Evaluation of rapid detection methods for H5N1 virus using biosensors: An AI-based study

Abstract

High mortality and zoonotic potential predispose the H5N1 avian influenza virus as a critical threat. Knowing that an epidemic could be occurring, quick and precise diagnostic techniques are essential for managing and containing possible epidemics. To detect H5N1 in saliva samples, this study investigates the theoretical design, simulation and evaluation of three kind of biosensors based on different technologies with potential as rapid identifications tools to diagnose quickly H5N1: Lateral Flow Tests (LFT), Field Effect transistors (FET) based electrochemical sensors and Quartz Crystal Microbalance (QCM) sensors. Through detailed AI-based simulations, we show the capabilities, sensitivities and specificities of these biosensors, highlighting their potential for applications in general biology as well as their suitability both for routine home practice and for applications by control entities in public settings. We therefore wish to pave the way to a framework for the quick creation of detection tools that can be swiftly implemented for rapid deployment in case of an outbreak of disease.

Full text

ISSN 0973-2063 (online) 0973-8894 (print)
©Biomedical Informatics (2024)
Bioinformation 20(11): 1516-1523 (2024)
1516
www.bioinformation.net
Research Article
Volume 20(11)
Received November 1, 2024; Revised November 30, 2024; Accepted November 30, 2024, Published November 30, 2024
DOI: 10.6026/9732063002001516
BIOINFORMATION 2022 Impact Factor (2023 release) is 1.9.
Declaration on Publication Ethics:
The author’s state that they adhere with COPE guidelines on publishing ethics as described elsewhere at https://publicationethics.org/. The authors
also undertake that they are not associated with any other third party (governmental or non-governmental agencies) linking with any form of
unethical issues connecting to this publication. The authors also declare that they are not withholding any information that is misleading to the
publisher in regard to this article.
Declaration on official E-mail:
The corresponding author declares that lifetime official e-mail from their institution is not available for all authors
License statement:
This is an Open Access article which permits unrestricted use, distribution, and reproduction in any medium, provided the ori ginal work is properly
credited. This is distributed under the terms of the Creative Commons Attribution License
Comments from readers:
Articles published in BIOINFORMATION are open for relevant post publication comments and criticisms, which will be published immediately
linking to the original article without open access charges. Comments should be concise, coherent and critical in less than 1000 words.
Disclaimer:
The views and opinions expressed are those of the author(s) and do not reflect the views or opinions of Bioinformation and (or) its publisher
Biomedical Informatics. Biomedical Informatics remains neutral and allows authors to specify their address and affiliation details including territory
where required. Bioinformation provides a platform for scholarly communication of data and information to create knowledge in the
Biological/Biomedical domain.
Edited by Francesco Chiappelli
Citation: Eggenhöffner et al. Bioinformation 20(11): 1516-1523 (2024)
Evaluation of rapid detection methods for H5N1
virus using biosensors: An AI-based study
Roberto Eggenhöffner1, Paola Ghisellini1, Cristina Rando1, Simonetta Papa2, Allen Khakshooy3
& Luca Giacomelli2,*
1Department of Surgical Sciences and Integrated Diagnostics (DISC), University of Genova, Corso Europa 30, Genova - 16132, Italy;
2Polistudium SRL, Milan, Italy; 3Department of Internal Medicine, Valley Hospital Medical Center, Las Vegas, NV - 89106, USA;
*Corresponding author
Affiliation URL:
https: //disc.unige.it
https://www.polistudium.it
https://www.valleyhospital.net/
Author contacts:
Roberto Eggenhöffner - E - mail: roberto.eggenhoffner@unige.it

ISSN 0973-2063 (online) 0973-8894 (print)
©Biomedical Informatics (2024)
Bioinformation 20(11): 1516-1523 (2024)
1517
Paola Ghisellini - E - mail: paola.ghisellini@unige.it
Cristina Rando - E - mail: cristina.rando@unige.it
akhakshooy@gmail.com Allen Khakshooy - E - mail:
Luca Giacomelli - E - mail: luca.giacomelli@polistudium.it
Simonetta Papa - E - mail: Simonetta.papa@polistudium.it
Preparing for the emerging H5N1 pandemic: Lessons learned from HIV/AIDS and SARSCov2/CoViD19
Abstract:
High mortality and zoonotic potential predispose the H5N1 avian influenza virus as a critical threat. Knowing that an epidemic could
be occurring, quick and precise diagnostic techniques are essential for managing and containing possible epidemics. To detect H5N1
in saliva samples, this study investigates the theoretical design, simulation and evaluation of three kind of biosensors based on
different technologies with potential as rapid identifications tools to diagnose quickly H5N1: Lateral Flow Tests (LFT), Field Effect
transistors (FET) based electrochemical sensors and Quartz Crystal Microbalance (QCM) sensors. Through detailed AI-based
simulations, we show the capabilities, sensitivities and specificities of these biosensors, highlighting their potential for applications in
general biology as well as their suitability both for routine home practice and for applications by control entities in public settings.
We therefore wish to pave the way to a framework for the quick creation of detection tools that can be swiftly implemented for rapid
deployment in case of an outbreak of disease.
Keywords: H5N1 virus detection, biosensors, lateral flow, field-effect transistor, quartz crystal microbalance
Background:
H5N1 is an avian virus, whose mutated variants circulates and
spreads amongst mammals. The influenza caused by this virus is
becoming a serious public health threat that can lead to severe
respiratory illness and death, in animals and in humans [1-3].
The key to managing viral outbreaks that may have epidemic or
pandemic potential is early identification and fast diagnosis, as
well as precluding further spread. Case in point, the Corona
Virus Disease-2019 (CoViD-19) pandemic has made evident the
need for rapid and reliable diagnostic tools to screen, diagnose
and manage viral infections [4]. Technologies such as lateral
flow tests (LFTs) and electrochemical sensors, adopted to face
the spread of previous pandemics such as CoViD-19, played
pivotal roles in the quick identification and isolation of infected
individuals in the last years. Accordingly, this study describes
the development of computer simulations and testing of LFT,
Field Effect transistors (FET)-based electrochemical and Quartz
Crystal microbalance (QCM) diagnostic sensors for H5N1
detection with an emphasis on rapid preparedness for
deployment in case of a H5N1 epidemic/pandemic outbreak.
Gold Nanoparticle-based LFT act as ideal biosensors for quick
and sensitive assessment of viral nucleic acids. A Gold
Nanoparticle-based LFT was also established for the detection of
fish nervous necrosis virus [5-6]. The choice of a FET biosensor is
recommended because it inherently permits an ultra-high
response speed [7]. There were few advanced versions
developed like graphene-based/MoS2/Silicon nanowires
exhibited high sensitivity for SARS-CoV-2 and other viral
nucleic acids detection. The biosensors based on the QCM are
introduced by coating antibody onto plates to detect viral
proteins and whole intact viruses with high specificity. For
example, a QCM biosensor was developed for detecting
antibodies against the African swine fever virus with high
sensitivity [8]. In other settings, aptamer-based QCM was
adopted for the sensitive detection of leukemia cells,
demonstrating potential for viral detection through specific
nucleic acid binding [9]. AI-based approaches have a mounting
role in the development of computer simulations, which guide
researchers and decision-makers in building hypotheses based
on a stronger scientific foundation. This methodology has been
applied in various fields, and biosensors are no different [10].
Our focus characterized the detection of H5N1 in saliva samples,
which offer a less invasive and more practical means of sample
collection for diagnosis. By simulating the performance of these
biosensors at various virus concentrations, we showed here their
potential capability for actual applications in the context of viral
infections and emphasize the importance of being prepared with
advanced diagnostic tools to quickly respond to emerging viral
threats. A comparison between LFT for H5N1 with the LFT for
SARS-CoV-2 is also provided.
Methods:
Development of lateral flow tests (LFT):
The test strip is composed of an H5N1 virus-specific capture
antibody-labeled conjugate zone, a test zone with anti-species
antibodies to verify the test's validity and a sample pad
collecting saliva on a nitrocellulose membrane. Monoclonal
antibodies are used to detect H5N1 targeting the hemagglutinin
(HA) and neuraminidase (NA) proteins. Saliva samples
prepared to express known concentrations of H5N1 from the
lowest to highest values are placed on the sample pad for the
generation of a standard curve. In the conjugate zone, the
unknown sample interacts with the labeled antibodies as it
migrates through the membrane via capillary action. A color
shift in the test and control lines is used to achieve visualization;
the test zone shows the presence of H5N1 and the control zone
verifies the test's functionality [11]. The study modeled the
optical density (OD) response of LFT biosensors for detecting
salivary H5N1 virus particles. The migration of samples through
the membrane plate was modeled using the Lucas-Washburn
equation, incorporating parameters such as surface tension, pore
radius and viscosity. The antibody-antigen interaction in LFT

ISSN 0973-2063 (online) 0973-8894 (print)
©Biomedical Informatics (2024)
Bioinformation 20(11): 1516-1523 (2024)
1518
was modeled by the Langmuir adsorption model, with specific
association and dissociation rate constants for H5N1 antibodies.
Development of FET-based electrochemical sensors:
The FET-based sensor exploits the unmatched electrical
properties of graphene as the substrate material in
semiconducting devices. Gold electrodes are deposited on the
graphene to serve as the source and drain, while the gate is
functionalized with monoclonal HA antibodies specific to H5N1.
Thiol groups are used to attach the antibodies to the graphene
gate. Saliva samples are applied to the functionalized gate,
where the binding of H5N1 to the antibodies causes a change in
the electrical properties of the graphene. This change is
measured as a variation in the current (ΔI) using a potentiostat,
providing a quantitative indication of the virus concentration.
The sensitivity of the FET sensor allows detecting concentrations
as low as 10 particles/mL the lowest limit [12].
Development of quartz crystal microbalance (QCM) sensors:
QCM sensors are based on the principle that a change in mass on
the surface of a vibrating quartz crystal affects its resonant
frequency. For H5N1 detection, the quartz crystal is coated with
monoclonal antibodies specific to the HA and NA proteins. As
H5N1 particles bind to these antibodies, the added mass causes a
measurable change in the crystal's frequency. The QCM sensor
setup involves preparing the quartz crystal with a thin gold
layer, onto which antibodies are immobilized using self-
assembled monolayers (SAMs) of thiols. Saliva samples are
applied to the sensor and the binding of H5N1 virus, when
present, produces a resulting frequency shift that can be
monitored with high accuracy using a frequency counter. In
turn, this method provides a direct and quantitative
measurement of the virus concentration [13]. Mass changes due
to virus adsorption on the functionalized gold surface are
measured as frequency shifts. Mass and frequency variations are
related throughout the following Sauerbry’s well-known
equation [14]:
Where Δf is the frequency change, Δm the mass change, f0 the
fundamental frequency of the quartz crystal (5 or 10 MHz
usually), A the area of the gold electrode on the crystal plate
(typical value 0.2 cm²), ρ the density of quartz plate (2.65 g/cm³),
μ the shear modulus of quartz (2.95 × 10¹¹ g/cm s²).
Simulations of detections with the three biosensors:
To evaluate the performance of the developed biosensors, we
conducted AI-based simulations considering different
concentrations of H5N1 in saliva samples and their respective
limits of detection (LOD). Table 1 summarizes the approach
used for each simulation. In the case of SARS-CoV-2, the median
viral load in posterior oropharyngeal saliva or other respiratory
specimens at presentation was approximately 5.2 log10
copies/mL, with an interquartile range (IQR) of 4.1–7.0 log10
copies/mL [15]. The average PCR cycle threshold values ranged
from were 29 to 31 log10 copies/ml for symptomatic and
asymptomatic cases respectively [16]. Fewer specific data on
salivary viral load are available for H5N1, compared to SARS-
CoV-2. However, similar methods of detection using saliva
samples were suggested based on respiratory tract sampling and
antibody presence [17]. It follows that the three sensors
considered in the present work should be expected to measure
virus concentrations ranging from 104.1 to 107.0 units/mL
accordingly to Kim et al. 2020 [18].
Results:
Simulations performed in the reported range confirmed the
concentration detection limits (i.e., LOD) 104.1 to 107.0 units/mL.
We report below the change in color
intensity/current/frequency measured to determine the sensor's
response to varying concentrations of H5N1.
Simulation of a lateral flow test (LFT) for H5N1 in saliva:
The simulation for the LFT involves testing H5N1 concentrations
in saliva at the limiting levels: low (104 particles/mL) and high
(107 particles/mL), with LOD at 104 particles/mL. The optical
density response was modeled over time, considering delays
before visibility for low (104 particles/mL) and high (107
particles/mL) concentrations (Appendix 1). The results showed
a significant optical density increase after 5 minutes for low
concentrations and 2 minutes for high concentrations. Enhanced
scaling factors ensured detectability even at low viral loads.
Given the human retina's logarithmic response to light, the
optical density responses were visualized on a logarithmic scale
for clear comparison. At low concentrations near the detection
limit, a slight discoloration in the test zone is expected, which
produces a faint but discernible stain on the optical response
chart. At high concentrations, much above the detection limit,
the test zone exhibits an intense and saturated stain, which
indicates a high viral load. These results predict the LFT's
capability for reliable initial screening correlating with virus
concentration. Figure 1 illustrates the optical density response of
an LFT biosensor over time for the above concentration limits of
salivary H5N1 virus particles. At low concentration (104
particles/mL, blue curve), optical density increases slightly and
then stabilizes at a lower intensity, indicating low viral particle
detection. At high concentration (107 particles/mL, orange
curve), the response shows a rapid and significant increase,
stabilizing at a higher intensity value. Taken together, this
signifies high viral particle detection, with a strong signal
maintained after the initial increase. Overall, the comparison of
the two curves in Figure 1 shows that the sensor can produce
discernible signals at various viral particle concentrations. The
response is faster and noticeably more intense at high
concentrations and slower and less intense at low
concentrations. The differentiation seen is essential for precise
viral load monitoring and detection in diagnostic applications.

ISSN 0973-2063 (online) 0973-8894 (print)
©Biomedical Informatics (2024)
Bioinformation 20(11): 1516-1523 (2024)
1519
Figure 1: Color Intensity Response (OD) vs time in minutes of a
LFT biosensors to Different Concentration of H5N1 Virus
Particles in Saliva (blue: low, orange: high) from the simulation
model. Curve blue: low concentration response; curve orange:
high concentration response.
Simulation of a FET-based electrochemical biosensor for H5N1
detection in saliva:
The graphene-based field-effect transistor (FET) or metal-oxide-
semiconductor field-effect transistor (MOSFET) sensor consists
of a graphene channel, gate dielectric, source and drain
electrodes and a passivation layer. Both FET and MOSFET are
types of transistors that control the flow of current between the
sources and drain electrodes via an electric field applied to the
gate. In these devices, the graphene channel, this is a single-layer
graphene sheet, acts as the conductive path between the sources
and drain electrodes. The gate dielectric, a combination of SiO₂
and graphene oxide, provides the necessary capacitance to
control the channel conductivity. The detection mechanism
involves the binding of H5N1 particles to specific receptors
(antibodies of H5N1) on the graphene surface, which changes
the surface charge density Δσ. This binding event modulates the
gate voltage (VGS), thereby affecting the channel conductivity.
The change in surface charge density due to the binding event
results in a measurable change in the drain current (ΔID), which
can be converted to a voltage through an amplification stage.
The simulations of these sensors are performed at the same
concentration levels as the above LFT, i.e. 104 particles/mL and
107 particles/mL concentration values. The FET sensor's limit of
detection limits (LOD) ranges from 10 to 103 depending on
experimental choices [12]. The change in charge density Δσ are
1.60×10−9 C/cm2 and 1.60×10−6 C/cm2 for low and high
concentration limits, respectively. The carrier mobility μ is very
high in graphene, of the order of 5000 cm²/V·s that was
observed to reduce in heavily deposited surfaces, as discussed in
Appendix 2. The main equation that relates the change in drain
current (ΔD) to the relevant physical parameters of the FET
biosensor is as follows:
ΔID= μ VDS Δσ (W/L)
Where VDS is the drain-source voltage, W and L are the width
and length of the channel, respectively. The equation adopted in
the simulations shows that the change in current is directly
proportional to the applied drain-source voltage, the change in
charge density due to viral particle binding and the aspect ratio
of the graphene channel. This relationship helps in
understanding how sensitive the sensor is to changes in viral
particle concentration and how it translates into measurable
electrical signals. The simulated ΔID current at the selected low
and high concentrations, as shown in Figure 2, starts rising at the
initial reaction time and quickly reaches stable saturation values
above the baseline, the detection limit (LOD) of the sensor. This
response reflects the possible sensor's detection of low
concentration of viral particles, with the signal stabilizing at a
low but detectable current value. The rapid rise indicates the
FET sensor's high sensitivity and quick response both for high
and also low levels of viral particles. The baseline current is the
value predicted in absence of external charges linked to surface
antibodies. The ΔID response at high concentrations of viral
particles shows the sensor's ability to handle high concentrations
and produce a strong signal, highlighting its suitability for
detecting high viral loads. A graphene-based FET/MOSFET
exhibits a good sensitivity also at low viral concentrations,
resulting in a significant change in drain current, crucial for early
detection of H5N1 particles in saliva.
Simulations for the quartz crystal microbalance (QCM)
biosensor:
In order to quantitatively simulate the performance of a QCM
biosensor for detecting salivary H5N1, we must focus on the
relationship between virus concentration and frequency shifts. In
our study, we calculated the frequency shifts for a QCM
biosensor to detect salivary H5N1 at two concentration levels:
low (104 particles/mL) and high (107 particles/mL). For the low
concentration, the number of particles is approximately 500 per
50-microliter. Given that each H5N1 particle has a mass of
approximately 100 femtograms (fg), the total mass added to the
sensor is 5 × 10^-11 grams. This results in a frequency shift of
approximately -2.9 Hz, which, although small, is within the
practical measurement capability for QCM sensors. For the high

ISSN 0973-2063 (online) 0973-8894 (print)
©Biomedical Informatics (2024)
Bioinformation 20(11): 1516-1523 (2024)
1520
concentration of 107 particles per milliliter, the number of
particles per 50-microliter drop sample is approximately 5×105.
The total mass added to the sensor is thus 5×10-8 grams,
resulting in a frequency shift of approximately -290 Hz. This
significant frequency shift is well within the measurable range of
QCM sensors, indicating in this case the very high presence of
H5N1. The above findings, displayed in Figure 3, show that the
QCM sensor can accurately quantify salivary H5N1
concentrations and produce detectable frequency shifts that
match the virus load. The sensor showcases its potential utility
in detecting high viral loads in clinical samples by effectively
identifying significant changes at higher concentrations.
Figure 2: Change in drain current (ΔID) over time for graphene-
based FET/MOSFET sensors exposed to different concentrations
of H5N1 particles. The blue line represents the response to a low
concentration of 104particles/mL, showing a significant change
in current at the reaction time, and stabilizing at approximately
24 nA, 10 nA above baseline. The orange line represents the
response to a high concentration of 107particles/mL, with a
rapid increase in current, indicating the sensor's sensitivity and
potential saturation at higher concentrations. The dashed gray
line represents the baseline current in the absence of H5N1
particles.
Discussion:
Performance and potential applications of the three sensors:
Through detailed computer simulations, we showed the
capabilities, sensitivities and specificities of three different
biosensors (LFT, FET and QCM) for the detection of H5N1 in
saliva samples. Our data show that salivary H5N1 can be
quickly and affordably detected using LFT. The easy-to-use
nature and rapid results make LFTs is appropriate for both field
settings and large-scale screenings. The sensitivity of the LFT is
limited at lower viral loads and its dynamic range is effective
within a narrow concentration window from 104 to 107
particles/mL. Thus, it provides qualitative results and therefore
is a highly valuable tool for initial screenings without any
discomfort to the tested subject. LFT may also be of great use for
extensive screening in less-developed countries [3] because it is a
practical protocol that is inexpensive and that requires little
training and no complex equipment. Widespread testing is
possible even in remote locations because they can be conducted
by healthcare professionals with only rudimentary laboratory
experience. Additionally, in situations where prompt diagnosis
can have a substantial impact on the management and
containment of infectious diseases, the quick results that LFTs
provide can be lifesaving. LFTs are especially well-suited for use
in field operations and mobile clinics due to their portability and
simplicity, which makes mass screenings during outbreaks
easier. By contrast, the more advanced method is offered by the
Quartz Crystal Microbalance (QCM) sensor, which uses
frequency shifts to identify mass changes on its surface. The
QCM sensor is suitable for in-depth laboratory analysis due to
its high sensitivity and capacity to generate quantitative data,
with broader dynamic range from 102 to107 particles/mL. By
identifying notable frequency shifts, the QCM sensor detects the
viral load at lower concentrations of H5N1, performing better
than the LFT but being less sensitive than the FET sensor. In
addition, when compared with LFT, QCM may be less useful for
on-site testing as the instrumentation is not easily portable.
Figure 3: Frequency decreases of the QCM Biosensor to Different
Concentrations of H5N1 Virus Particles in Saliva (blue: low
concentration, orange: high concentration)
Given its high sensitivity and the broadest dynamic range of all
three sensors, spanning from 10 particles/mL (LOD) to 107

ISSN 0973-2063 (online) 0973-8894 (print)
©Biomedical Informatics (2024)
Bioinformation 20(11): 1516-1523 (2024)
1521
particles/mL the FET can provide accurate readings and detect
viral load at very low concentrations. The FET biosensor is
perfect for confirming diagnoses in a clinical setting, even
though it does require specific equipment and expertise. Its
sensitivity and accuracy make it a vital tool for confirming the
presence of the virus. In brief, the distinct qualities of each
sensor make for a wide variety of different applications. The LFT
provides rapid, on-the-spot screening yielding results quickly
and clearly. For in-depth laboratory work, the QCM sensor is
ideal for extensive quantitative analysis due to its high
sensitivity which requires handling by a highly trained
researcher due its complex hardware setting. Lastly, the
extraordinary sensitivity and accurate measurement capabilities
of the FET biosensor make it especially helpful in clinical settings
for confirmatory testing and does not require operation by a
dedicated technician. Overall, despite the speed, simplicity and
affordability of LFTs, its shortcomings in terms of sensitivity and
lack of quantitative output draw attention to the necessity of
complementary technologies like QCM and FET sensors, which
provide a more thorough assessment in both clinical and
laboratory settings. Comprehensive studies of viral load are
made possible by the QCM sensor's great precision in detecting
mass changes, even though its portability for accurate
instrumentation may be restricting. The FET biosensor's
unparalleled sensitivity and quantitative precision make it an
indispensable tool for confirmatory testing. In the end, this
multifaceted approach can improve pandemic preparedness and
control efforts by enhancing diagnostic accuracy and response
times [19]. The above findings concerning the three sensors’
properties are summarized in Table 2.
Comparison of LFT for H5N1 with the LFT for SARS-CoV-2:
The SARS-CoV-2 lateral flow biosensor used for the rapid test
consists of a sample pad that collects the salivary or post-nasal
specimen, a conjugate pad with antibodies to gold nanoparticles
targeting the SARS-CoV-2 virus, a nitrocellulose membrane with
a test line featuring immobilized antibodies specific to the SARS-
CoV-2 antigen with a control line of anti-species antibodies and
an absorbent pad to absorb any excess. Visualization is achieved
through a color change in the test and control lines. As presented
in Table 3, the SARS-CoV-2 and H5N1 lateral flow biosensors
jointly share high sensitivity and specificity, fast reaction times,
and visualization mechanisms. Despite being tailored for distinct
viruses and bio-components, their shared features make them
crucial instruments for promptly identifying and treating these
infectious illnesses, significantly supporting public health efforts.
Table 1: AI-based approach for each simulation and key results the measured parameters, response values, sensitivity and dynamic range of each sensor based on
experimental data are focused. The dynamic range data illustrate the span between the detectable minimum and maximum concentrations, highlighting each sensor's
versatility
LFT
FET
QCM
Objective
To model the optical density response of
LFT biosensors for detecting H5N1 virus
particles in saliva.
To measure the change in current (ΔI) as an
indicator of H5N1 concentration in saliva.
To quantify the relationship between virus
concentration and frequency shifts for H5N1
detection in saliva.
Components
and Process
The LFT uses a test strip with an H5N1-
specific capture antibody-labeled
conjugate zone, a test zone with anti-
species antibodies, and a sample pad
collecting saliva on a nitrocellulose
membrane. Sample migration and
antibody-antigen interaction are
modeled.
The FET-based sensor consists of a graphene
substrate with gold electrodes. The gate is
functionalized with HA antibodies specific to
H5N1. Saliva samples are applied to the
functionalized gate. Binding of H5N1 to the
antibodies causes a change in the electrical
properties, measured as a variation in the current
using a potentiostat.
The QCM sensor uses a quartz crystal coated
with antibodies specific to HA and NA proteins
of H5N1. Saliva samples are applied to the
sensor, where the binding of H5N1 causes a
measurable change in the crystal's frequency.
The mass change is measured as a frequency
shift using a frequency counter.
Simulation
Parameters
Surface tension: 0.072 N/m - Pore radius:
1×10−6m - Viscosity: 1.5×10−3 Pa.s -
Association rate constant (ka): Specific to
H5N1 antibodies - Dissociation rate
constant (kd): Specific to H5N1 antibodies
Drain-source voltage 0.1 V - graphene channel
width and length: 10-7 and 10−6 meters - graphene
carrier mobility 5000 cm²/V·s - rate constant ()
108 M-1s-1 (for viral particle binding to antibodies) -
Viral particle concentrations 10−12M (104
particles/mL) and 10−9 M for high concentration
(107 particles/mL).
Fundamental frequency (f0): 5 or 10 MHz -
Gold electrode area: 0.2 cm² - Quartz density
(ρ): 2.65 g/cm³ - Shear modulus (μ): 2.95×1011
/cm·s² - Low concentration: 104 particles/mL -
High concentration: 107 particles/mL
Table 2: Performance summary of three sensors in detecting particle concentrations of 104 and 107 particles/ml
Sensor
Parameter Measured
Response at 104
particles/mL
Response at 107
particles/mL
Sensitivity
Dynamic Range
Lateral Flow
Optical Density (O.D.)
0.1 a.u.
0.6-0.7 a.u.
Moderate (~0.5-0.6 a.u.
difference)
Effective for 104 to 107
particles/mL
Graphene
FET
Change of Drain
Current (ΔD)
24 nA
160 nA
High
(136 nA difference)
Broad; effective from <103 to 107
particles/mL
QCM
Frequency Shift (Δf)
~0 Hz
-300 Hz
High at 107 (300 Hz shift);
low at 104
Effective for 102 to 107
particles/mL
Table 3: Comparison of COVID-19 and H5N1 LFT
Feature
SARS-CoV-2 Lateral Flow Biosensor
H5N1 Lateral Flow Biosensor
Target Virus
SARS-CoV-2
H5N1
Sample Type
Saliva/Nasal Swab
Saliva
Conjugate Pad
Antibodies conjugated to gold nanoparticles for SARS-CoV-2
Antibodies conjugated to gold nanoparticles for H5N1
Test Line
Immobilized antibodies specific to SARS-CoV-2 antigen
Immobilized antibodies specific to H5N1 antigen

ISSN 0973-2063 (online) 0973-8894 (print)
©Biomedical Informatics (2024)
Bioinformation 20(11): 1516-1523 (2024)
1522
Control Line
Anti-species antibodies
Anti-species antibodies
Visualization
Color change (gold nanoparticle aggregation)
Color change (gold nanoparticle aggregation)
Sensitivity
High (LOD ~ 100 particles/mL)
High (LOD ~ 100 particles/mL
Specificity
High, with cross-reactivity tests for other coronaviruses
High, with cross-reactivity tests for other influenzas
Response Time
15-30 minutes
15-30 minutes
Conclusions:
The present study shows that there is great potential for
improving pandemic preparedness through the development
and assessment of quick H5N1 virus detection techniques
utilizing biosensors as preventive measures. Our research
concentrates on three different biosensors, each with their own
advantages and critical issues: LFT, QCM sensors and FET based
electrochemical sensors. These biosensors possess distinctive
features and prospective uses; their integration into public
health initiatives could significantly enhance our ability to
respond promptly to emerging viral hazards. The CoViD-19
pandemic has provided important insights into the need for
accessibility, flexibility and scalability of diagnostic tools. By
creating and testing these biosensors in time, we can enhance
both early detection and prompt reactions, which can ultimately
lessen the negative impact outbreaks and even prevent
pandemics. The analysis we provide presents a path for the
creation of cutting-edge diagnostic tools ready to uphold the
protection of the world's health. A comprehensive and reliable
detection system for H5N1 and other emerging infections can be
built by utilizing the unique advantages that each type of
biosensor has to offer. The timely need for further research and
development into this field is crucial for attaining a state of
preparedness and resilience against impending public health
emergencies.
Conflicts of interest: None
Ethics approval: Not required.
References:
[1] Bonilla-Aldana DK et al. New Microbes New Infect. 2024
60:101439 [PMID:38911488]
[2] Olaimat AN et al. Life (Basel). 2024 14:190 [PMID:38398699]
[3] Mallapaty S. Nature. 2024 [PMID:38997549]
[4] Qasem A et al. Curr Issues Mol Biol. 2021 43:728
[PMID:34287238]
[5] Toubanaki DK et al. Sci Rep. 2020 10:6529 [PMID:32300204]
[6] Qiu W et al. Biosens Bioelectron. 2015 64:367 [PMID:25262062]
[7] Sreejith S et al. Measurement. 2023 206:112202.
[DOI:10.1016/j.measurement.2022.112202]
[8] Abad JM et al. Anal Chimica Acta. 1998 368:183 [
DOI:10.1016/S0003-2670(98)00205-0]
[9] Shan W et al. Talanta. 2014 126:130 [PMID:24881543]
[10] Zhang JZ et al. Nat Biotechnol. 2022 40:1336 [PMID:35484405]
[11] Grant BD et al. Anal Chem. 2020 92:11305 [PMID:32605363]
[12] Seo G et al. ACS Nano. 2020 14:5135 [PMID:32293168]
[13] Fu X et al. Int J Mol Sci. 2023 24:17157 [PMID:38138987]
[14] O’Sullivan CK & Guilbault GG. Biosensors Bioelectronics.
1999 14:663.[DOI:10.1016/S0956-5663(99)00040-8]
[15] To K et al. Lancet Infect Dis. 2020 20:565 [PMID:32213337]
[16] Bhatta A et al. medRxiv. 2021
DOI:10.1101/2021.02.12.21251229[ ]
[17] Rowe T et al. J Clin Microbiol. 1999 37:937 [PMID: 10074505]
[18] Kim S et al. J Korean Med Sci. 2020 35:e287 [PMID: 32776725]
[19] Narayanasamy S et al. Clin Infect Dis. 2023 77:1635 [PMID:
37435958]
Appendix 1:
The migration of the sample through the membrane in lateral flow tests can be described using the Lucas-Washburn equation, which models
capillary-driven flow in porous media. The Lucas-Washburn equation is given by:
L (t) = 

Where L (t) is the distance traveled by the fluid in the membrane at time t, g is the surface tension of the fluid, R is the pore radius of the
membrane, and η is the viscosity of the fluid. In this study, we used this equation to model the migration of saliva samples containing H5N1
virus particles through the membrane. The parameters were chosen to approximate the properties of saliva and the materials used in the
lateral flow tests. Specifically, we used a surface tension of 0.072 N/m, a pore radius of 1×10−6 m, and a viscosity of 1.5×10−3 Pa.s.
Additionally, the interaction between the H5N1 virus particles and the antibodies in the conjugate zone was modeled using the Langmuir
adsorption model. The interaction between the H5N1 virus particles and the antibodies in the same zone was modeled using the Langmuir
adsorption model. This model describes the binding kinetics of antigens to antibodies and is given by
󰇟󰇠
󰇟󰇠󰇟󰇠󰇟󰇠
Where [A] is the concentration of H5N1 particles, [B] is the concentration of unbound antibodies, [AB] is the concentration of H5N1-antibody
complexes, ka is the association rate constant, kd is the dissociation rate constant. At equilibrium, the fraction of occupied binding sites  can
be described by the Langmuir isotherm equation:

ISSN 0973-2063 (online) 0973-8894 (print)
©Biomedical Informatics (2024)
Bioinformation 20(11): 1516-1523 (2024)
1523
󰇟󰇠
󰇟󰇠
Where Kd is the dissociation constant, defined by


The concentrations of H5N1 particles ranged from 104 to 107 particles/mL, the antibody concentration [B] was selected to the value of 5×10−2
M to ensure detection.
Appendix 2:
Calculations and model features used in our study on the graphene-based FET/MOSFET biosensor for H5N1 detection in saliva samples are
reported. Constants and parameters, derivation of equations for the change in charge density, the effective mobility, the maximum current
changes, and the time-dependent change in current are considered. We also discuss the application of the Langmuir adsorption model for
modeling the binding kinetics of viral particles to antibodies on the graphene surface.
Constants and parameters:
The graphene-based MOSFET biosensor operates with several key parameters. The drain-source voltage () is set at 0.1 V. The width ()
and length () of the graphene channel are estimated 10 10−6 meters and 10−6 meters, respectively. The carrier mobility () of the graphene is
5000 cm²/Vs. To account for the realistic conditions at high viral concentrations, a reduction factor of 0.1 is applied to the mobility. The
baseline current due to antibodies only (b) is simulated at 9.6 nA. The association rate constant () for the binding of viral particles to
antibodies is 108 M-1s-1. The viral particle concentrations considered are 10−12M for low concentration (equivalent to104 particles/mL) and 10−9
M for high concentration (equivalent to 107 particles/mL).
Change in charge density:
The binding of viral particles to the graphene surface results in a change in charge density (Δ). This change depends on the concentration of
viral particles and the effective charge each particle contributes. For low and high concentrations, the change in charge density is calculated
as follows:
[1] For low concentration: Δ=1.60×10−9 C/cm2;
[2] For high concentration: Δ=1.60×10−6 C/cm2
Maximum current changes:
The maximum change in drain current due to the binding of viral particles is directly proportional to the drain-source voltage, the change in
charge density, and the aspect ratio of the graphene channel. For low and high concentrations, the maximum current changes are:
[1] For low concentration: Δmax, low=×Δlow×(/)=0.1×1.60×10−9×10×109=16 nA
[2] For high concentration: Δmax, high=×Δhigh×(/)×=0.1×1.60×10−6×10×109×0.1= 160 nA
Time-dependent change in current:
To model the time-dependent change in current, we use the kinetics of the binding process, described by the Langmuir adsorption model.
The change in current (Δ () over time due to the binding of viral particles is expressed as:
Δ() = Δmax x (1−exp(− [] )).
Including the baseline current, the total current at any time  is given by:
[1] For low concentration: low()=16 nA×(1−exp(−0.1))+9.6 nA
[2] For high concentration: high()=160 nA×(1−exp(−100))+9.6 nA
Langmuir adsorption kinetics:
The binding kinetics of viral particles to the antibodies on the graphene surface is described using the Langmuir adsorption model. This
model, as in the Appendix 1, takes into account the rate of association and dissociation of the viral particles with the binding sites and the
fraction of occupied binding sites (θ) at equilibrium. Given the reduced dimensions of the present devices, these effects are less relevant in the
present simulation with respect to the electrical properties.