-dependent response curve in real-time (bottom)
Figure 12: Block diagram of instrument having SAW sensor as the detector 12
Figure 13: Schematic illustration of antigen detection-based LFIA Test 18
Figure 14: Block Diagram of a Saliva self-test kit 19
Figure 15: GeNose device 22
Figure 16: Circuitry connection of GeNose System 23
Figure 17: Block Diagram of GeNose System 23
Figure 18: Schematic illustration of BreFence Go System 25
Figure 19: TVD-2 Breathalyser 26
Figure 20: Working Principle of TVD-1 27
Figure 21: CoronaCheck 28
Figure 22: Exhaled Breath offers a powerful medium for collection of a range of 29
biomarkers and pathogens found in respiratory tract
Figure 23: Working Principle of Corona Check 30
Figure 24: Block diagram of proposed design. Red boxes indicate design 37
improvement
Figure 25: Circuitry connection of proposed design. Red colour indicates design 37
improvement
Figure 26: Example sample gas cooler for pre-cooling gas sample in our design 39
Figure 27: Integration of Artificial Intelligence in classifying result of VOC 41
Figure 28: Google Colaboratory platform by Google 41
Figure 29: Example of COVID-19 test result updates at MySejahtera application 43

List of Tables

Table 1: Comparison of conventional PCR and RT-qPCR 5


Table 2: Normalization of PT-qPCR 5
Table 3: Comparison of RTK-Ag across different countries 13
Table 4: Comparison of RT-qPCR and RTK-Ag Test 14
Table 5: A comparison of the sensitivity, specificity rates and result time of a few 20
of the Saliva Self-test Kits approved for sale in pharmacies in Malaysia
Table 6: Advantages and disadvantages of saliva test kit 21
Table 7: Comparison of four breath-based COVID-19 point-of-care tests 31
Table 8: Comparison of PCR, RTK, and GeNose in Indonesia 32
SECTION I: COVID-19 Detection Systems

Chapter 1: Polymerase Chain Reaction (PCR)

1.1 Introduction to PCR

As the coronavirus that causes the COVID-19 disease spreads over the world, many nations
are using polymerase chain reaction (PCR), one of the most accurate laboratory methods for
detecting, tracking, and investigating the COVID-19 coronavirus. Previously, other diseases
such as the Ebola and Zika viruses, have also been diagnosed using PCR tests (Jawerth,
2020). PCR tests can be further branched into several variations, such as real time reverse
transcription–polymerase chain reaction (real time RT–PCR) and quantitative PCR (qPCR)
tests. The most common method of detection is reverse-transcription quantitative PCR
(RT-qPCR).
The polymerase chain reaction (PCR) can be understood as one of the methods of
detecting a specific microorganism such as viruses. During the pandemic, PCR test was used
for detection of COVID-19 viruses, specifically ribonucleic acid or RNA of SARS-CoV-2.
During the test, bits of RNA from the specimen are amplified into deoxyribonucleic acid
(DNA), which is then replicated until SARS-CoV-2 is identifiable if present, using PCR
technique. PCR test remains one of the stable methods of COVID-19 diagnoses ever since it
obtains acknowledgement and approval in 2020 (Cleveland Clinic, 2021).

1.2 Basic Working Principle of PCR

With reference from (Kadri, 2019), PCR procedure is split into three steps: Denaturation,
Annealing and Extension. The block diagram for the process of PCR is depicted in Figure 1.

Figure 1: Block diagram of working principle for PCR

The first stage is denaturation, which is achieved by raising the temperature and
separating the two strands of DNA. The first period is carried out at 94°C, which is known as
the denaturation temperature. At this temperature, the matrix DNA, which acts as the
replication matrix, is denatured and converted into single-stranded DNA (single-stranded
DNA).

1
The hybridization process is the next phase. It is performed at a temperature of 40 to
70°C, known as the primer hybridization temperature. As the temperature drops, the
hydrogen bonds rebuild, allowing the complementary strands to hybridise. The primers,
which are short single-strand sequences that are complementary to the flanking sections of
the DNA to be amplified, hybridise more easily than the long strand matrix DNA.
The third stage is carried out at 72°C, often known as the elongation temperature. It is
the synthesis of the complementary strand. Taq polymerase attaches to the single-stranded
DNAs at 72°C and catalyses replication using deoxyribonucleoside triphosphates from the
reaction mixture. The downstream portions of the template DNA are so preferentially
produced.

Figure 2: Molecular illustration of polymerase chain reaction (Karki, 2017)

The reaction is returned to the denaturation phase after extension, and the PCR
procedure is repeated. Because a new strand of DNA acts as a template for replication in the
following cycle, each cycle roughly doubles the quantity of DNA. The amount of DNA
grows exponentially as a result of this. Summary of molecular illustration for PCR is depicted
in Figure 2.

1.3 Variations of PCR

1.3.1 Variation of PCR: RT-PCR

The real-time polymerase chain reaction (RT–PCR) is a derivative of the polymerase


chain reaction (PCR). The two techniques are similar, but RT–PCR adds a reverse
transcription of RNA to DNA, or RT, to allow amplification. RNA is used as a template in
RT-PCR, in which the template's quality and purity is critical for the performance of the test.

2
Firstly, a DNA hybrid is created by using RNase H, a reverse transcriptase that degrades the
RNA in the hybrid and converts DNA into complementary DNA (cDNA). The cDNA is then
amplified using the conventional PCR technique (Neidler, 2017).
RT-PCR can be carried out in one or two phases. In one-step RT-PCR, the RT reaction
and the PCR reaction are both carried out in the same tube. In a two-step RT-PCR technique,
the produced cDNA is transferred into a second tube for PCR. Oligo (dT), random hexamer,
or gene-specific primers can be used to accelerate the process. One-step reactions are easier
to set up and are better suited to high throughput screening. Two-step reactions are
appropriate for identifying several messages in a single RNA sample (Jalali, Zaborowska, &
Jalali, 2017).

1.3.2 Variation of PCR: qPCR

In recent years, qPCR, a PCR technology advancement, is becoming increasingly


significant in clinical diagnosis. As the reaction develops, qPCR enables the identification
and quantification of the target DNA. Reverse transcription followed by quantitative
polymerase chain reaction analysis, or qRT-PCR, is a highly sensitive and low-cost approach
for measuring gene transcripts from cells (Udvardi, Czechowski, & Scheible, 2008).
RT-qPCR amplifies cDNA in three stages, like in regular PCR: denaturation, annealing, and
elongation.

Figure 3: Schematic diagram of dye-based (left) and probe-based


(right) qPCR detection (Cao et al., 2020)
There are two kinds of qPCR: dye-based and probe-based. SYBR Green I is the most
widely used fluorescent DNA binding dye. When attached to double-stranded DNA, it
produces fluorescence, and the intensity of the fluorescence rises according to the
concentration of PCR product (Jalali, Zaborowska, & Jalali, 2017). Using probe-based qPCR,

3
several targets may be recognised simultaneously in each sample, but this needs the
optimization and creation of a target-specific probe(s) in addition to primers. The most
common type of probe is a hydrolysis probe, which employs the use of a fluorophore and a
quencher (Neidler, 2017).

1.4 Application of RT-qPCR in COVID-19 Detection

The most common method for detecting a coronavirus infection and diagnosing COVID-19 is
to use molecular assays, which identify virus genes in patient samples. All of these tests are
Nucleic Acid Amplification Tests (NAATs), and they all rely on reverse transcription (RT) of
viral RNA into DNA, followed by PCR amplification and detection of the DNA.
Reverse-transcription quantitative PCR (RT-qPCR) is the most popular technique of detection
(Pohl, 2020). The block diagram for the overall process of RT-qPCR is illustrated in Figure 4.

Figure 4: Block diagram for COVID-19 detection using RT-qPCR

To collect viral particles and virus-infected mucosa cells, a nose or throat swab is
used. After that, the swab sample is lysed. Viruses can be freed from the host cell by lysis,
which destroys the cell by shattering its membrane and, if present, its cell wall. Many
bacterial and animal viruses have this property. Moving on, the viral RNA is removed, and
the cDNA is reverse-transcribed. The presence of viral genome and confirmation of infection
with the virus are detected using qPCR on the cDNA in the processed sample.

1.5 Comparison of Conventional PCR and RT-qPCR - Advantages and Limitations

Because of its simplicity, reproducibility, and high throughput, reverse transcription


quantitative PCR (RT-qPCR) is frequently used in gene expression studies. By comparing the
expression of target genes to reference genes across several samples, it provides a quick way
to study gene variation in distinct organisms/organs. Furthermore, RT-qPCR has become a
low-cost approach, allowing its usage in laboratories to grow more widespread. Several
parameters, such as sample volume, RNA integrity, and cDNA quality, must be controlled

4
when analysing RT-qPCR data because they impact the quantitative determination of gene
expression. The comparison of RT-qPCR relative to PCR is illustrated in Table 1.

Table 1: Comparison of conventional PCR and RT-qPCR


Conventional PCR Real Time qPCR

Advantage - Early confirmation diagnosis - Allows detection and


- Multiplexing potential amplification at the same time
- Widely used molecular diagnosis - Monitors the amplification in
format real time.
- Quantitative
- Lower contamination
- Increased sensitivity

Limitation - Qualitative - Expensive detection


- Post PCR handling leads to equipments compared to
contamination conventional PCR
- Missing borderline cases of low - Requires fluorescent dye /
gene copy numbers results in less probe
sensitivity
- Requires more time to process the
result

1.6 Normalization Techniques of RT-qPCR

Although RT-qPCR is frequently employed to justify physiologically relevant differences in


mRNA levels, it has a few drawbacks. These include RNA's intrinsic variability, variations in
extraction techniques that might copurify inhibitors, and varying reverse transcription and
PCR efficiency. Several ways for normalising real-time RT-qPCR data have been presented in
Table 2.

Table 2: Normalization of PT-qPCR (Hugget et al., 2015)


Normalisation Pros Cons
strategy

It may be difficult to assess sample


Identical sample
Relatively simple. size, and/or it may not be biologically
size
representative.

Errors introduced during reverse


Ensures that reverse
Total RNA transcription and PCR are not
transcriptase input is consistent.
controlled.

Give an estimate of the size of


Genomic DNA RNA extraction is difficult.
the cellular sample.

5
Internal control that is exposed Validation must be done using the
Reference genes
to the same circumstances as same experimental samples. The
ribosomal RNAs
the target mRNA. Real-time inaccuracy of the reference gene
(rRNA)
data is also taken into account. determines the assay resolution.

Internal control that is exposed Validation must be done using the


Reference genes
to the same circumstances as same experimental samples. The
messenger
the target mRNA. Real-time inaccuracy of the reference gene
RNAs (mRNA)
data is also taken into account. determines the assay resolution.

Internal control that is


susceptible to the same It must be found, cloned, or
circumstances as the target synthesised. The RNA of interest, on
Alien molecules
mRNA. It doesn't have the the other hand, is not extracted from
biological diversity of a within the cells.
reference gene.

6
Chapter 2: Rapid Test Kit - Antigen (RTK-Ag)

2.1 Introduction to RTK

Notwithstanding the fact that RT-PCR remains as the gold standard for identification of the
SARS-CoV-2 virus, an alternative diagnosis method such as Rapid Test Kit - Antigen
(RTK-Ag) has become more widely accepted due to faster result and cheaper cost. This in
turn assisted with faster decisions for subsequent contact tracing, isolation and treatment of
the COVID-19 patients. The use of RTK-Ag for diagnosis of COVID-19 is guided by the
World Health Organization (WHO). Figure 5 shows the algorithm for diagnosis of the disease
based on either symptomatic case, suspected outbreak setting or asymptomatic case.

Figure 5: SARS-CoV-2 antigen-detecting rapid-diagnostic-tests algorithm (World Health


Organization, 2021)
As of 29 October 2021, the Malaysian Ministry of Health has listed 119 RTK-Ag
products for recommendation for professional use within Malaysia (Ministry of Health
Malaysia, 2021). These test kits were manufactured in China (67 products), South Korea (25
products), Malaysia (8 products), the USA (8 products), Germany (4 products), Canada (2
products), India (2 products) and Finland, Saudi Arabia and Singapore (1 product each). The
sample types for these test kits are deep throat saliva (DTS), nasopharyngeal (NPS) or
oropharyngeal (OPS). In general, an RTK-Ag biosensor consists of receptors, a transducer
and a reader. The receptors, or recognition molecules, are specific to identify the disease by
recognizing the analyte; the transducer converts the recognition event into a measurable
signal; and the reader gives the output result understood by the user (Rasmi et al., 2021).

7
2.2 Basic Working Principle of RTK-Ag for COVID-19 Detection

The underlying working principle of a COVID-19 RTK-Ag biosensor is not different from
the general working principle of a biosensor whereby the process begins with the analyte
diffusing to the surface of the biosensor where the recognition molecules are immobilized.
Once the analyte reacts with the recognition molecule, the reaction changes the
physicochemical properties of the transducer surface. This leads to a change in the properties
of the transducer surface, and depending on the type of transducer the type of properties
change is measured or converted into an electrical signal which is then amplified, processed
and displayed. See the block diagram depicted in Figure 6 below. Further descriptions for the
different types of transducers for a COVID-19 RTK-Ag biosensor is provided in Sections
2.3.1 through 2.3.3.

Figure 6: General illustration of a typical biosensor (block diagram)


The analyte in this case is a specific part of the SARS-CoV-2 virus. The spike (S)
protein, the nucleocapsid (N) protein, the membrane (M) protein, and the envelope (E)
protein are all required for the production of a structurally complete viral particle and are
encoded by the coronaviral genome (Schoeman & Fielding, 2019). In the design of a
COVID-19 RTK-Ag biosensor, the common target antigen biomarkers are the N and S
proteins because the N protein is involved in the viral RNA replication and assembly and is
high in level during the early stage of infection, and the S protein facilitates viral entry into
infected cells. Choice of recognition molecules for the RTK-Ag biosensor design generally
considers its small size relative to the virus particle, ease of production and affinity to the
target antigen (Limsakul et al., 2021).

8
Figure 7: Coronavirus antigen rapid test kit
The coronavirus antigen rapid test kit is a qualitative lateral flow assay for identifying
N protein in upper respiratory samples. When a specimen is put on the sample pad of a test
cassette, it interacts with colloidal gold-labeled SARS-CoV-2 N protein (Rabbit monoclonal)
antibodies to form an antibody-antigen (Ab-Ag) complex. The Ab-Ag complex migrates to
the test line under capillary action, and the SARS-CoV-2 N protein antibody catches it. The
test line will generate a red band if the sample is positive for COVID-19 nucleocapsid
protein. If the material does not include any coronavirus antigen (N protein) or the antigen
level is below the detection limit, no colour band will develop on the test line. The process
flow diagram for COVID-19 detection using the RTK test kit is depicted in Figure 8.

Figure 8: Process flow diagram for COVID-19 detection using RTK-Ag Test Kit

The concept of this design is applied in the study of Iravani (2020), where the
detection of the virus’s spike (S1) protein was carried out by applying the cellular angiotensin
converting enzyme 2 (ACE 2) receptor that produces a matched pair with antibody mAb. The
ACE 2 and S1-mAb were paired with each other for capture and detection using lateral flow
immunoassay. The design included a sample pad, conjugate pad, nitrocellulose membrane
and absorbent pad. Nasal swab sample from a subject is placed onto the sample pad. As it
diffuses from the conjugate pad at one end to the absorbent pad at the other end, it crosses the
test line and control line. The test line contains ACE 2 for detection of the SARS-CoV-2
spike antigen, and the control line contains the anti-IgG antibody.

9
2.3 Variations of RTK-Ag

2.3.1 Variation of RTK-Ag Test Kit - Electrochemically Based

Figure 9: Schematic illustration of designing an electrochemical immunosensor for


SARS-CoV-2 detection (Singh et al., 2021)
With reference from Singh et al. (2021), the example biosensor is electrochemically-based.
As illustrated in Figure 9, graphene is deposited with gold nanoparticles to increase the
surface area and conductivity. As denoted by d in the same illustration, the surface is
immobilized with an antibody that will detect S protein in the SARS-Cov-2 virus. If the virus
is present in the sample, then its S protein will conjugate with the antibody. This recognition
event, i.e. conjugation between the S protein and the antibody molecule, changes the
distribution of charge in the potentiometer set-up (denoted by g in the above diagram).
Changes in current flow due to this conjugation event changes the potential and is measured.
The signal is then processed by correlating the amount of the recognition event to the
measured potential, and consequently read out by the computer screen. In this example, the
analyte is the S protein from the SARS-CoV-2 virus, the sensing element is the antibody
deposited on the electrode, the transducer is the potentiometer, and the computer carries out
the signal processing, and the reader is the computer screen.

10
Figure 10: Circuitry for an electrochemically-based
RTK-Ag
The circuitry functionality of an electrochemically-based RTK-Ag can be represented
by Figure 10. As shown, a voltage is applied across the semiconductor to produce a current.
Reaction between the analyte (the virus’s S protein) and the biorecognition element (the
antibody) incurs changes to the charge distribution on silicon dioxide which results with
changes in the current. This electrical signal can thus be measured, or amplified, processed
and displayed to show the system’s positive detection of the virus.

2.3.2 Variation of RTK-Ag Test Kit - Utilizing Field-effect transistor (FET)

As presented by Seo et al. (2020), the biosensing device uses a field-effect transistor (FET) to
detect the S protein in the SARS-CoV-2 virus. See Figure 11 on the overleaf page. In this
design, graphene sheets of FET are coated with an antibody specific against the S protein in
SARS-CoV-2. The system detects presence of the virus based on changes in the channel
surface potential and the corresponding effects on the electrical response. The FET sensor,
acting as a transducer, converts the signal to electrical current, amplified, processed and read
out by the computer screen. Thus, not only the virus is qualitatively detected, but it can be
quantitatively estimated based on the electrical current output.

11
Figure 11: FET sensing operation procedure for COVID-19 in
schematic diagram(top) and Response of COVID-19 FET toward
SARS-CoV-2 antigen protein in universal transport media and related
dose-dependent response curve in real-time (bottom) (Seo et al., 2020)

2.3.3 Variation of RTK-Ag Test Kit - Surface Acoustic Wave (SAW)

Figure 12: Block diagram of instrument having SAW sensor


as the detector (Shachar et al., 2021)
Alternatively, the RTK-Ag biosensor can be based on surface acoustic wave (SAW) as
presented in Shachar et al. (2021). Figure 12 shows the block diagram of an RTK-Ag having
an SAW sensor as the detector (Shachar et al., 2021). In this SAW-based biosensor, the
propagation loss of the wave along the membrane from input transducer to output transducer
is measured and is in correlation to the amount of the SARS-CoV-2 virus detection. In this
design example, the SAW reader is coupled to a sensor interface and provides an excitation
signal to the SAW package. The reader package included a USB interface, power, serial data

12
interface, display control and wireless communication. The microfluidic controller includes
processor control of a laser and disc motor. In this arrangement, the configuration performs
immunofluorescence detection using microarrays with a SAW sensor.

2.4 Assessing Sensitivity of RTK-Ag

RTK-Ag is able to detect COVID-19 disease in a more rapid, less laborious and less
expensive way. Numerous studies have been performed and reported by researchers from
around the globe with relation to RT-PCR as the reference model.

Table 3: Comparison of RTK-Ag across different countries


Country Assessment Ref.

Spain Baro et al.


(2021)

Germany Corman et
al. (2021)

Belgium - Product: Biotical, Panbio, Healgen and Roche Favresse


(immunochromatography based) and automated antigen et al.
detection test: Vitros (chemiluminescence based). (2021)
- Findings: With Ct values 25 and below, the four
RTK-Ag’s performed with sensitivities between 93.1%
and 96.6%; Vitros showed a 100% sensitivity for Ct value
33 and below and specificity 100%, thus it fully aligned
with RT-PCR.

Nether- - Product: Veritor by Becton Dickinson and Biosensor by Schuit et


lands Roche-Diagnostics. al. (2021)
- Among the symptomatic subjects, the sensitivities were
84.2% and 73.3% for Veritor and Biosensor respectively.
However, among the asymptomatic subjects the
sensitivities were lower, at 58.7% for Veritor and 59.4%
for Biosensor. In all analyses for both the Veritor and
Biosensor, specificities were more than 99%, positive
predictive values were more than 90% and negative
predictive values were more than 95%.

13
We can therefore conclude that RTK-Ag would probably be suitable for mass
screening of SARS-CoV-2 infection in the general population, especially where there are
limited resources to adopt the RT-PCR method whether due to the high volume of testing
required in that population at that time or due to absence of sufficient RT-PCR facilities.
Nevertheless, use of RTK-Ag does not come without caution. It has a narrow time
frame whereby it is predominantly useful within the first week of symptoms when the viral
load is reasonably high. This agrees with the findings by Singh et al. (2021) that viral load of
SARS-CoV-2 reduces after the first week of infection, thus the viral protein amount to be
detected also reduces and renders the detection difficult. Understandably, Corman et al.
(2021) in their article had suggested that guidelines for use of RTK-Ag should mention that a
negative result might reflect low sensitivity as symptoms could occur soon after testing.
Interestingly, a meta-analysis by Khandker et al. (2021) involving 17,171 subjects
reported in 29 selected studies revealed that the RTK-Ag’s showed better performances in
sensitivities for the European and American populations (70.0% and 74.1% respectively)
compared to the African and Asian populations (56.4% and 65%). This demonstrates that
there is a wide range of products in the market for RTK-Ag’s and their performances can vary
greatly. These variations in performance could be due to the products’ quality, or they could
also be attributed to the differences in sampling procedure or differences in prevalence of the
disease in the population where the study was conducted.
Furthermore, use of RTK-Ag biosensor may not be able to effectively detect
mutations in the virus, an area where gene sequencing such as using RT-PCR is more
suitable. In the article by Xi et al. (2021) where microarray technologies in biosensors were
reviewed for rapid detection and early clinical diagnosis of SARS-CoV-2 mutations, the
authors recognized that application of biosensor based on gene level is more promising than
on protein level for the detection of SARS-CoV-2 mutations.

2.5 Comparison of RT-qPCR Test and RTK-Ag Test (Malaysia)

Table 4: Comparison of RT-qPCR and RTK-Ag Test (Caring Pharmacy, 2021; Rahman, 2021)
Aspects RT-qPCR RTK-Ag Test

Purpose Identify presence of virus Identify presence of virus protein


RNA

MOH Recognition Gold standard for Recognized by MOH

14
COVID-19 Testing

Testing Procedure Nasal and throat swab Nasal swab

Turnaround Time 24-72 hours 12-30 minutes

Accuracy 99.9% 90.0%

Advantage Sensitivity > 90%, Specificity >97.0%,


Specificity >97.0% Able to detect COVID-19 outbreak
quickly in large population

Disadvantage Risk of virus gene mutation Poor sensitivity,


giving false negative result Higher risk of false negative,
Need to be confirmed with
RT-qPCR test

Cost Higher Lower

New Development N/A Available over the counter,


(as of July 2021) Cheaper and more sensitive and
specific kits has been developed

15
Chapter 3: Saliva Self-test kit

3.1 Introduction to Saliva Test Kit

The Malaysian health ministry gave conditional approval for trade in and supply of
COVID-19 self-test kits on 20th of July, 2021. The first self- test kit to be approved in
Malaysia was Salixium Covid-19 Rapid Antigen Test by Reszon Diagnostic International Sdn
Bhd Malaysia, an in vitro diagnostic rapid test kit producer company. Salixium Covid-19
Rapid Antigen Test used nasal and saliva swabs as a mix specimen to detect SARS-CoV-2. In
fact, the first saliva self-test kit to be approved in Malaysia was Gmate Covid-19 Rapid test
by Philosys Co Ltd in South Korea (Jayatilaka, 2021). Currently, there are a total of 64
self-test kits approved by the health ministry in Malaysia out of which 31 are self-test kits
which use saliva as the sole specimen (Zaini, 2021).
Quick and precise diagnostic tests are crucial for monitoring the COVID-19
pandemic. However, to deliver testing for a huge population can be a significant challenge
when it comes to collecting the biological specimen (Landry et al., 2020, Ott et al., 2020,
Yokota et al., 2020). Utilization of self-collected samples for COVID-19 serves numerous
benefits, especially minimizing the risk of exposure to the virus for healthcare workers as
trained personnel would not be involved in the process of collecting the sample.
(Valentine-Graves et al., 2020, Hall et al., 2020). Furthermore, the usage of saliva samples for
diagnosis of SARS-CoV-2 has a great advantage as opposed to nasal or throat swab collection
since saliva collection is a non-invasive procedure and so avoids patient discomfort
(Fakheran et al., 2020). A comparably sensitive and cheap alternative that can substitute
nasopharyngeal swabs for assortment of clinical samples for SARS-CoV-2 testing would be
by using saliva samples (Bastos et al., 2021). In fact, a study suggested that making use of
self-collected saliva specimen and an RNA extraction-free RT-PCR method, demonstrated a
diagnostic accuracy similar to that of oro/nasopharyngeal swab and so supported the
application of this protocol as an alternative instead (de Oliveira et al., 2021).
Interestingly, saliva tests have a great accuracy in the first days of the symptoms with
a 92% sensitivity and 97% specificity and so this makes it useful for detecting SARS-CoV-2
in children as well (Al Suwaidi et al., 2021). The viral load detected in saliva in the early
symptomatic stage of COVID-19 may be a prognostic indicator and can be used to monitor
the course of the disease in patients over 45 years of age as shown by a study (Aydin et al.,
2021). For early stage and presymptomatic Covid-19, saliva maintains high clinical

16
sensitivity minimizing the need for personal protective equipment and making it a feasible
choice for population scale testing (Johnson et al., 2021). Nevertheless, a study carried out to
compare the sensitivity of self-collected nasal and saliva samples in mild COVID-19 patients
showed that saliva had a sensitivity of 90% and nasal samples had a sensitivity of 99%
(Alemany et al., 2021). The reliability of saliva for detecting SARS-CoV-2 was studied in
symptomatic patients and asymptomatic patients. It was found that the symptomatic patients
showed a higher sensitivity of 80% versus the 36% in asymptomatic patients. In high risk
(symptomatic, feverish and/or with comorbidities), a sensitivity of 82% was recorded versus
the low risk (asymptomatic, not feverish and no comorbidities) which had a sensitivity of
22% (Alkhateeb et al., 2021).
In the journal Nature Medicine, a study was published which demonstrated that
SARS-CoV-2, the virus which leads to COVID-19 can actively affect the cells lining the
mouth and the salivary glands. This is explanatory for the manifestation of oral symptoms of
COVID-19 such as dry mouth, loss of taste and oral ulcers and the reason for how saliva tests
have helped in detection of COVID-19. Moreover, the findings indicate that the mouth plays
a significant role in the transmission of the SARS-CoV-2 virus and when the saliva loaded
with the virus is ingested or inhaled, can spread to the oesophagus, the respiratory system or
the digestive system (Huang et al., 2021). As a matter of fact, Chen et al. (2020) found that
the salivary glands expressed the likely cell receptor of SARS-CoV-2, angiotensin-
converting enzyme II (ACE2) and was able to detect the virus in saliva samples of patients.
A study showed a simple, suitable and a low-cost alternative to the conventional nasal
and pharyngeal swab-based tests is the self-collected saliva-based test (Vaz et al., 2020). The
advantages of using a novel SARS-CoV-2 detection kit over direct PCR are that the time for
detection is shorter and it excludes the steps required for RNA extraction and purification and
at the same time, does not impair the diagnostic accuracy (Fukumoto et al., 2020). The newly
built SARS-CoV-2 antigen lateral flow devices (LFDs) detect the occurrence of particular
viral proteins making use of conjugated antibodies to bind spike, envelope, membrane or
nucleocapsid proteins. These antigen tests can identify viral proteins directly unlike the
IgM/IgG “antibody tests” and these antigen tests do not depend on the host’s immune
response. In comparison to the RT-PCR, the results from LFDs are analyzed in 10 to 30
minutes depending on the device, and so opens a window for early interventions to stop the
chain of transmission earlier in the course of the disease when individuals are most infectious
(Guglielmi, 2020).

17
3.2 Basic Working Principle of Saliva Test Kit

The Saliva self-test kit for COVID-19 is an in vitro immunoassay which identifies
SARS-CoV-2 variants. The variants of SARS-CoV-2 can be either alpha, delta, gamma,
kappa and beta. The kit helps to detect COVID-19 by directly and qualitatively identifying
the SARS-CoV-2 viral nucleoprotein antigens in the human saliva samples from individuals
who suspect they may have COVID-19 within the first week of symptom onset. Basically, the
COVID-19 Antigen Saliva Test Kit is designed to be used by lay persons and allows for a
self-test at home. During the acute phase of the infection is when the antigen of SARS-CoV-2
is usually detectable in the saliva samples (COVID-19 Antigen Saliva Test Kit Self Test,
2021). There are numerous COVID-19 saliva self-test kits with different designs and
protocols to follow. However, the working principle is the same for all.
The COVID-19 Antigen Saliva Test Kit to detect SARS-CoV-2 is based on a lateral
flow immunoassay (LFIA) and is an antigen detection-based biosensor. The principle of a
Lateral Flow Assay (LFA) uses the propagation of the liquid sample containing the analyte
through a polymeric strip coated with molecules that interact with the analyte, producing a
signal that can be visually detected (Koczula & Gallotta, 2016). The lateral flow device
consists of a control line (C) which ensures that the device is working properly and one or
two test lines (T). These kinds of tests are qualitative and are read visually when the color
band develops on the lines serving as the indicator for the test ("What is a lateral flow test
and how does the technology work?", 2021).

Figure 13: Schematic illustration of antigen detection-based LFIA Test


(Hsiao et al., 2021)

In a COVID-19 Antigen Saliva Test Kit, SARS-CoV-2 nucleoprotein viral antigens


are detected through visual inspection of the development of color. The test region (T) of the
nitrocellulose membrane consists of anti-SARS-CoV-2 antibodies which are immobilized on

18
it (e.g. monoclonal antibodies against SARS-CoV-2 antigen). The conjugated pad contains
anti-SARS-CoV-2 antibodies conjugated to colored particles which are immobilized (e.g.the
colloidal-gold conjugated with monoclonal antibodies specific to SARS-CoV-2 antigen and
anti-mouse polyclonal antibodies). When the specimens are processed and placed on the
sample pad of the test device, the SARS-CoV-2 antigens will bind to anti-SARS-CoV-2
antibodies conjugated to the colored particles. By capillary action, as the specimen moves
along the nitrocellulose membrane, it interacts with the reagents on the membrane and the
complex formed is captured by the anti-SARS-CoV-2 antibodies at the test region (T). The
colored particles in excess are captured at the control zone (e.g. the control zone has
anti-mouse IgG antibodies to bind to remaining conjugates) . If there is a colored band in the
test region, the test will be positive for SARS-CoV-2 and its absence indicates a negative test
result. The purpose of the control zone is to ensure that a proper volume of the specimen is
added and affirms the validity of the test (Orawell COVID-19 Ag Rapid Saliva Test Device,
2021).
The block diagram in Figure 14 (shown below) presents a general approach for testing
of SARS-CoV-2 using saliva self-test kits. In fact, there are various designs of COVID-19
saliva self-test kits which differ in terms of the protocol followed, the methodology of
collection of the sample, the source of anti-SARS-CoV-2 antibody used (Anti-SARS-CoV-2
nucleocapsid protein mouse antibody, anti-SARS-CoV-2 nucleocapsid protein rabbit
antibody) and the overall body of the test kit. However, the inner workings are the same
which basically make use of an antigen detection-based method and qualitative based
membrane immunoassay. The target analyte is the N-protein (Nucleocapsid) of SARS-CoV-2
which is the core part of SARS-CoV-2 located inside the virus.

Figure 14: Block Diagram of a Saliva self-test kit

19
According to Table 5 below, the saliva self-test kit with the highest sensitivity rate at
96.8% is Gruenbanka SARS-CoV-2 Antigen Detection Kit (Colloidal Gold Method) and
belongs to a manufacturer in China which runs by the name,NingboNingbo Lvtang
Biotechnology Co. Ltd while the specificity rate for most of the kits is recorded to be 100%.
The time to result is 15 minutes for all kits except for one, Longsee 2019-nCoV Ag &
Influenza A/B Rapid Co-Detection Kit which gives results at 5 minutes. The lowest
sensitivity tabulated is 90.1%.

Table 5: A comparison of the sensitivity, specificity rates and result time of a few of the
Saliva Self-test Kits approved for sale in pharmacies in Malaysia ("Covid-19 self-test kits in
Malaysia: A guide to the types, efficacy rates, and where to buy them", 2021)
No. Name Sensitivity Specificity Time to Manufacturer
rate (%) rate(%) Result(min)

1. GMate Covid-19 Ag 90.9 100 15 Citymedic Sdn


Bhd, Malaysia

2. Beright Covid-19 Antigen 93.1 100 15


Rapid Test Device

3. All Test Covid-19 Antigen 91.38 100 15


Rapid Test

4. JusChek COVID-19 90.1 99.3 15


Antigen Rapid Test

5. 94.3 100 5 Guangdong Longsee


Biomedical Co.
Ltd.,Guangzhou,
China

6. 96.5 100 15 SD Biosensor, Inc,


Korea

7. 93 99 15

20
8. 93 99 15

9. 96.8 99.6 15

*The sensitivity of a test determines how frequently it correctly generates a positive result, whereas
the specificity of a test determines how often it correctly generates a negative response.

The advantages and disadvantages of saliva test kits are depicted in Table 6 below.

Table 6: Advantages and disadvantages of saliva test kit


Testing Advantages Disadvantages
Device

Saliva - User friendly, - Limited detection range and sensitivity


Test Kit requires
minimal - Can only be used to detect COVID-19 within the
Training first seven days of symptom onset and so can give
false negative if not done within the seven
- Non-invasive,
minimize - Inaccurate results if conduct incorrect procedure
discomfort
- Possible false negative if the antigen concentration
- Portable is below the detection limit or if the specimen is not
collected properly.
- Low cost
- Positive results do not rule out co-infections with
- Rapid results other pathogens.
- A confirmatory test is required to define a
COVID-19 case.
- The test is not reliable in asymptomatic individuals
and in the later phase of infection.

21
SECTION 2: Breath-based COVID-19 Tests

Chapter 4: Comparison of Technologies

4.1 GeNose C19 - Indonesia

4.1.1 Introduction to GeNose C19 - Indonesia

In Indonesia, GeNose, an electronic nose, was created as an option to screen patients infected
with SARS-CoV-2 (Saki, Setiawan, & Wicaksono, 2021). The device has received
distribution permits from the government and recent diagnostic trials on 1999 patients
showed 89-92% sensitivity for detecting positive COVID-19 cases. The GeNose C19 is built
on an AI-powered sensory system that was previously used to analyse food, beverages, and
identification of tuberculosis. For COVID-19 detection, GeNose C19 acts as an electronic
nose that mimics the human nose's process. The machine-learning-based AI system has been
trained to recognise an object's specific reaction pattern and identifies the disease through
exhaled volatile organic molecules (Nurputra et al., 2021).

Figure 15: GeNose device (Sipayung, 2021)


Generally, GeNose is a device that is able to rapidly detect the presence of COVID-19
in a short amount of time with high accuracy, affordable cost and it is also non-invasive.
Compared to swab tests, GeNose does not require any chemical solution or swab sample to
perform tests (Simanjuntak, Octavia, & Sinaga, 2021). The main objective for developing
GeNose is to achieve rapid and accuracy screening tests with incorporation of AI to
automatically classify positive and negative COVID-19 cases.

22
4.1.2 Working Principle of GeNose

With reference from the GeNose manual guide, the entire system consists of 7 main
components: GeNose, Hepa Filter Unit, AI Unit (Computer), Adapter 12VDC 5A
(maximum), USB 2.0 Cable, Serial Cable, and Sampling Kit . Figure 16 and Figure 17 show
the connection circuitry and block diagram for the GeNose system respectively.

Figure 16: Circuitry connection of GeNose System

Figure 17: Block Diagram of GeNose System


The corona virus is identified by GeNose by analysing Volatile Organic Compounds
(VOC) patterns from the patient's breath. An array of chemiresistive sensors is used in the
GeNose design to detect the VOCs in the breath sample. These chemiresistive sensors work
based on equilibrium shift of the surface chemisorbed oxygen reaction in response to VOCs
thus producing flow of free electrons that can be measured. The integrated responses of these
sensors toward the complex VOCs produce a qualitative pattern that is analysed using
artificial intelligence (AI) algorithms. The Deep Neural Network (DNN) model has been
found to produce the most stable and optimum machine learning algorithm, yielding 95.5%
sensitivity and 95.7% specificity (Nurputra et al., 2021, 14).
Accordingly, subjects to be tested with GeNose are instructed to exhale into a
sampling bag during the examination. After that, the breath sample is fed into the GeNose

23
unit where the sensors will detect VOCs. The VOC data in the form of a profile is then
analysed with a computer by using artificial intelligence to give the predicted results of
whether the subject is COVID-19 positive or otherwise. Specific steps of using GeNose are
as follows:
1. Start the device 30 minutes prior to the test.
2. With mask on, deeply inhale by using nose and exhale by using mouth three times.
3. Deeply inhale by using the nose and exhale into a sample kit until the bag is fully
filled. Lock T-valve of airbag.
4. Install sample kit into HEPA filter that is connected to the sensor. Unlock the T-valve
of the airbag and press “Analyze” on GeNose AI Dashboard.
The performance of chemiresistive sensors is highly dependent on temperature,
humidity and background gases. Therefore, the GeNose design has incorporated a constant
temperature and humidity control within the test chamber. The GeNose unit needs to be
pre-conditioned before use for about 30 minutes to produce a baseline of the background
gases, and if the background air is too polluted (resulting with an output signal higher than
3000 mV) then the unit needs to be moved to another cleaner area (GeNose C19-S, 2021).

4.2 BreFence Go COVID-19 Breath Test System - Breahonix, Singapore

4.2.1 Introduction to BreFence Go COVID-19 Breath Test System - Breahonix, Singapore

Breathonix, a company from Singapore, had developed the BreFence Go COVID-19 Breath
Test System and received provisional approval by Health Sciences Authority (HSA) of
Singapore in May 2021. According to results from a pilot study conducted in Singapore with
180 individuals, a machine learning algorithm shows that the system is able to achieve 93
percent sensitivity and 95 percent specificity. Breathonix and the Singapore Ministry of
Health (MOH) are now planning a deployment experiment at the Tuas Checkpoint, where
visitors entering the nation would be checked using the breath test equipment (Verdict, 2021).
Accredify and Breathonix, both located in Singapore, have announced a partnership to
provide an on-site COVID-19 rapid-breath test technology that gives digitally verified
COVID-19 test results in less than one minute, complying with nations' international travel
and local health standards (BioSpectrum, 2021). In May 2021, MY EG Services Bhd (MyEG)
partnered with Breathonix Pte Ltd of Singapore to bring the breath test equipment to
Malaysia for the screening of the Covid-19 virus (Adilla, 2021).

24
4.2.2 Working Principle of BreFence Go System

Figure 18: Schematic illustration of BreFence Go


System (Breathonix, 2021)
BreFence Go System detects volatile organic compounds (VOCs) in the subject's
exhaled breath. The user must blow into a disposable one-way valve mouthpiece linked to a
breath sampler to perform the BreFence test. The exhaled breath is put into a mass
spectrometer (MS) to quantify VOCs. Following that, a machine learning software
programme evaluates the VOC biomarkers and provides results in under a minute. Any
qualified individual may do the basic test as operation of the system does not require
medically skilled professionals or laboratory processing. A positive breath test result will
necessitate confirmation with a polymerase chain reaction (PCR) Covid-19 swab test.

4.3 Trace Virus Detection (TVD)-2 Breathalyser - GreyScan, Australia

4.3.1 Introduction to Trace Virus Detection (TVD)-2 Breathalyser - GreyScan, Australia

When it comes to testing for the COVID-19 causing virus SARS-CoV-2, GreyScan's existing
Capillary Zone Electrophoresis (CZE) technology is combined with existing methods of lung
disease analysis to give the TVD-2 Breathalyser the advantage of speed, comfort, and
accuracy.

25
Figure 19: TVD-2 Breathalyser (GreyScan, 2021)
Capillary zone electrophoresis (CZE) uses a high voltage to separate molecules in a
liquid. It's also a well-established approach for biomolecule analysis, having first been used
to sequence the human genome, and it's demonstrated to have a great deal of promise for
detecting complex biological assemblies. The existence of different surface charges and zeta
potential allows native virus and subviral particles to be distinguished.
Previously in 2020, TVD-1 (trace virus detection) was developed to be the first
fieldable, quick, and reliable screening tool capable of detecting SARS-CoV-2 virus presence
on surfaces in community settings.The TVD-1's operation will be intended for simplicity of
usage in the field by first responders as well as cleaning and disinfection service providers
across the world (Tocco, 2020).
In 2021, TVD-2 Breathalyser was developed. At present, GreyScan is still pending
approval from Therapeutic Goods Administration (TGA), the Food and Drug Administration
and funding by the Australian government in releasing their product (Mcphee, 2021).
Current diagnostics, such as polymerase chain reaction (PCR), immunoassay, and
antibodies, are unable to distinguish active virus particles from non-contagious dead particles
and are prone to false positives. The company claims that the portable and simple to operate
TVD-2 Breathalyser is the sole on-the-ground test that can offer rapid, comfortable, and
accurate findings to assist protect the safety of the general public, whether it's used in crowds,
at border crossings, or in frequently travelled regions (GreyScan, 2021).

26
4.3.2 Working Principle of TVD

Figure 20: Working Principle of TVD-1 (Breadmore, 2020)


Previously, the intention of creating TVD-1 was to allow screening of potential biological
threats in aviation. CZE was combined with TVD-1 to successfully separate proteins,
peptides, and nucleic acids. The general concept of TVD is shown in Figure 20. Firstly, the
user will need to breathe into a single-use disposable mouthpiece which contains a filter to
capture virus. Moving on, the filter will be transferred into the TVD-1 device for analysis.
Biological interpretation can be shown within 5 minutes and display of red color shows
potential virus detection while green color shows none virus was detected.
TVD-1 is portable and able to be set up within 5 minutes. It can be used at
checkpoints, as a pre-checkpoint screening tool, or as an on-demand check in any area that
needs testing after cleaning, or in an outbreak zone, such as COVID-19 red zones. It will look
and operate similarly to the current fleet of explosive detection systems (ETDs) found at
airport checkpoints, with red and green screen alarms and operational commands similar to
ETDs.

4.4 CoronaCheck - Exhalation Technology (ETL), United Kingdom (UK)

4.4.1 Introduction to CoronaCheck - Exhalation Technology (ETL), United Kingdom (UK)

Exhalation Technology (ETL), located in Cambridge, UK, has announced the launch of a
major human clinical research with their unique CoronaCheck diagnostic test for Covid-19 in
exhaled breath condensate (EBC). The device is designed to satisfy the requirement for
speedy and reliable testing in a variety of settings such as hospitals, airports, retail stores,
schools, and similar venues. CoronaCheck has been evaluated with both live and inactivated

27
viruses, as well as in both controlled and "virus-spiked" exhaled breath condensate (EBC)
(Mason, 2020).
According to Exhalation Technology (2021), CoronaCheck is an evolution of an
existing technology, the Inflammacheck®, which is the only device on the market designed to
collect Exhaled Breath Condensate (EBC) and test for compounds directly in it at the point of
care.

Figure 21: CoronaCheck


(Exhale Technology, 2021)
ETL reported that with 62 patients tested thus far, CoronaCheck has recognised those
individuals who are either positive or negative for COVID-19 with 100 percent specificity
and 100 percent sensitivity. Some of the individuals who tested positive for COVID-19 were
even asymptomatic (Select Science, 2021). Exhaled breath moisture is a highly clean
biological sample with little interfering particles, unlike blood, urine, saliva, swabs, and etc,
hence CoronaCheck technology is very rapid and efficient.
In addition to that, CoronaCheck does not identify any of the typical respiratory
viruses or bacteria that cause colds and seasonal flu but SARS-CoV-2 viruses and is thought
to identify the majority mutations discovered so far. CoronaCheck can identify Covid 19 and
related variations in both symptomatic and asymptomatic people. It can be operated without
special training, and is appropriate for use in schools, airports, workplaces, public venues etc
as well as at hospitals and doctors offices.

4.4.2 Working Principle of CoronaCheck

CoronaCheck catches small amounts of exhaled breath which includes numerous


SARS-CoV-2 virus particles in an infected person within minutes. CoronaCheck employs a
new biosensor contained in a safe breath analysis device designed for quick point-of-care

28
coronavirus detection. In contrast to blood antibodies, the test directly detects the presence of
virus in the breath.
The basic working principle of CoronaCheck is by collecting the exhaled breath
condensate sample from the patient (see Figures 22 & 23) and analyse the sample using an
immuno electrochemical sensor. Thus, unlike GeNose and BreFence that analyse the VOCs
in the breath, the CoronaCheck directly detects the presence of the SARS-CoV-2 virus .

Figure 22: Exhaled Breath offers a powerful medium for collection of a range of
biomarkers and pathogens found in respiratory tract (Exhalation Technology, &
The Sage Group,, 2021)
The immuno electrochemical sensor in CoronaCheck is functionalized with a
macromolecule directed to bind with the surface of the virus, if present. The binding event
takes about 2 - 5 minutes and will give out an electrochemical signal that is read by the reader
within a few seconds. The reader is enhanced with a machine learning algorithm to maximise
its sensitivity and specificity (Exhalation Technology, & The Sage Group, 2021).
Based on the explanation on Inflammacheck by Maniscalco et al. (2021), the
temperature for the breath collection unit is below 6 degree Celsius to allow condensation to
happen when breathing into the disposable mouthpiece of the breath collection bag. Cooling
of the collection unit is initiated by means of the Peltier element. After the breath has been
condensed into water droplets, the water droplets which contain the viruses flow to the
sensor. The sample is then analyzed and test data is generated.

29
Figure 23: Working Principle of Corona Check (Maniscalco et al., 2021)
CoronaCheck does not necessitate the use of laboratory equipment, nor does it
necessitate any specific handling or training. The biosensor is housed in a test cartridge that is
introduced into the device into which the subject breathes, yielding findings in minutes before
the test cartridge is securely destroyed and the device is ready for a fresh test cartridge and
subject (Mason, 2020).

30
4.5 Comparison of breath-based COVID-19 point-of-care tests
Table 7: Comparison of four breath-based COVID-19 point-of-care tests
Aspects GeNose C19 BreFence Go System TVD-2 (*) CoronaCheck

General Qualitative VOC pattern (detected Quantitative VOC pattern Immuno electrochemical
Principle by chemoresistive sensors) with (measured by MS) with machine sensor that detects the virus,
machine learning learning and enhanced with machine
learning

Cost $0.7-$1.7 $5-$20 N/A < $10

Time 3 minutes < 1 minute 3 minutes < 5 minutes

Specificity 95.7% 97% N/A 98%

Sensitivity 89-92% 85.3% N/A 98%

Advantages Cheap, Cheap, Rapid Result, User Friendly; Reproducible, Direct detection of virus; Can
Rapid Result, Machine learning on the MS results Rapid Result, be adapted to other respiratory
User Friendly. may possibly identify the specific Able to differentiate carriers infectious diseases by using
biomarkers for COVID-19 when of live virus and fragments dual biosensors within the
more tests data are accumulated. of dead virus. same unit;

Dis- Requires pre-conditioning to Investment cost for mass Amplitude of electroosmotic Electrochemical immunoassay
advantages produce a baseline of background spectrometry is high; Indirect flow (EOF) might vary, for analytical applications has
gases; Different VOC patterns interpretation of results based on which can impair migration typically small voltage
resulting from population VOC. time and peak area windows, the need for proper
difference have not been validated; repeatability; Highly sample extraction, and the risk
Indirect interpretation of result sensitive to pH value of of poisoning the detector
based on VOC. breath. surface.

Regulatory Currently obtaining Indonesian Health Sciences Authority (HSA) N/A Currently obtaining FDA and
Approval National Standard (SNI) SG has granted provisional CE mark
authorisation
Note (*): please see Appendix A where we have attempted our inquiry to Greyscan on TVD-2 but have not received their response.

31
With reference from Chew (2021), each test for BreFence Go System costs between $5
and $20, depending on the volume of the deployment, which is significantly less than the cost
of PCR testing in Singapore. On the other hand (refer Table 8), the cost of a test for the
GeNose breathanalyser is about $0.7-$1.7 (Nurputra et al., 2021). Based on (Mason, 2020),
the CoronaCheck breathanalyser typically costs about less than $10 per test. Unfortunately,
the price for the TVD-2 is not available online and the email enquiry was not replied.

Table 8: Comparison of PCR, RTK, and GeNose in Indonesia (Nurputra et al., 2021)

Among the four breath-based tests discussed in Table 7, the GeNose and BreFence
analyse the VOCs present in the breath sample, while the TVD-2 and CoronaCheck perform
direct detection of the virus protein. BreFence is bulky due to the inclusion of a mass
spectrometry. While it is understood that the product can be easily applied within Singapore
due to the nation’s small and dense population, its application may be hindered in rural areas
in Malaysia where the population is more dispersed. As for GeNose, it relies heavily on the
machine learning algorithm for it to function, without which the VOC pattern has little
meaning. The problem with this approach is that a very large and diverse sample data is
required to validate its performance. For example, if the unit is tested in Malaysia it may
result with different sensitivity and specificity due to the different VOC patterns of the people
impacted by the type of food, race and metabolism. TVD-2 and CoronaCheck in many
respects work on a similar principle to that found in existing antigen rapid test kits, but with
the added advantage of higher sensitivity and specificity due to enhancement by machine

32
learning algorithms. Furthermore, the TVD-2 and CoronaCheck are small in size and portable
making them useful in many point-of-care situations. However, due to the use of immuno
sensors, these become consumable parts that need to be replaced after every test hence their
operational costs can be higher compared to GeNose and BreFence.

33
Chapter 5: Improvement for Breath-based COVID-19 tests

5.1 Introduction

The COVID-19 epidemic has highlighted the need for rapid, non-invasive methods for
screening persons who may have been exposed to pathogens. In contrast to bacterial
infections, where toxins produced by the infectious agent cause much of the pathology,
COVID-19 symptoms are caused by the host's reaction to the virus.
Controlling the pandemic will need rapid COVID-19 screening. Current nucleic acid
amplification, on the other hand, necessitates time-consuming processes in addition to the
inconvenience of obtaining throat/nasal swabs. The combination of breath-borne volatile
organic compound (VOC) biomarkers and machine learning has the potential to be employed
for COVID-19 point-of-care screening. SARS-CoV-2 infection may produce distinct
breath-borne VOC profiles that can be utilised to test COVID-19 quickly.
With reference to Chen et al., (2021), by using a commercial gas chromatograph-ion
mobility spectrometer, researchers discovered greater amounts of propanol in the exhaled
breath of COVID-19 patients with non-COVID-19 respiratory illnesses than in
non-COVID-19 patients. In contrast, COVID-19 patients had considerably lower levels of
breath-borne acetone than other participants.
Machine learning models (support vector machines, gradient boosting machines
(GBMs), and Random Forests) were shown to have great accuracy in discriminating
COVID-19 from respiratory disease inside the spectrometer, ranging from 91 percent to 100
percent. The GBM and Random Forests algorithms can also accurately differentiate
respiratory infected individuals from healthy persons (Chen et al., 2021).
According to the study of Snitz et al. (2021), a deep learning classifier was used to
analyse eNose data and result shows that eNose able to diagnose SARS CoV-2 infection in
real time at a level that was much better than chance for both symptomatic and
non-symptomatic subjects. This proof of concept with a generic eNose suggests that an
optimised eNose might allow for successful real-time diagnosis in the Covid-19 epidemic,
providing significant relief.
Generally, multiple COVID-19 breath analysers are now being developed by various
organisations in various countries. This unique noninvasive sensor has the potential to
identify novel coronaviruses early, allowing (i) quick large-scale population screening in a
short period of time, (ii) active-case hunting in the community, and (iii) a cheap point-of-care

34
diagnostic tool. A small investigation found that the breathalyzer tests had high specificity
and sensitivity for discriminating and screening COVID-19 and lung infection as well as
healthy controls at the same time ( Ministry of Health, Malaysia, 2020). However, more
investigation, validation, and verification with a larger sample size are necessary to determine
its effectiveness and safety.
Among four of the popular breathalyzers are GeNose C19, BreFence Go System,
TVD-2, and CoronaCheck. Generally, they all have one common feature, which is rapid
screening. For instance, the screening time for CoronaCheck is the longest (<5 mins),
followed by GeNose C19 and TVD-2 (3 mins) and lastly BreFence Go System (<1 min).
Compared to the conventional RTK-Ag or Saliva Test Kit which is approximately 15 minutes
to 30 minutes, breathalyzer is able to provide rapid results. However, there are also some
limitations to the current design of breathanalyzer. For instance, the composition of volatile
organic compounds (VOCs) in a gas sample will differ based on temperature and humidity.
Hence, this might contribute to false positive and false negative results.
For the case of BreFence Go System, the machine cost of the device is relatively high
due to the inclusion of mass spectrometry in detecting the COVID-19 virus. In reality, the
device for breathanalyser should be as minimal as possible, this is to make sure that it can be
mass produced to be distributed in different areas and countries, especially developing or
least developed countries. For the case of TVD-2, which involves the use of CZE, the
electroosmotic flow (EOF) is at the heart of capillary electrophoresis (CE). One problem of
executing CZE in the presence of EOF is that the amplitude of the EOF might vary, which
can impair migration time and peak area repeatability. Furthermore, modest pH changes
affect a molecule's charge and flow in CE, hence small changes in pH have a higher impact
on CE. Therefore, if the patient does not have a normal pH value on the breath exhaled, the
result might be inaccurate (Chromatography Today, 2014).
Electrochemical immunosensor is utilized in CoronaCheck. Electrochemical
immunoassays provide several benefits, including speed, accuracy, and precision.
Furthermore, electrochemical detection setups have intrinsic properties such as cheap cost,
easy scalability, and simple operation, which are frequently referred to as vital in the design
of biosensors. However, electrochemical immunoassay for analytical applications has several
limitations due to the typically small voltage window, the need for proper sample extraction,
and the risk of poisoning the detector surface (Wu & Ju, 2012). Sole usage of electrochemical
immuno biosensor might result in inaccurate results as the detector itself might be poisoned
or contaminated anytime throughout the period of usage.

35
Upon considering and levelling the advantages and disadvantages of the mentioned
devices, this case study will be focusing on improving the existing breathalyser, specifically
combining and innovating ideas from both GeNose C19 and CoronaCheck used in the
industry for detection of COVID-19 viral infection, to adapt with the situation and
background of Malaysia.

5.2 Objective

The main objective of this case study is to develop a novel paradigm which improves the
current innovative for breath-based Covid test Point-of-care. The study will include
improvement in the aspect of the working technique, design, cost, and advantages in detail.
Specific objective of the case study is as following:
1. To improvise and introduce a novel breathanalyser system with sufficient
effectiveness, safety and cost-effectiveness for the detection of COVID-19 in
Malaysia.
2. To provide detailed design of breathanalyser which incorporates the block diagram
and circuitry design of the device.
3. To employ the transfer learning technique of CNN to build a VOC pattern classifier to
allow automatic classification of positive and negative cases of COVID-19.

5.3 Proposed Design

5.3.1 General Working Principle

Our design proposal for breath analyser possesses a similar working principle to that
elaborated under Section 4.1 for GeNose C19, but with two distinct improvement features
related to sample pre-processing and data integration. Figure 24 and Figure 25 illustrate the
block diagram of the proposed design and circuitry connection for this case study
respectively. Red colour in these figures represent our design improvement features.

36
Figure 24: Block diagram of proposed design. Red boxes indicate design improvement

Figure 25: Circuitry connection of proposed design. Red colour indicates design
improvement
By referring to Figures 24 and Figure 25, first of all, the exhaled breath samples are
collected from a subject using a sampling kit. The breath samples are securely contained in
the sampling kit and connected to a HEPA filter for physical separation of large particles
from the gas sample.

37
This is followed by a process that we propose as an improvement to the design
whereby the gas is further cooled in a precooler unit which aims to remove moisture and
soluble macromolecules. The precooler unit acts as a dehumidifier which separates liquid
from gas (breath). Exiting from the top outlet of the precooler is the dry and cooled gas from
the breath sample. At the same time, exiting from the bottom of the precooler is the
condensed liquid from the breath sample, that can optionally be transferred to an
electrochemical immunosensor (working principle refer section 7.2 by CoronaCheck).
The resultant dry and cooled gas exiting from the top of the precooler unit is then fed
into the GeNose unit whereby it is conditioned again to 40 ± 1 degree Celsius according to
the sensors’ operating requirements. Within the GeNose unit is an array of chemiresistive
sensors made from metal oxide semiconductors. These sensors will change in their
conductivities in the presence of target gases due to redox reactions between the metal and
gas molecules. The change in movement of free electrons on the metal oxide semiconductors
produces a pattern of electrical currents that is shown as a VOC profile. Later, by using
artificial intelligence, the Deep Learning Neural (DNN) model is employed to recognize
patterns of results.
The second part of our design improvement involves using artificial intelligence not
only from VOC pattern but by using results from both the chemiresistive sensors and
electrochemical immunosensor to differentiate a COVID-19 patient from a non- COVID-19
subject. The result automatically generated from machine learning will be cross-validated
between the two sensors and stored locally in the laptop. Later, as per our design proposal,
the final result of the test is sent to the cloud server and automatically reported in the
MySejahtera application.

5.3.2 Sample Pre-processing - Additional Installation of Precooler Unit

It is well understood that the composition of volatile organic compounds (VOCs) in a gas
sample will differ based on temperature and humidity. Furthermore, the sensitivity of
chemiresistive sensors is negatively affected by humidity (Wang et al., 2010). Therefore, it is
vital to remove air moisture and heavier components from the breath sample before feeding it
to the chemiresistive sensors for analysis.
This can be achieved by using an indirect pre-cooling unit, also known as sample gas
cooler, powered by electricity. An example sample gas cooler is shown in Figure 26. The
breath sample will be pre-cooled to 6 degree Celsius to remove from its vapour phase, most

38
of the moisture content and any heavier components (such as proteins, bacterial and viral
particles, macromolecules and drugs) that are present in the sample which interfere with the
VOC pattern.

Figure 26: Example sample gas cooler for


pre-cooling gas sample in our design (Ahlers,
2017)
A sample gas cooler works on the principle of psychrometric properties whereby
condensable components in the sample, including moisture, can be separated out by cooling
down the temperature. To be more precise, at 35 degree Celsius every one kilogram of
saturated air can hold 36.6 grams of water in its vapour phase; but as it is cooled down to 6
degree Celsius, one kilogram of saturated air can only hold 5.8 grams of water in its vapour
phase. The difference of 30.8 grams of water is thus condensed out in liquid form and can be
separated from the air via this working principle. An online calculator to perform the
psychrometric calculation is available at Calculator: Saturated Humid Air Table.
After the condensable components turn into liquid for draining out from the cooler
unit, a dry and non-interfering gas-phase of the sample can be obtained. This “pre-processed”
breath sample will be more suitable to come into direct contact with the gas sensors, and in
particular metal oxide semiconductor type of gas sensors as those existing within the GeNose
unit.
The bottom or condensate phase from the same pre-cooling process in liquid form can
either be disposed of (to reduce cost), or optionally fed into a separate detection system such
as immuno electrochemical sensor. This is because the liquid that is drained out from the
pre-cooling unit will carry the viral particles from the breath sample. Immunosensors are
biosensors that work by detecting interactions between an antibody and an antigen on the
surface of a transducer. The species immobilised on the transducer to detect antigen or
antibody might be either antibody or antigen. Electrochemical immunosensors investigate
electrical signal measurements produced by an electrochemical transductor. By referring to

39
Mojsoska et al. (2021), spike surface protein of the virus was used to detect the SARS-CoV-2
virus. The sensor consists of a graphene working electrode that has been functionalized with
anti-spike antibodies, which allows detection. Immunosensors that use electrochemical
detection have been studied in a variety of studies because they are specific, simple, portable,
and typically disposable, and they can perform in situ or automated testing (Felix & Angnes,
2018).
The expected outcome from this design improvement will be reducing interferences in
the VOC patterns that are caused by fluctuations in humidity of the tested gas sample and
presence of heavy components in the gas sample. The proposed inclusion of the
electrochemical immunosensor in this design aims to allow cross-validation between the
output from both sensors to ensure accurate results. By improving the quality of the gas
sample, higher sensitivity and specificity results can be expected when the VOC profiles are
analysed using the Deep Neural Network (DNN) machine learning algorithm.

5.3.3 Data Integration and Machine Learning

The software for GeNose is developed to produce results or outputs in digital form. These
results can be easily exported to a centralised health monitoring system in order to maximise
its positive effects for controlling the COVID-19 pandemic. One such system that has been
successfully rolled out in Malaysia is the MySejahtera application. Our design proposal
includes integration of the results obtained from GeNose testings to the MySejahtera
applications, thereby showing the health risk status on a person’s MySejahtera application if
that person is tested positive from GeNose.
Machine learning can be understood as a large category of algorithms and modelling
tools used for a wide range of data processing tasks, which recently has gained traction in
many scientific areas (Carleo et al., 2019). Machine learns in three ways, either supervised,
unsupervised or semi supervised. Feature learning is achieved by either building a full
convolutional neural network (CNN) from scratch or by transfer learning, which involves
modifying a pretrained CNN in the classification or prediction for a specific aspect. With
finely tuned hyperparameters, transfer learning is usually much more accurate, faster and
easier than learning or training the network from the scratch (Lopes, 2018).

40
Figure 27: Integration of Artificial Intelligence in classifying result of VOC
The sensors respond to the various fractions of VOCs in a specific way during their
exposure to the breath. Each odour, which represents a specific VOC combination, may then
produce a sensor signal pattern that is unique to that odour. Complex VOC combinations
may thus be differentiated and categorised at high throughput using pattern recognition
methods such as machine learning, without identifying the individual molecule components.

Figure 28: Google Colaboratory platform by Google


In this case study, we proposed using the Resnet50 D-CNN model as a baseline in
deep learning at Google Colaboratory Platform, which is a free platform that allows users to
conduct machine learning with local runtime GPU. Transfer learning is used instead of
training CNN from scratch. As such, the D-CNN model is pre-trained with torchvision from
fastai library and will be transferred to conduct the classification.
Results from the study of He et al. (2016) demonstrate that ResNet, even with far
more layers than typical CNN, is able to maintain stability. According to the study of Purva
(2020), another important aspect to notice is that the ResNet creators believe the more layers
that are being stacked, the better the model will perform, which is basically identical to
VGG16. However, instead of just stacking the layers on top of each other, ResNet modifies

41
the underlying mapping. ResNet50 is also one of the most popular models available, with a
top-5 error rate of about 5%.
The VOC pattern generated by the chemiresistive biosensor is captured and sent to a
machine learning algorithm to conduct auto segmentation of positive and negative cases.
Optimization algorithm using Adam is included to enhance the effectiveness of the model. In
addition to that, ReLu is activated to prevent the computation required to run the neural
network from growing exponentially. Batch Normalization is activated to enable each layer
of the network to conduct learning more independently by re-centering and re-scaling the
layers' inputs to improve the speed and stability of the network. Finally, a confusion matrix is
included to allow generalization of output of machine learning to demonstrate the accuracy,
specificity, and precision.
There are many types of biosensors that can be used to detect the composition of
VOC in the breathanalyser. Chemiresistive metal oxide semiconductor gas sensors have
shown to be quite effective, notably in the detection of volatile organic compounds (VOCs),
due to their good properties (e.g., cheap cost, quick reaction time, easy measurement setup,
high resilience, and extended lifetime). Their distinct signals can subsequently be detected
and integrated via artificial intelligence-based data post processing. As a result,
chemiresistive metal oxide semiconductor sensors have been used in breathalysers for a range
of environmental monitoring applications.
Compared to the ideology of using mass spectrometry method to analyze the
composition of VOC from the subject, GeNose C19 places a greater emphasis on AI-based
pattern analysis of integrated sensor responses to complex VOCs emerging from COVID19's
extrapulmonary metabolic and gastrointestinal symptoms. As a result, the analysis conducted
for the composition of VOC as well as the final result of the outcome of analysis may be done
in a more straightforward and time-efficient manner, while maintaining high detection
accuracy.
Usually in machine learning, a confusion matrix will be utilized in order to identify
the outcome of the learning. One of the major drawbacks of using AI is the need to train the
machine with a huge pool of information to pick up the features of VOC for a COVID-19
patient accurately. As mentioned earlier, in order to ensure that the machine detects VOC
accurately, the breath VOC samples from citizens of Malaysia must be taken because the
VOC of the subjects in Malaysia may differ according to the type of food, race and
metabolism rate. The machine can only be put into practice upon sufficient training of data.

42
In the proposed work, upon obtaining the result generated by the machine (positive or
negative COVID-19), the result will be automatically sent to the cloud server to be stored,
which will then be utilized to update the MySejahtera COVID-19 test result automatically for
the patient within minutes. See Figure 29.

Figure 29: Example of COVID-19 test result


updates at MySejahtera application

5.4 Costing

A typical GeNose test costs about USD 0.7 usd to USD 1.7 (Nurputra et al., 2021). The cost
for creating the app for surveillance after results from GeNose are fed into the app is
negligible since an app already existent is the MySejahtera app making the design
resourceful. A sample gas cooler unit typically costs about USD 2,000 (Ahlers, 2017). The
selection of the type of sample gas cooler unit to be installed may be influenced by a variety
of factors such as the resources available, the investors at hand and most importantly, the
compatibility of the sample gas cooler with the overall design to retrieve optimum benefits
and results. Nevertheless, the financial benefits of the design must exceed the investment into
the design for it to be a design that is worthwhile. This basically means if the financial return
of the new product design is much greater than the cost of the whole design project, then the
new project design is considered. Taking into account the indicated price of a sample gas
cooler published in Ahlers (2017) as USD 2,000, and by assuming 50 tests/day multiplied by
300 operating days per year, the additional fixed cost incurred by the addition of our
proposed sample gas cooler is only USD 0.13 per test to break-even in one year. Since the
GeNose test costs between USD 0.7 and USD 1.7 per test (Nurputra et al., 2021), then our
design will only increase it to about USD 0.83 to USD 1.83 per test. This is still
comparatively cheaper than the BreFence (USD 5-20 /test) and CoronaCheck tests (USD 10

43
/test). In return, we expect better reliability from the analysis result by the GeNose unit after
our proposed installation. In our above estimation, we have assumed negligible installation
cost and operation cost from the sample gas cooler, which is justifiable considering the tens
of thousands of the population that would undergo the screening test. However, since it is a
new design, it may take some time to adapt to the new test and therefore, the payback period
may be relatively longer compared to the typical at-point of care tests.

5.5 Advantages and Limitations of the New Design

5.5.1 Advantages

The proposed new design has numerous advantages. The topmost advantage is the removal of
the moisture content from the sample which consequently works out better for the
chemiresistors as chemiresistors are highly sensitive to water content which in turn causes a
deterioration in the sensitivity of the chemiresistors. The reduction in sensitivity due to the
water content is because the water molecules and surface oxygen react with each other
causing the baseline resistance of the chemiresistor to decrease. Moreover, the sensitivity also
reduces due to water content because the tin oxide (SnO2) surface area which is responsible
for the sensor response lessens as water molecules are adsorbed on the surface and therefore,
there will be less chemisorption of oxygen species. Not only does the moisture content reduce
sensitivity of the chemiresistor, but it also increases the response time because the water
molecules act as a barrier for acetylene (C2H2) adsorption (Wang et al., 2010). The reduction
in sensitivity of the chemiresistor affects the overall result and therefore, by adding the new
component to the design, the pre-cooler unit, helps in removal of the moisture content and
thereby improving the quality of the gas sample which is greatly enhanced and so helps to
attain highly sensitive and highly specific results with a faster response time. In addition to
this, the pre-cooler unit also assists in removal of larger molecules which otherwise would
cause interferences in VOC patterns and affect the accuracy of the results. All in all, the
design is quite simple since installing the pre-cooler unit is easy and so is the ease of
maintenance which can be beneficial considering the overall product life cycle of the design
product.

44
5.5.2 Limitations

Every design is bound to have flaws and drawbacks. In our proposed design, the limitation
comes to the detection of the virus, which is not direct and so, to prove its reliability,
numerous tests must be undergone to fit more data into the machine learning model since in
this design, the detection of the virus solely relies on the machine learning model alone. The
deep learning machine model which is quite complex for anyone who happens to be a
bystander having come across such a complex computerized program would have to learn
that addition or removal of certain components connected to it would require making
modifications to the machine learning algorithms. The addition of the pre-cooler unit would
mean that the machine learning model must be equipped well enough to handle new process
data to give results, so more knowledge must be fed into the machine learning algorithm
which can be quite troublesome if one is not so familiar with machine learning. Furthermore,
any new design would have to be approved by relevant authorities before it goes into the
market and even after it does get into the market, it would take time for consumers to get
around to the idea of it and therefore, the payback period for the new design may take
relatively longer.

45
References

Adilla, F. (2021). MyEG, Singapore's Breathonix to offer Covid-19 breath test product. New
Straits Time.
https://www.nst.com.my/business/2021/05/692661/myeg-singapores-breathonix-offer-
covid-19-breath-test-product
Advantages and Disadvantages of Capillary Electrophoresis Chromatography Today. (2014,
November 3). Chromatography Today.
https://www.chromatographytoday.com/news/electrophoretic-separations/35/breaking
-news/advantages-and-disadvantages-of-capillary-electrophoresis/32342
Ahlers, P. (2017, February 2). New gas sample cooler TC-Double+ by Buhler. Elemental
Analytics. Retrieved November 18, 2021, from
https://e-analytics.co.za/new-gas-sample-cooler-tc-double-by-buhler/
Al Suwaidi, H., Senok, A., Varghese, R., Deesi, Z., Khansaheb, H., & Pokasirakath, S. et al.
(2021). Saliva for molecular detection of SARS-CoV-2 in school-age children.
Clinical Microbiology And Infection, 27(9), 1330-1335.
https://doi.org/10.1016/j.cmi.2021.02.009
Alemany, A., Millat-Martinez, P., Ouchi, D., Corbacho-Monné, M., Bordoy, A., & Esteban,
C. et al. (2021). Self-collected mid-nasal swabs and saliva specimens, compared with
nasopharyngeal swabs, for SARS-CoV-2 detection in mild COVID-19 patients.
Journal Of Infection. https://doi.org/10.1016/j.jinf.2021.09.012
Alkhateeb, K., Cahill, M., Ross, A., Arnold, F., & Snyder, J. (2021). The reliability of saliva
for the detection of SARS-CoV-2 in symptomatic and asymptomatic patients: Insights
on the diagnostic performance and utility for COVID-19 screening. Diagnostic
Microbiology And Infectious Disease, 101(3), 115450.
https://doi.org/10.1016/j.diagmicrobio.2021.115450
Aydin, S., Benk, I., & Geckil, A. (2021). May viral load detected in saliva in the early stages
of infection be a prognostic indicator in COVID-19 patients?. Journal Of Virological
Methods, 294, 114198. https://doi.org/10.1016/j.jviromet.2021.114198
Baro, B., Rodo, P., Ouchi, D., Bordoy, A. E., Amaro, E. N. S., Salsench, S. V., Molinos, S.,
Alemany, A., Ubals, M., Corbacho-Monne, M., Prat, N., Estrada, O., Vilar, M., Ara,
J., Vall-Mayans, M., G-Beiras, C., Bassat, Q., Blanco, I., & Mitja, O. (2021).
Performance characteristics of five antigen-detecting rapid diagnostic test (AG-RDT)
for SARS-CoV-2 asymptomatic infection: a head-to-head benchmark comparison.
Journal of Infection, 82, 269-275. 10.1016/j.jinf.2021.04.009
Bastos, M., Perlman-Arrow, S., Menzies, D., & Campbell, J. (2021). The sensitivity and costs
of testing for SARS-CoV-2 infection with saliva versus nasopharyngeal swabs. Annals
Of Internal Medicine, 174(4), 501-510. https://doi.org/10.7326/m20-6569
Breadmore, M. (2020). Repurposing technology : When explosive detection morphs into virus
detection. Transport Security International.
https://www.tsi-mag.com/repurposing-technology-when-explosive-detection-morphs-i
nto-virus-detection/

46
BreFence™ Go COVID-19 breath test system. (2021). Breathonix.
https://breathonix.com/covid-19/
Cao, Y., Yu, M., Dong, G., Chen, B., & Zhang, B. (2020). Digital PCR as an emerging tool
for monitoring of microbial biodegradation. Molecules, 25(3), 706.
Calculator: Saturated Humid Air Table. (n.d.). TLV. Retrieved November 19, 2021, from
https://www.tlv.com/global/TI/calculator/humid-air-table.html?advanced=on
Carleo, G., Cirac, I., Cranmer, K., Daudet, L., Schuld, M., Tishby, N., ... & Zdeborová, L.
(2019). Machine learning and the physical sciences. Reviews of Modern Physics,
91(4), 045002.
Castillo-Henriquez, L., Brenes-Acuna, M., Castro-Rojas, A., Cordero-Salmeron, R.,
Lopretti-Correa, M., & Vega-Baudrit, J. R. (2020). Biosensors for the detection of
bacterial and viral clinical pathogens. Sensors, 20, 2692. 10.3390/s20236926
Chen, H., Qi, X., Zhang, L., Li, X., Ma, J., Zhang, C., ... & Yao, M. (2021). COVID-19
screening using breath-borne volatile organic compounds. Journal of Breath
Research.
Chen, L., Zhao, J., Peng, J., Li, X., Deng, X., & Geng, Z. et al. (2020). Detection of
SARS‐CoV‐2 in saliva and characterization of oral symptoms in COVID‐19
patients. Cell Proliferation, 53(12). https://doi.org/10.1111/cpr.12923
Chew, H. M. (2021, June 14). Singapore firm that developed COVID-19 breath test pivoted
from detecting lung cancer to coronavirus. CNA.
https://www.channelnewsasia.com/singapore/covid-19-breath-test-breathonix-nus-sin
gapore-lung-cancer-1840826
Corman, V. M., Haage, V. C., Bleicker, T., Schmidt, M. L., Muhlemann, B., Zuchowski, M.,
Jo, W. K., Tscheak, P., Moncke-Buchner, E., Muller, M. A., Krumbholz, A., Drexler,
J., & Drosten, C. (2021). Comparison of seven commercial SARS-CoV-2 rapid
point-of-care antigen tests: a single-centre laboratory evaluation study. Lancet
Microbe, 1-9. 10.1016/S2666-5247(21)00056-2
CoronaCheck. (2021). Exhalation Medical Technology.
https://www.exhalationmedicaltechnology.com/
COVID-19 and PCR Testing. (2021). Cleveland Clinic.
https://my.clevelandclinic.org/health/diagnostics/21462-covid-19-and-pcr-testing
COVID-19 Antigen Saliva Test Kit Self Test. (2021). [Ebook]. Retrieved 4 November 2021,
from
https://www.tga.gov.au/sites/default/files/covid-19-rapid-antigen-self-tests-are-approv
ed-australia-ifu-372335.pdf.
Covid-19 Breathanalyser Screening Tools. (2020). Ministry of Health, Malaysia.
https://covid-19.moh.gov.my/kajian-dan-penyelidikan/mahtas-covid-19-rapid-evidenc
e-updates/10_COVID-19_Breathanalyser_screening_tools.pdf
Covid-19 self-test kits in Malaysia: A guide to the types, efficacy rates, and where to buy
them. BURO. (2021). Retrieved 10 November 2021, from
https://www.buro247.my/lifestyle/news/covid-19-self-test-kits-guide-malaysia.html.
Covid-19 Tests: RT-PCR vs RTK antigen test. (2021). Caring Pharmacy.
https://caring2u.com/health_info/covid-19-tests-rt-pcr-vs-rtk-antigen/

47
Covid-19: Why test? Who to test? How to test?. (2020), 204(9), e9-e10.
https://doi.org/10.1016/j.banm.2020.05.006
de Oliveira, C., Brochi, L., Scarpelli, L., Lopes, A., & Levi, J. (2021). SARS-CoV-2 saliva
testing is a useful tool for Covid-19 diagnosis. Journal Of Virological Methods, 296,
114241. https://doi.org/10.1016/j.jviromet.2021.114241
Exhalation Technology,, & The Sage Group,. (2021). CoronaCheck Update - Unique
Commercial Opportunity for Rapid (<5mins) Covid-19 Virus Detection. United
Kingdom.
Fakheran, O., Dehghannejad, M., & Khademi, A. (2020). Saliva as a diagnostic specimen for
detection of SARS-CoV-2 in suspected patients: a scoping review. Infectious Diseases
Of Poverty, 9(1). https://doi.org/10.1186/s40249-020-00728-w
Favresse, J., Gillot, C., Oliveira, M., Cadrobbi, J., Elsen, M., Eucher, C., Laffineur, K.,
Rosseels, C., Eeckhoudt, S. V., Nicolas, J.-B., Morimont, L., Dogne, J.-M., &
Douxfils, J. (2021). Head-to-head comparison of rapid and automated antigen
detection tests for the diagnosis of SARS-CoV-2 infection. Journal of Clinical
Medicine, 10, 265. 10.3390/jcm10020265
Febrian, R. (2021). How Does GeNose Work, Why Is It So Much Criticized? VOI.
https://voi.id/en/bernas/41898/how-does-genose-work-why-is-it-so-much-criticized
Felix, F. S., & Angnes, L. (2018). Electrochemical immunosensors–a powerful tool for
analytical applications. Biosensors and Bioelectronics, 102, 470-478.
Fukumoto, T., Iwasaki, S., Fujisawa, S., Hayasaka, K., Sato, K., & Oguri, S. et al. (2020).
Efficacy of a novel SARS-CoV-2 detection kit without RNA extraction and
purification. International Journal Of Infectious Diseases, 98, 16-17.
https://doi.org/10.1016/j.ijid.2020.06.074
GeNose C19-S,. (2021). Buku Manual dan Spesifikasi (Issue User Versi 1.0.7). Indonesia.
Retrieved November 9, 2021, from
https://ditpui.ugm.ac.id/wp-content/uploads/sites/322/2021/01/Manual-Book-GeNose-
v.1.0.7-1.pdf
Guglielmi, G. (2020). Fast coronavirus tests: what they can and can’t do. Nature, 585(7826),
496-498. https://doi.org/10.1038/d41586-020-02661-2
Hall, E., Luisi, N., Zlotorzynska, M., Wilde, G., Sullivan, P., & Sanchez, T. et al. (2020).
Willingness to use home collection methods to provide specimens for
SARS-CoV-2/COVID-19 research: survey study. Journal Of Medical Internet
Research, 22(9), e19471. https://doi.org/10.2196/19471
He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In
Proceedings of the IEEE conference on computer vision and pattern recognition (pp.
770-778).
Hsiao, W., Le, T., Pham, D., Ko, H., Chang, H., & Lee, C. et al. (2021). Recent Advances in
Novel Lateral Flow Technologies for Detection of COVID-19. Biosensors, 11(9), 295.
https://doi.org/10.3390/bios11090295
Huang, N., Pérez, P., Kato, T., Mikami, Y., Okuda, K., & Gilmore, R. et al. (2021).
SARS-CoV-2 infection of the oral cavity and saliva. Nature Medicine, 27(5), 892-903.
https://doi.org/10.1038/s41591-021-01296-8

48
Huggett, J., Dheda, K., Bustin, S., & Zumla, A. (2005). Real-time RT-PCR normalisation;
strategies and considerations. Genes & Immunity, 6(4), 279-284.
Industry News: Groundbreaking clinical trial for CoronaCheck: A rapid COVID-19 breath
test from ETL. (2021). Select Science.
https://www.selectscience.net/industry-news/groundbreaking-clinical-trial-for-coronac
heck-a-rapid-covid-19-breath-test-from-etl/?artID=53469
Iravani, S. (2020). Nano and biosensors for detection of SARS-CoV-2: challenges and
opportunities. Materials Advances, 1-37. 10.1039/D0MA00702A
Jalali, M., Zaborowska, J., & Jalali, M. (2017). The polymerase chain reaction: PCR, qPCR,
and RT-PCR. In Basic science methods for clinical researchers (pp. 1-18). Academic
Press.
Jawerth, N. (2020). How is the COVID-19 Virus Detected using Real Time RT-PCR?
International Atomic Energy Agency.
https://www.iaea.org/newscenter/news/how-is-the-covid-19-virus-detected-using-real-
time-rt-pcr
Jayatilaka, T. (2021). Covid-19: self-test kits in Malaysia & how they work. Tatler Asia.
Retrieved 11 November 2021, from
https://www.tatlerasia.com/style/wellness/covid-19-self-test-kit-moh-malaysia
Johnson, A., Zhou, S., Hoops, S., Hillmann, B., Schomaker, M., & Kincaid, R. et al. (2021).
Saliva testing is accurate for early-stage and presymptomatic COVID-19.
Microbiology Spectrum, 9(1). https://doi.org/10.1128/spectrum.00086-21
Kadri, K. (2019). Polymerase chain reaction (PCR): principle and applications. In Synthetic
Biology-New Interdisciplinary Science. IntechOpen.
Karki, G. (2017). Polymerase chain reaction (PCR): Principle, procedure or steps, types and
application. Online Biology Notes.
https://www.onlinebiologynotes.com/polymerase-chain-reaction-pcr-principle-proced
ure-steps-types-application/
Kesumapramudya, D. (2021, March 2). Indonesian-made COVID-19 breathalyser sensitivity
comparable to RT-PCR. The Conversation.
https://theconversation.com/indonesian-made-covid-19-breathalyser-sensitivity-comp
arable-to-rt-pcr-155497
Khandker, S. S., Nik Hashim, N. H. H., Deris, Z. Z., Shueb, R. H., & Islam, M. A. (2021).
Diagnostic accuracy of rapid antigen test kits for detecting SARS-CoV-2: a systematic
review and meta-analysis of 17,171 suspected COVID-19 patients. Journal of Clinical
Medicine, 10, 3493. 10.3390/jcm10163493
Koczula, K., & Gallotta, A. (2016). Lateral flow assays. Essays In Biochemistry, 60(1),
111-120. https://doi.org/10.1042/ebc20150012
Landry, M., Criscuolo, J., & Peaper, D. (2020). Challenges in use of saliva for detection of
SARS CoV-2 RNA in symptomatic outpatients. Journal Of Clinical Virology, 130,
104567. https://doi.org/10.1016/j.jcv.2020.104567
Limsakul, P., Charupanit, K., Moonla, C., & Jeerapan, I. (2021). Advances in emergent
biological recognition elements and bioelectronics for diagnosing COVID-19.
Emergent Materials, 4, 231-247. 10.1007/s42247-021-00175-9

49
Lopes, R. d. l. F. (2018). Wild data part 3: Transfer learning. Stratio.
https://blog.stratio.com/wild-data-part-three-transfer-learning/
Mason, B. (2020). Exhalation Technology (ETL) announces new clinical trial for
CoronaCheck™ – a rapid Covid-19 test for exhaled breath. Biospace.
https://www.biospace.com/article/exhalation-technology-etl-announces-new-clinical-t
rial-for-coronacheck-a-rapid-covid-19-test-for-exhaled-breath/
Mcphee, E. (2021). Australian company develops test that will tell you if you have
coronavirus in just three minutes - and all you have to do is breathe into the device.
Mail Online.
https://www.dailymail.co.uk/news/article-9263791/Melbourne-based-company-GreyS
can-develops-breath-test-Covid-19-results-three-minutes.html
Ministry of Health Malaysia. (2021, Oct 29). List of Recommended for Use of Covid-19 IVD
Test kit (Professional use only). Medical Device Authority Malaysia Official Portal.
Retrieved Oct 29, 2021, from
https://www.mda.gov.my/announcement/596-list-of-recommended-for-use-of-covid-1
9-ivd-test-kit.html
Mojsoska, B., Larsen, S., Olsen, D. A., Madsen, J. S., Brandslund, I., & Alatraktchi, F. A.
(2021). Rapid SARS-CoV-2 detection using electrochemical immunosensor. Sensors,
21(2), 390.
Neidler, S. (2017). What are the differences between PCR, RT-PCR, qPCR, and RT-qPCR?
ENZO.
https://www.enzolifesciences.com/science-center/technotes/2017/march/what-are-the-
differences-between-pcr-rt-pcr-qpcr-and-rt-qpcr?/
Nurputra, D. K., Kusumaatmadja, A., Hakim, M. S., Hidayat, S. N., Julian, T., Sumanto, B.,
Mahendradhata, Y., Saktiawati, A. M. I., Wasisto, H. S., & Triyana, K. (2021). Fast
and noninvasive electronic nose for sniffing out COVID-19 based on exhaled
breath-print recognition. Research Square, 1-37. 10.21203/rs.3.rs-750988/v1
Nurputra, D. K., Triyana, K., & Sasongko, T. H. (2021). Indonesian-made COVID-19
breathalyser sensitivity comparable to RT-PCR.
https://theconversation.com/indonesian-made-covid-19-breathalyser-sensitivity-comp
arable-to-rt-pcr-155497
Orawell COVID-19 Ag Rapid Saliva Test Device. (2021). [Ebook]. Retrieved 14 November
2021, from
https://www.tga.gov.au/sites/default/files/covid-19-rapid-antigen-self-tests-are-approv
ed-australia-ifu-377136.pdf.
Ott, I., Strine, M., Watkins, A., Boot, M., Kalinich, C., & Harden, C. et al. (2020). Simply
saliva: stability of SARS-CoV-2 detection negates the need for expensive collection
devices. https://doi.org/10.1101/2020.08.03.20165233
Pohl, H. (2020). How do Coronavirus Disease (COVID-19) Tests Work? ENZO.
https://www.enzolifesciences.com/science-center/technotes/2020/july/how-do-corona
virus-disease-covid-19-tests-work?/
Precooler TS 10. (2021). [Ebook]. Retrieved 17 November 2021, from
https://www.buehler-technologies.com/uploads/tx_szdownloadcenter/DA450011_TS1
0.pdf.

50
Purva, H. (2020). Top 4 pre-trained models for image classification with Python code.
Analytics Vidhya.
https://www.analyticsvidhya.com/blog/2020/08/top-4-pre-trained-models-for-image-c
lassification-with-python-code/
Rahman, R. (2021). COVID-19 Swab Tests in Malaysia: What Is It and Where Can You Get It
Done? Health Facilities.
https://www.homage.com.my/resources/covid-swab-test-malaysia/
Rasmi, Y., Li, X., Khan, J., Ozer, T., & Choi, J. R. (2021). Emerging point-of-care biosensors
for rapid diagnosis of COVID-19: current progress, challenges, and future prospects.
Analytical and Bioanalytical Chemistry. 10.1007/s00216-021-03377-6
Saki, E. F., Setiawan, S. A., & Wicaksono, D. H. (2021). Portable tools for COVID-19
point-of-care detection: a review. IEEE Sensors Journal.
Schoeman, D., & Fielding, B. C. (2019). Coronavirus envelope protein: current knowledge.
Virology Journal, 16(1), 1-22.
Schuit, E., Veldhuijzen, I. K., Venekamp, R. P., Bijllaardt, W. v. d., Pas, S. D., Lodder, E. B.,
Molenkamp, R., GeurtsvanKessel, C. H., Velzing, J., Huisman, R. C., Brouwer, L.,
Boelsums, T. L., Sips, G. J., Benschop, K. S. M., Hooft, L., van de Wijgert, J. H. H.
M., van den Hof, S., & Moons, K. G. M. (2021). Diagnostic accuracy of rapid antigen
tests in asymptomatic and presymptomatic close contacts of individuals with
confirmed SARS-CoV-2 infection: cross sectional study. BMJ, 374:n1676.
10.1136/bmj.n1676
Seo, G., Lee, G., Kim, M. J., Baek, S.-H., Choi, M., Ku, K. B., Lee, C.-S., Jun, S., Park, D.,
Kim, H. G., Kim, S.-J., Lee, J.-O., Kim, B. T., Park, E. C., & Kim, S. I. (2020). Rapid
detection of COVID-19 causative virus (SARS-CoV-2) in human nasopharyngeal
swab specimens using field-effect transistor-based biosensor. ACS Nano, 14,
5135-5142. 10.1021/acsnano.0c02823
Shachar, J., Felgner, P., & Madou, M. (2021). Automated, cloud-based, point-of-care (POC)
pathogen and antibody array detection system and method (Issue US Patent No.
11,130,994 B2).
Simanjuntak, E. Y., Octavia, Y. T., & Sinaga, J. (2021). Optimalisasi pencegahan melalui
deteksi dini penularan Covid-19 menggunakan Genose C19. JUKESHUM: Jurnal
Pengabdian Masyarakat, 1(2), 101-106.
Singapore to issue digital verifiable COVID-19 breath test results. (2021). BioSpectrum.
https://www.biospectrumasia.com/news/54/18653/singapore-to-issue-digital-verifiabl
e-covid-19-breath-test-results-.html
Singh, B., Datta, B., Ashish, A., & Dutta, G. (2021). A comprehensive review on current
COVID-19 detection methods: from lab care to point of care diagnosis. Sensors
International, 2(100119). 10.1016/j.sintl.2021.100119
Sipayung, I. (2021). Uniting Aviation. A game changer for air travel: Indonesia’s COVID-19
breath analyzer.
https://unitingaviation.com/news/safety/a-game-change-for-air-travel-indonesias-covi
d-19-breath-analyzer/

51
Snitz, K., Andelman-Gur, M., Pinchover, L., Weissgross, R., Weissbrod, A., Mishor, E., ... &
Sobel, N. (2021). Proof of concept for real-time detection of SARS CoV-2 infection
with an electronic nose. PloS one, 16(6), e0252121.
Tocco, J. (2020). GreyScan announces Development of First Fieldable Trace Virus Detection
(TVD-1). Grey Innovation.
https://www.greyinnovation.com/newsroom/article/20/greyscan-announces-developm
ent-of-first-fieldable-trace-virus-detection-tvd-1.html
TVD-2 Breathalyser. (2021). GreyScan.
https://greyscandetection.com/products/details/24/tvd-2-breathalyser.html
Udvardi, M. K., Czechowski, T., & Scheible, W. R. (2008). Eleven golden rules of
quantitative RT-PCR. The Plant Cell, 20(7), 1736-1737.
Valentine-Graves, M., Hall, E., Guest, J., Adam, E., Valencia, R., & Shinn, K. et al. (2020).
At-home self-collection of saliva, oropharyngeal swabs and dried blood spots for
SARS-CoV-2 diagnosis and serology: Post-collection acceptability of specimen
collection process and patient confidence in specimens. PLOS ONE, 15(8), e0236775.
https://doi.org/10.1371/journal.pone.0236775
Vaz, S., Santana, D., Netto, E., Pedroso, C., Wang, W., Santos, F., & Brites, C. (2020). Saliva
is a reliable, non-invasive specimen for SARS-CoV-2 detection. The Brazilian
Journal Of Infectious Diseases, 24(5), 422-427.
https://doi.org/10.1016/j.bjid.2020.08.001
Verdict. (2021). Breathonix’s Covid-19 breath test gets provisional approval in Singapore.
https://www.medicaldevice-network.com/news/breathonix-breath-test-singapore/
Wang, C., Yin, L., Zhang, L., & Gao, R. (2010). Metal oxide gas sensors: sensitivity and
influencing factors. Sensors, 10(3), 2088-2106. 10.3390/s100302088
What is a lateral flow test and how does the technology work?. Abingdon Health plc. (2021).
Retrieved 4 November 2021, from
https://www.abingdonhealth.com/services/what-is-lateral-flow-immunoassay/.
World Health Organization. (2021, October 6). Antigen-detection in the diagnosis of
SARS-CoV-2 infection: Interim guidance.
https://www.who.int/publications/i/item/antigen-detection-in-the-diagnosis-of-sars-co
v-2infection-using-rapid-immunoassays
Wu, J., & Ju, H.X. (2012). Comprehensive Sampling and Sample Preparation. Reference
Module in Chemistry, Molecular Sciences and Chemical Engineering, 3.
https://www.sciencedirect.com/science/article/pii/B9780123813732000715
Xi, H., Jiang, H., Juhas, M., & Zhang, Y. (2021). Multiplex biosensing for simultaneous
detection of mutations in SARS-CoV-2. ACS Omega, 6, 25846-25859.
10.1021/acsomega.1c04024
Xi, H., Jiang, H., Juhas, M., & Zhang, Y. (2021). Multiplex biosensing for simultaneous
detection of mutations in SARS-CoV-2. ACS Omega, 6, 25846-25859.
10.1021/acsomega.1c04024
Yokota, I., Shane, P., Okada, K., Unoki, Y., Yang, Y., & Inao, T. et al. (2020). Mass screening
of asymptomatic persons for SARS-CoV-2 using saliva. SSRN Electronic Journal.
https://doi.org/10.2139/ssrn.3668435

52
Zaini, Z. (2021). Self-test COVID-19 test kit for conditional approval (approved) - Medical
Device Authority (MDA). Portal.mda.gov.my. Retrieved 11 November 2021, from
https://portal.mda.gov.my/announcement/631-self-test-covid-19-test-kit-for-condition
al-approval-approved.html.

53
Appendices

Appendix A: Request for TVD-2 Breathanalyser Information via Email to


GreyScan

Appendix B: Request for BreFence Information via Email to Brethnoxix

54
Appendix C: Reply on CoronaCheck Information via Email from ETL

55

You might also like