MBE, ():

DOI:
Received: November 2023
Revised:
Accepted:
http://www.aimspress.com/journal/mbe Published:

1 Type of article

2 A multi-scale approach driven by population demographic and

3 socio-cultural data to simulate setting-specific respiratory infectious disease
4 transmission patterns: An application to COVID-19 in a shopping center

5 Dramane Sam Idris Kanté1, * , and Aissam Jebrane2

1
6 LAMAI, Department of Mathematics, Faculty of Sciences and Technologies, Cadi Ayyad
7 University, Marrakesh 40140, Morocco
2
8 Complex Systems and Interactions Research Center, Ecole Centrale Casablanca, Ville Verte,
9 Bouskoura 27182, Morocco

10 * Correspondence: [email protected]

Abstract: We suggest a multi-scale approach for estimating the risk of contracting a respiratory dis-
ease in a social context. The method takes both direct and indirect modes of transmission into account.
A social force model that is guided by population demographic and socio-cultural data particular to the
social setting, such as desired speeds, interpersonal distances, visit duration, and population density,
describes interactions between populations. The spread of covid-19 among customers at a Moroccan
shopping center is examined as a case study. Results of conducted simulations show that visitors be-
tween the ages of 10 and 15 have the highest risk of contracting an infection due to their high levels of
activity and frequent contact with people of different ages. People over 40, who move slowly and spend
11
a lot of time with their neighbors, are next. We estimated the probability of each age group transmitting
the disease. Findings indicate that agents older than 40 may be superspreaders. The assessment of the
effect of the virus deposition range on the risk of indirect infection revealed that an expansion would
induce growths in the risk of indirect infection that are over 90 %. This study also demonstrates that
social isolation reduces the risk of direct infection while maintaining the risk of indirect transmission.
The current study provides the path for the estimation of infection risk matrices and for a deeper com-
prehension of the key drivers of the Covid-19 epidemic’s transmission dynamics. Furthermore, it can
be used to design effective interventions to thwart epidemic spread within each social context.

12 Keywords: Crowd dynamics; Social mixing patterns; COVID-19; multi-scale model;

13 setting-specific contact matrices; infection risk; socio-cultural data

14
 2

1 1. Introduction

2 Understanding the patterns that underlie the transmission of respiratory infectious disease, par-
3 ticularly indoors or in a moving and interacting population, is critical to forecasting these illnesses.
4 Changes in these patterns are brought about by the habits and socio-cultural behaviours of the popula-
5 tion under study, in addition to the demographic structure or health status of the population. Facilities
6 frequented by a large number of people including schools, public transport, health services and shop-
7 ping centers are considered to be the most vulnerable to the spread of infectious respiratory diseases
8 such as COVID-19. The social interactions and behaviors of agents in these social contexts affect the
9 dynamics of virus infections [1, 2], which change from one place of activity to another.
10 The most common method of capturing social interactions is the so-called “contact matrix” which
11 estimates the rate of contact between different age groups visiting a place of activity [3]. This method
12 relies on questionnaires where two participants are considered to be in contact if they initiate a con-
13 versation that lasts longer than three words. The contact matrices of this study were projected in 152
14 countries [4] and recently extended to 177 geographical regions for the COVID-19 era [5]. These
15 matrices are integrated into epidemiological models such as [6, 7, 8] to forecast disease spread like
16 COVID-19. Although these setting-specific contact matrices reveal a particular interest in the math-
17 ematical modeling of infectious diseases, questionnaire-based methods do not account for temporal
18 changes in the dynamics of social contact patterns [9], which is required for anticipating the evolution
19 of an epidemic and its seasonality[10]. Furthermore, it is difficult to capture the effects of situations
20 such as rush hours, or physical distancing using these methods. In addition, questionnaires are ex-
21 pensive and impossible to regularly conduct to accurately track contacts. Another class of approaches
22 consisting of coupling surveys and numerical simulations has been developed by researchers for es-
23 timating contact matrices. Authors in [11] use an agent-based model informed by the schedule of
24 Italians and socio-demographic data. The model simulates the interactions of the Italian population
25 as a synthetic society for 24 hours in 144 time slots of 10 minutes each. Agent interactions for each
26 time slot are captured via survey and two persons are in contact when they share the same place (bus,
27 office, etc.). A similar approach has been also adopted in [12] to quantify the mixing patterns of 26
28 European countries. In this approach’s formalism, it is challenging to capture the dynamic of contacts,
29 notwithstanding the work’s contributions to the improvement of reflections on the estimate of contact
30 matrices.
31 To address this limitation, a social force model for crowd dynamics improved from [13, 14] is used
32 for modeling indoor transmission risks of diseases during people’s social activities [15]. In addition to
33 estimating contact matrices, this method may also look at how non-pharmaceutical interventions affect
34 contact patterns. Afterwards, authors in [16] developed a multi-scale model to simulate the dynamics
35 of Covid-19 transmission in a synthetic population. Both direct and indirect disease transmission
36 modes are considered via a spatio-temporal model. Population interactions are captured by a modified
37 social forces model [17] that incorporates socio-cultural and demographic data about the community
38 under investigation. This approach has been implemented with Moroccan and Vietnamese populations
39 [18, 19]. The approach allows for the accurate capture of age-structured contact matrices and infection
40 risk of a virtual population as well as the evaluation of non-pharmaceutical intervention measures.
41 Social networks and graph models are other methods that help integrate interactions between agents.
42 Researchers in [20] developed an individual-based model that uses close contact tracing to study the

Mathematical Biosciences and Engineering Volume , Issue ,

 3

1 effect of non-pharmaceutical measures. A similar approach is used in [21]. A graph model that allows
2 for the integration of social distance and contact duration was proposed in [22]. However, researchers
3 stated that they were not able to consider factors such as demography and population density that affect
4 the number of cases of an infection.
5 In epidemiological models, the risk of viral infection or the so-called probability of virus transmis-
6 sion from an infectious to a susceptible agent is defined by the product of two parameters : i) the contact
7 rate estimated via contact matrices, and ii) the contact effectiveness or the probability of transmission
8 on contact which must also be estimated separately [8]. It seems more appropriate to estimate infection
9 risk matrices that account for disease characteristics for example the modes of transmission, and pop-
10 ulation interactions rather than contact matrices that refer only to the agent’s displacement. Thus, this
11 work deals with the study of patterns underlying transmission in a virtual Moroccan shopping center
12 using an improved multi-scale approach [23] for Covid-19 transmission. A shopping center is one of
13 the places of activity in society [11], and the growing interest in this location in the Moroccan society
14 due to the emergence of mall culture [24], makes it an interesting case study. Indeed, after collecting
15 age structures, socio-cultural and demographic data on the Moroccan mall-visiting population, such as
16 desired social distances [25, 26], desired walking speeds [27], the duration and frequency of mall visits
17 [28, 29, 30]. These are integrated into the social forces model, while virus production and transmission
18 are captured by a diffusion reaction equation.
19 The organization of this manuscript is as follows: section 2 concerns the presentation of the setting-
20 specific multi-scale approach to infectious respiratory disease transmission. Section 3 focuses on socio-
21 cultural data. Section 4 concerns the numerical simulations with an application to Covid-19 disease in
22 a Moroccan shopping Center.

23 2. Materials and methods

24 In this section, we will discuss the method for infectious respiratory diseases transmission that we
25 have selected. This method is based on a multi-scale model of COVID-19 dynamics that we have
26 designed recently [16]. The multi-scale model considers both direct and indirect transmission. This
27 approach relies on a virus concentration equation and a modified social force model that describes
28 agent displacement.

29 2.1. Agent’s displacement model

30 The social force model is a well calibrated and often utilized framework to understand human
31 displacement in various contexts [13, 31, 32]. Previous models are not well suited to situations where
32 agents move in different directions and their paths cross or are opposite [33, 34]. Pedestrians cannot
33 avoid multiple pedestrians simultaneously with these methods especially in dense crowds [35]. [36]
34 modified the repulsive forces to account for agents’ personal spaces and long-range interactions. This
35 modified social forces model that was employed is briefly described below. We consider a population
36 of n agents where the i—th agent is assimilated to a disk of center xi , a radius ri and mass mi . The
37 equation governing the at-a-distance interactions of the i–th agent with its environment is the following:
38
dxi
= vi
(
dt (2.1)
mi dv
dt
i
= fisel f + fisoc + fiobs ,

Mathematical Biosciences and Engineering Volume , Issue ,

 4

Figure 1. Schematic representation of the model parameters defining the interactions be-
tween two agents.

1 where, vi is his velocity, fisel f is the self-driven force which allows the agent to regulate his actual
2 velocity via his desired velocity vd,i . The self-driven force is given as follows:
vd,i − vi
fisel f = mi , (2.2)
τi
3 with, vd,i = vd,i ed,i , vd,i (resp ed,i ) is the desired speed (resp desired direction) of the i–th agent and τi is
4 his relaxation time.
5 Agent-agent interactions are described by a repulsive force or the so-called socio-psychological force
6 given by:
X
fisoc = j ,
fisoc (2.3)
7 where fisoc
j is the social psychological force between the i–th and j–th agents, described as follows:

1+cos ϕ
  d j −dsoc   
 A soc exp iβsoc
 γ + (1 − γ) 2 i j ei j , i f di j < d soc
fi j =  ,
soc

(2.4)
 0,
 elsewhere
8 Here, A soc represents the amplitude of the social psychological force, di j is the distance between the
9 i–th and the j–th agent, d soc is the desired distance which describes the distance that agents tend to
10 keep between them, β soc is the falloff length of the social psychological force, γ describes the effect of
11 interactions taking place behind the agent, while ϕi j represents the angle between the desired velocity
12 and the actual one.
obs
13 The repulsive force between agents and walls fiw allowing to avoid collisions is similar to fisoc
j
14 Figure 1. More details on the parameters of the model are provided in [37].

Mathematical Biosciences and Engineering Volume , Issue ,

 5

1 2.2. Transmission dynamic model

2 An infectious agent can directly transmit the virus to a susceptible agent by direct contact, ie, if
3 the two remain close enough for a given duration. The probability of direct infection Pd follows a
4 Bernouilli law. To consider the indirect transmission, we describe the spatio-temporal dynamics of the
5 virus concentration C in a social context by the following equation:

∂C
(x, t) = D∆C(x, t) + W(x) − σC(x, t) (2.5)
∂t

6 where, D is the diffusion coefficient, σ is the decay rate of the virus on surfaces and W refers to virus
7 produced by infected agents on surfaces. When an infectious agent coughs, they secrete droplets in an
8 area of radius d0 . Thus we define W using a radial-basis function [16]:

 4 1 + cos πd(x)
 w  
, d ≤ d0
W(x) = 

d0 (2.6)
 0, d ≥ d0

9 where d(x) is the distance between the position of an infectious coughing agent and the surrounding
10 position on the mesh and w is the virus production rate by the coughing agent. The probability of
11 getting infected from a surface is Pind = λC(x), where λ is the index of indirect contamination per day.

12 2.3. Aggregate socio-cultural and demographic data

13 We first gathered socio-cultural data relevant to the studied social context (in this case, a Moroccan
14 shopping centre), such as the duration of the visit, the social distances, the desired speeds, the age
15 structure of the visitors, etc. This is because interactions between agents can vary from one social
16 context to another as well as from one age group to another.
17 After that, we divided the population into 16 groups of five-year ages ( 0–5, 5–10,..., 75–80 )
18 and identified the parameters that were age-dependent or not. The values of the age-dependent and
19 age-independent parameters were then presented in Tables 2 and 3. The social forces model finally
20 incorporates these data.
21 The desired interpersonal distances or so-called social distances depend on the social context as
22 well as the age of the agents. In [25], authors estimated four interpersonal distances that depend on the
23 relationship. We compute personal distances estimated per age group in [26, 25]. The desired speed
24 also depends on an agent’s age and gender, as well as the location where they walk. In this study
25 we use estimates provided in [27, 38]. The relaxation time depends on the age and was calibrated in
26 previous studies [39, 40]. The percentage of shopping center visitors by age group was estimated in
27 [30, 41, 29]. Another characteristic of the population visiting a shopping center is the weight which
28 also varies according to age [42, 43, 44, 45]. We summarize the sources of each parameter and their
29 dependency on age in Table 1, their values are given in Table 2.
30 The parameters intrinsic to the social force model, such as the amplitudes, the falloff lengths of
31 the social psychological forces are calibrated in several studies [46, 31, 32]. The visit duration was
32 estimated in [28, 29]. We retain the corresponding values in a normal situation (without panic) Table
33 3.

Mathematical Biosciences and Engineering Volume , Issue ,

 6

Table 1. The model parameters dependency on age and their sources.

Parameter Age- source
dependent
vd,i yes adapted to the demographic structure [27]
τi yes [39, 40]
d soc yes adapted to the demographic structure [26]
dobs no [36]
A soc no [17, 32, 31]
Aobs no [17, 32, 31]
βobs no [17, 32, 31]
β soc no [17, 32, 31]
γ no [17]
mi yes [42, 44, 45]
visit duration no [28, 29]

Table 2. Values of data that are dependent on the age in a Moroccan shopping center.
Age bracket Demography (%) Weight (kg) Desired speed (m/s) τi (s) Desired distance (m)
[0,5) 0.0 N(9.79; 6.1) N(0.65; 0.292)
N(1.26; 0.125)
[5,10) 0.6 N(21.129; 2.816) N(1.08; 0.292) 0.5
[10,15) 1 N(34.525; 6.553) N(1.33; 0.292)
N(1.31; 0.135)
[15,20) 12.75 N(53.456; 8.798) N(1.45; 0.323)
[20,25) 12.25 N(58.08; 13.1) N(1.44; 0.323)
N(1.35; 0.141)
[25,30) 12.50 N(64.32; 13.1) N(1.39; 0.246) 0.54
[30,35) 12.04 N(68.4; 10.30) N(1.36; 0.246)
N(1.38; 0.108)
[35,40) 11.42 N(71.5; 10.30) N(1.33; 0.246)
[40,45) 10.03 N(59.1; 10.80) N(1.3; 0.246)
N(1.36; 0.128)
[45,50) 6.05 N(65.3; 10.80) N(1.26; 0.2246)
[50,55) 5.43 N(58.99; 10.13) N(1.22; 0.243)
N(1.22; 0.015)
[55,60) 5.11 N(60.36; 10.03) N(1.2; 0.243)
0.71
[60,65) 4.49 N(72.1; 12.02) N(1.2; 0.243)
N(1.37; 0.129)
[65,70) 3.44 N(71.995; 13.63) N(1.18; 0.255)
[70,75) 2.08 N(67.005; 15.50) N(1.08; 0.255)
N(1.49; 0.1309)
[75,80) 1.35 N(70.39; 16.71) N(0.92; 0.255)

Table 3. Social force model parameters’ values that are independent of age in a Moroccan
shopping center.
Parameter dobs A soc Aobs βobs β soc
visit γ
duration
Value 1 m 700 N 1000 N 0.8 m 0.8 m [0,1] 1h30

Mathematical Biosciences and Engineering Volume , Issue ,

 7

1 2.4. Numerical simulations

2 The computational shopping center Figure 2 has a size 200 m × 200 m, and it accommodates 1000
3 agents. At the beginning of the simulation, the positions are random, and the initial velocities are
4 set to zero. The initial directions are random, and 30% of the agents change their routes every 10
5 seconds. 150 simulations are performed for each study. We will simulate the spread of COVID-19 in a
6 virtual shopping center using the framework described before. We apply Dirichlet boundary conditions
7 C = 0 on the walls of the shopping center. Our simulations include two classes of people, symptomatic
8 infectious individuals and susceptible individuals. We estimate the risk of being contaminated by
9 an infectious individual in a social context where time is reduced. The simulations last 10 minutes
10 and individuals do not transition from one class to another. The risk of direct infection Pd follows a
11 Bernoulli law. The radius of a contaminated surface is d0 , and the virus production rate of a coughing
12 agent is w. The index of daily indirect contamination is λ, and the virus decay rate is σ. COVID-
13 19 aerosol diffusivity is D. Table 4 lists the values for these parameters. Three infectious persons
14 are chosen among the agents at the beginning of the simulations despite their age, socio-cultural and
15 demographic characteristics [47, 48]. Finally, we assess the risk of contamination in each age group.
16 We use a Runge Kutta method of order 4 to solve the modified social force model and a finite difference
17 method to solve the diffusion reaction equation.

Figure 2. Snapshot of the computational shopping center

Table 4. Epidemiological parameters of the model.

Parameter Value Source
Pd 0.04 [16]
d0 1m [49]
w 0.00639 virus h−1 [50]
σ 0.042 h−1 [51]
λ 0.015 [16]
D 0.001m2 /s [52]

Mathematical Biosciences and Engineering Volume , Issue ,

 8

1 3. Results

2 3.1. Risk of infection in the Moroccan shopping center

3 Using data in Tables 2, 3 and 4, we ran simulations 150 times with three different infectious agents at
4 different initial positions each time. We estimate the risk of infection P = Pd + Pind in each age group.
5 Results in Figure 3 show that people between 10–15 years have a higher risk of being contaminated.
6 They have the smallest average desired distance 1.31 m and a small desired velocity 1.33 m/s which
7 increases their exposure period to the virus when they establish contact with an infectious agent or
8 a contaminated surface. They are followed by age groups 70–75 and 75–80 who have the smallest
9 velocities in the simulations. With the exception of age group 10–15, young age groups have a low
10 risk of contracting the disease because of their high velocities that decrease their exposure period to
11 the virus.

30
mean risk of infection
95% confidence interval
25

20
P (%)

15

10

0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
age (years)
Figure 3. Risk of infection per age group in the shopping center with the data presented in
Table 1.

12 3.2. Social distancing reduces the risk of direct infection

13 To determine whether social distance affects the risk of direct infection, we consider parameters as
14 presented in Table 2, and we vary interpersonal distances from 0.3 to 1.70 m. Results in Figure 4 A)
15 show a decrease in the risk of direct infection when we increase the desired distance. However, the
16 indirect infection risk is not correlated with social distancing Figure 4 B).

Mathematical Biosciences and Engineering Volume , Issue ,

 9

(A) (B)
80 74.5

70 74

60 73.5

50 73

Pind (%)
Pd (%)

40 72.5

30 72

20 71.5

10 71

0 70.5
0 0.5 1 1.5 2 0 0.5 1 1.5 2
dsoc (m) dsoc (m)

Figure 4. Effect of social distancing on the risk of infection of the population frequenting
Moroccan shopping centers. A) the risk of direct infection Pd . (B) the risk of indirect infec-
tion Pind .

1 3.3. The reduction in contact duration and surface disinfection reduce the risk of infection

2 First, we investigate the effect of weight and desired velocity on the risk of infection. We set the
3 desired distance to 1.22 m [25] and the desired velocity to 1.34 m/s [13], then we alter the agents’
4 weights from 45 to 178 kg. Simulations show that when you gain weight, the probability of infection
5 rises Figure 5 A). The social force model can be used to explain this by stating that as mass increases,
6 the magnitude of the repulsive force (or socio-psychological force) weakens and agents are no longer
7 able to maintain a suitable social distance from one another. This increases the risk of infection. Figure
8 5 B) depicts the impact of the desired velocity which varies from 0.5 to 3.3 m/s on the risk of infection.
9 Results show that when the desired velocity increases the risk of infection decreases. This occurs
10 because the agent is exposed to the virus for a shorter time when the desired velocity (and therefore the
11 actual velocity) is high. Indeed, it shortens the time spent in direct and indirect contact with infectious
12 agents and contaminated surfaces.

13 Second, the assessment of the virus decay rate reveals that public spaces should also be disinfected
14 to avoid contamination Figure 6.

Mathematical Biosciences and Engineering Volume , Issue ,

 10

(A) (B)
60 80

70
50

60

40
50
P (%)

30 40

30
20

20

10
10

0 0
50 100 150 200 0 1 2 3

weight (kg) vd,i (m/s)

Figure 5. Impact of the weight and the desired velocity on the risk of infection. (A) Estimate
of the impact of the weight on the risk of infection. All parameters are homogenized the
desired distance d soc = 1.22 m, the desired velocity vd,i = 1.34 m/s and we vary the weight
m. (B) Impact of an increase in the desired velocity. We consider parameters in Table 1, the
desired velocity is homogeneous in each simulation and we vary it from 0.5 to 3.3 m/s.

Mathematical Biosciences and Engineering Volume , Issue ,

 11

40

35

30
Pind (%)

25

20

15

10
0 0.5 1 1.5 2 2.5 3
-1 -3
(s ) 10

Figure 6. Effect of the virus decay rate on the risk of indirect infection.

1 3.4. People are more likely to be infected by old persons over 40 years old

2 Several studies analyzed the crucial traits of COVID-19 superspreaders. Some of them search for
3 genomic characteristics, whereas others demonstrated that advanced age may be a representative trait
4 [53]. With information from Tables 2, 3 and 4, we look at how likely it is for each age group in the
5 mall to transmit the infection. We chose 1 infectious agent at random from each of the 16 age groups
6 out of 1000 total agents. The simulation is run 150 times, and the findings are depicted in Figure7.
7 Results indicate that those between the ages of 55 and 60 are more likely to spread the disease to other
8 age groups. This result can be explained by their low velocity of 1.2 m/s and short desired distance of
9 1.22 m. Considering the population visiting the shopping center, old people over 40 years represent
10 the greatest threat to the health of a susceptible person compared to young people under 40 years old.

Mathematical Biosciences and Engineering Volume , Issue ,

 12

16

14

12
mean lhd
10 95 % confidence interval
lhd (%)

0
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80
age (years)
Figure 7. Probability of transmitting the disease. We estimate the likelihood ( lhd ) as the
number of time the age group has been the principal threat of infection to a susceptible person
(stocked in a matrix) divided by the number of susceptible person. The findings suggest old
people are most likely to infect others.

1 3.5. Impact of droplet travel distance on the indirect contamination

2 It was reported in [52] that aerosol particles could travel over 30m. In another study [54] they
3 investigated the aerodynamic nature of COVID-19 and reported aerosol deposition in different corners
4 inside a room of the intensive care unit of Renmin hospital in China more than 2m away from a patient’s
5 bed. They suggested that virus-laden aerosol deposition may have a role in surface contamination and
6 subsequent contact with susceptible people resulting in infection. In this section we study the impact
7 of particle deposits’ range on the risk of indirect transmission. To model such effects, we vary the
8 parameter d0 in Eq 2.6 which defines the contaminated surface. In the simulations we only account for
9 indirect transmission. Parameters for visitors’ displacement are presented in Tables 2, 3. We estimate
10 the mean risk of indirect infection for each age group Figure 8 A) and the average risk Figure 8 B)
11 for three values of d0 0.5 m, 1.5 m and 2 m. Results indicate a drop in indirect transmission risk of
12 58.21% when particles can reach surfaces that are 1.5 m away and a decrease of 80.11% when they
13 can only reach distances of 0.5 m, compared to the case where particles reach surfaces 2 m away from
14 an infectious individual.

15 4. Discussion and conclusions

16 Forecasting the spread of respiratory infectious diseases, particularly indoors or in a moving and
17 interacting population, requires an understanding of the mechanisms underlying their transmission.
18 Habits, social contacts, and the population’s demographic makeup all have an impact on the propa-

Mathematical Biosciences and Engineering Volume , Issue ,

 13

(A) (B)
100 35
d0 =0.5 m
90 d0 =1.5 m
30
d0 =2 m
80

70 25

60

Pind (%)
Pind (%)

20

50

15
40

30
10

20
5
10

0 0
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 0.5 1.5 2
age (years) d0 (m)

Figure 8. Impact of the particles travel distance.

(A) The risk of infection per age group for d0 = 2 m; d0 = 1 m; and d0 = 0.5 m. (B) the
average risk of indirect transmission for the values of d0 and the confidence intervals

1 gation of the virus. In order to solve these dependencies in a virtual mall in Morocco, we developed
2 a hybrid multi-scale model. The social force model, which is calibrated and takes into consideration
3 various psychological concepts particular to human interactions, is part of the multi-scale framework.
4 In order to recreate their movement, we acquired socio-cultural and demographic information on the
5 Moroccan population visiting the shopping center by age group. We examined the impact of social
6 isolation on the risk of infection and determined which age groups are most likely to transmit the
7 infection. Each age group’s infection risk was also calculated. The model enabled us to analyse the
8 effect of the viral particle deposit range on the risk of infection.
9 We found that people in the age group 10–15 had the greatest risk of getting infected, followed by
10 people aged 70–75 and 75–80. The age group 10–15 has the smallest desired distance 1.31 m in the
11 population and a small desired velocity 1.33 m/s. Old people have in general small desired velocities
12 which increase their risk of infection. People between 15 and 40 have high desired velocities coupled
13 with high desired distances, which reduces their exposure time to infectious agents. Researchers in
14 [3] estimated high numbers of contacts for adolescents and children. They simulated the initial phase
15 of a hypothetical respiratory infection epidemic. Their results suggest that young people between 5
16 and 19 years old will suffer the highest burden of respiratory infection during an initial spread due to
17 their number of contacts and tendency to interact within their age group. However the incidence of
18 COVID-19 cases in this age group is low and most cases are asymptomatic, suggesting that they are
19 less susceptible to the disease [55]. The present model still has room for improvement in this regard as
20 the probabilities of infection could be ameliorated to account for the strength of the immune system.

Mathematical Biosciences and Engineering Volume , Issue ,

 14

1 Nevertheless, the model indicates that people over 40 years are most likely to be contaminated. This
2 is due to the demography of the virtual shopping center ( the risk of infection of the age group 5-10 is
3 not important due to his low frequency ). This result is supported by the incidence of cases ( the mean
4 age of cases is 50.15 ± 18.97 ) in March 2020 in Morocco [56].
5

6 It was reported in [57, 58] that being obese increases the risk of COVID-19 severity. Some
7 experimental studies showed that this severity could be related to the inhaled dose of virus [59, 60, 61].
8 The duration of exposure to the virus is a key factor in the risk of infection for both direct contact (with
9 the infectious agent) and indirect contact (with contaminated surfaces). According to the social force
10 model, overweight people move slowly and do not change their directions of movement regularly,
11 which explains why they are more exposed to the virus. Thus, this lead to a higher risk of infection for
12 overweight people in the model. Even if findings seem to accord with the studies discussed above, we
13 cannot affirm their results because our model does not describe the response of the immune system.
14 The results presented in this study support the idea that long periods of exposure to an infectious person
15 would favor the accumulation of the virus within a susceptible agent. Contact durations were estimated
16 in previous studies [3, 11]. They were used to evaluate the impact of different control strategies of the
17 COVID-19 virus in a hospital in South Korea [62], and fitted to serological data of other infections [11].
18

19 The COVID-19 disease is reminiscent of the severe acute respiratory syndrome (SARS) epidemic
20 of 2002 to 2004. Superspreaders greatly favored the spread of the virus (SARS-CoV-1) during that
21 period. Investigations revealed that they were elderly agents (over 40 years old) [63, 64]. Many ev-
22 idences suggest that older agents are also playing the role of superspreaders of the Covid-19 virus.
23 Indeed, authors in [65] showed advanced age is a possible characteristic of COVID-19 superspreaders.
24 Furthermore, researchers in [66] performed a meta-analysis of the published literature on household
25 Covid-19 transmission clusters from 12 countries including Morocco. Agents over 40 years old repre-
26 sent 68 % of the index cases, and the mean age is 56 years old. Our approach enabled us to estimate
27 the likelihood of being infected by an infectious agent in each age group. The obtained results are con-
28 sistent with data in the literature [66]. They suggest that agents older than 40 are most likely to infect
29 a susceptible person. This finding also opens the door to the estimation of risk of infection matrices in
30 which a cell contains an estimate of a population’s risk of infection by one age group.
31 Indirect contamination of COVID-19 is neglected [67], [54] reported aerosol particles deposition
32 on surfaces over 2 m in Renmin hospital in China. They suggested that virus-laden aerosol deposition
33 may have a role in surface contamination and subsequent contact with susceptible people resulting
34 in infection. The results obtained in this study suggest an increase in the risk of infection that is
35 greater than 90 % for deposit ranges over 0.5 m in this study. Hence wearing of face masks and social
36 distancing are not enough, the disinfection of surfaces would also help tame the disease. The model
37 provides a tool for the study of the effect of ventilation on particles long range travel and deposition on
38 surfaces, therefore its effect on the ensuing risk of indirect contamination.
39 We demonstrated that social distancing effectively lowers the risk of infection. We also proved that
40 reducing contact duration will lower the risk of infection. Furthermore, we revealed that seniors over
41 40 are potential superspreaders.
42 To improve simulations in the virtual shopping center, one perspective of this study is to integrate
43 some features of a moving and interacting crowd such as waiting pedestrians [68] and group behavior

Mathematical Biosciences and Engineering Volume , Issue ,

 15

1 [69]. Another perspective is the modification of the formulas of the risks of infection to account for
2 both airborne and indirect transmission [70]. The concentration of the virus could also depend on
3 the viral load of an infectious person [71, 72]. Though, the purpose of our research is to propose an
4 idea for developing a method to better understand patterns underlying transmission and incorporating
5 them into epidemiological models. This research presents a setting-specific conceptual framework
6 for simulating infection risk in the study population. In order to recreate the mapping of the entire
7 community, one perspective on this study is to expand it to other social situations, such as schools,
8 residential neighbourhoods, etc. The framework provided in this work can be used to comprehend the
9 microscopic mechanisms controlling transmission dynamics.

10 Use of AI tools declaration

11 The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

12 Conflict of interest

13 The authors declare there is no conflict of interest.

14 References

15 1. A. Mousa, P. Winskill, O. J. Watson, O. Ratmann, M. Monod, M. Ajelli et al., Social contact pat-
16 terns and implications for infectious disease transmission – a systematic review and meta-analysis
17 of contact surveys, eLife, 10 (2021), e70294, URL https://doi.org/10.7554/eLife.70294.
18 2. D. M. Feehan and A. S. Mahmud, Quantifying population contact patterns in the united states
19 during the covid-19 pandemic, Nature communications, 12 (2021), 893.
20 3. J. Mossong, N. Hens, M. Jit, P. Beutels, K. Auranen, R. Mikolajczyk et al., Social contacts and
21 mixing patterns relevant to the spread of infectious diseases, PLOS Medicine, 5 (2008), e74.
22 4. K. Prem, A. R. Cook and M. Jit, Projecting social contact matrices in 152 countries using contact
23 surveys and demographic data, PLOS Computational Biology, 13 (2017), 1–21, URL https:
24 //doi.org/10.1371/journal.pcbi.1005697.
25 5. K. Prem, K. v. Zandvoort, P. Klepac, R. M. Eggo, N. G. Davies, C. for the Mathematical
26 Modelling of Infectious Diseases COVID-19 Working Group, A. R. Cook and M. Jit, Pro-
27 jecting contact matrices in 177 geographical regions: An update and comparison with empiri-
28 cal data for the covid-19 era, PLOS Computational Biology, 17 (2021), 1–19, URL https:
29 //doi.org/10.1371/journal.pcbi.1009098.
30 6. M. Kimathi, S. Mwalili, V. Ojiambo and D. K. Gathungu, Age-structured model for covid-19:
31 Effectiveness of social distancing and contact reduction in kenya, Infectious Disease Modelling, 6
32 (2021), 15–23.
33 7. N. Parolini, L. Dede’, G. Ardenghi and A. Quarteroni, Modelling the covid-19 epidemic and
34 the vaccination campaign in italy by the suihter model, Infectious Disease Modelling, 7 (2022),
35 45–63.

Mathematical Biosciences and Engineering Volume , Issue ,

 16

1 8. R. Singh and R. Adhikari, Age-structured impact of social distancing on the covid-19 epidemic
2 in india, 2020, URL https://arxiv.org/abs/2003.12055.
3 9. Y. Huang, X. Cai, B. Zhang, G. Zhu, T. Liu, P. Guo et al., Spatiotemporal heterogeneity of social
4 contact patterns related to infectious diseases in the guangdong province, china, Scientific reports,
5 10 (2020), 6119.
6 10. A. M. Dorélien, A. Ramen, I. Swanson and R. Hill, Analyzing the demographic, spatial, and
7 temporal factors influencing social contact patterns in us and implications for infectious disease
8 spread, BMC infectious diseases, 21 (2021), 1–19.
9 11. F. Iozzi, F. Trusiano, M. Chinazzi, F. C. Billari, E. Zagheni, S. Merler et al., Little italy: An agent-
10 based approach to the estimation of contact patterns- fitting predicted matrices to serological data,
11 PLoS Computational Biology, 6 (2010), 1–10, URL https://doi.org/10.1371/journal.
12 pcbi.1001021.
13 12. L. Fumanelli, M. Ajelli, P. Manfredi, A. Vespignani and S. Merler, Inferring the structure of social
14 contacts from demographic data in the analysis of infectious diseases spread, PLoS Computational
15 Biology, 8 (2012), 1–10, URL https://doi.org/10.1371/journal.pcbi.1002673.
16 13. D. Helbing and P. Molnár, Social force model for pedestrian dynamics, Physical review E, 51
17 (1995), 4282–4286.
18 14. D. Helbing, I. Farkas, P. Molnar and T. Vicsek, Simulation of pedestrian crowds in normal and
19 evacuation situations, in Pedestrian and Evacuation Dynamics (eds. M. Schreckenberg and S. D.
20 Sharma), Springer, Berlin, Germany, 2002, 21–58.
21 15. Y. Xiao, M. Yang, Z. Zhu, H. Yang, L. Zhang and S. Ghader, Modeling indoor-level non-
22 pharmaceutical interventions during the covid-19 pandemic: A pedestrian dynamics-based mi-
23 croscopic simulation approach, Transport Policy, 109 (2021), 12–23.
24 16. A. Bouchnita and A. Jebrane, A hybrid multi-scale model of covid-19 transmission dynamics
25 to assess the potential of non-pharmaceutical interventions, Chaos Solitons Fractals, 138 (2020),
26 109941.
27 17. B. Kabalan, P. Argoul, A. Jebrane, G. Cumunel and S. Erlicher, A crowd movement model for
28 pedestrian flow through bottlenecks, Annals of Solid and Structural Mechanics, 8 (2016), 1–15.
29 18. A. Bouchnita and A. Jebrane, A multi-scale model quantifies the impact of limited movement
30 of the population and mandatory wearing of face masks in containing the covid-19 epidemic in
31 morocco, Mathematical Modelling of Natural Phenomena, 15 (2020), 31.
32 19. A. Bouchnita, A. Chekroun and A. Jebrane, Mathematical modeling predicts that strict social
33 distancing measures would be needed to shorten the duration of waves of covid-19 infections in
34 vietnam, Frontiers in Public Health, 8 (2021), 559693, URL https://www.frontiersin.org/
35 article/10.3389/fpubh.2020.559693.
36 20. C. Xu, Y. Pei, S. Liu and J. Lei, Effectiveness of non-pharmaceutical interventions against local
37 transmission of covid-19: An individual-based modelling study, Infectious Disease Modelling, 6
38 (2021), 848–858.
39 21. O. Krivorotko, M. Sosnovskaia, I. Vashchenko, C. Kerr and D. Lesnic, Agent-based modeling

Mathematical Biosciences and Engineering Volume , Issue ,

 17

1 of covid-19 outbreaks for new york state and uk: Parameter identification algorithm, Infectious
2 Disease Modelling, 7 (2022), 30–44.
3 22. R. Alguliyev, R. Aliguliyev and F. Yusifov, Graph modelling for tracking the covid-19 pandemic
4 spread, Infectious disease modelling, 6 (2021), 112–122.
5 23. D. S. I. Kanté, A. Jebrane, A. Bouchnita and A. Hakim, Estimating the risk of contracting covid-
6 19 in different settings using a multiscale transmission dynamics model, Mathematics, 11 (2023),
7 254.
8 24. Z. Laroussi, The rise of consumer society in morocco, Journal of Applied Language and Culture
9 Studies, 5 (2022), 177–207, URL https://revues.imist.ma/index.php/JALCS/article/
10 view/24718.
11 25. E. T. Hall, The hidden dimension, Garden City, N.Y., Doubleday, New York, USA, 1966.
12 26. A. Sorokowska, P. Sorokowski, P. Hilpert, K. Cantarero, T. Frackowiak, K. Ahmadi et al., Pre-
13 ferred interpersonal distances: A global comparison, Journal of Cross-Cultural Psychology, 48
14 (2017), 577–592, URL https://doi.org/10.1177/0022022117698039.
15 27. E. Bosina and U. Weidmann, Estimating pedestrian speed using aggregated literature data, Physica
16 A: Statistical Mechanics and its Applications, 468 (2017), 1–29.
17 28. A. Williams, M. J. Ostwald, G. Brewer and R. Le Goff, Sustainable retail spaces: Establishing
18 the profile of the ”green” population, An International Journal of Urban & Extra Urban Studies,
19 4 (2014), 77–90.
20 29. Wall Street Journal Custom Content, Available online: https://partners.wsj.com/icsc/
21 shopping-for-the-truth/time-well-spent/article, 2017, Accessed March 7, 2023.
22 30. G. Shaw and A. M. Williams, Tourism and Tourism Spaces, SAGE Publications, Thousand Oaks,
23 CA, USA, 2004.
24 31. N. Taherifar, H. Hamedmoghadam, S. Sree and M. Saberi, A macroscopic approach for calibration
25 and validation of a modified social force model for bidirectional pedestrian streams, Transport-
26 metrica A: Transport Science, 15 (2018), 1637–1661.
27 32. S. Seer, C. Rudloff, T. Matyus and N. Brändle, Validating social force based models with com-
28 prehensive real world motion data, Transportation Research Procedia, 2 (2014), 724–732.
29 33. S. Heliövaara, T. Korhonen, S. Hostikka and H. Ehtamo, Counterflow model for agent-based
30 simulation of crowd dynamics, Building and Environment, 48 (2012), 89–100, URL https:
31 //www.sciencedirect.com/science/article/pii/S0360132311002630.
32 34. J. Lee, T. Kim, J.-H. Chung and J. Kim, Modeling lane formation in pedestrian counter flow and
33 its effect on capacity, KSCE Journal of Civil Engineering, 20 (2016), 1099–1108.
34 35. X. Chen, M. Treiber, V. Kanagaraj and H. Li, Social force models for pedestrian traffic–state of
35 the art, Transport reviews, 38 (2018), 625–653.
36 36. P. Wang, Understanding social-force model in psychological principles of collective behavior,
37 2016, URL https://arxiv.org/abs/1605.05146.
38 37. D. S. I. Kanté, A. Jebrane, A. Hakim and A. Boukamel, Characterization of superspreaders
39 movement in a bidirectional corridor using a social force model, Frontiers in Public Health, 11.

Mathematical Biosciences and Engineering Volume , Issue ,

 18

1 38. U. Weidmann, Transporttechnik der fußgänger: transporttechnische eigenschaften des

2 fußgängerverkehrs, literaturauswertung, IVT Schriftenreihe, 90.
3 39. M. Moussaı̈d, D. Helbing, S. Garnier, A. Johansson, M. Combe and G. Theraulaz, Experimen-
4 tal study of the behavioural mechanisms underlying self-organization in human crowds, Pro-
5 ceedings of the Royal Society B: Biological Sciences, 276 (2009), 2755–2762, URL https:
6 //royalsocietypublishing.org/doi/abs/10.1098/rspb.2009.0405.
7 40. J. Ma, W. guo Song, Z. ming Fang, S. ming Lo and G. xuan Liao, Experimental study on micro-
8 scopic moving characteristics of pedestrians in built corridor based on digital image processing,
9 Building and Environment, 45 (2010), 2160–2169.
10 41. PopulationPyramid.net, Available online: https://www.populationpyramid.net/morocco/
11 2021/, 2021, Accessed March 7, 2023.
12 42. Haut Commissariat du Plan, Available online: https://www.hcp.ma/
13 Les-Cahiers-du-Plan-N-43-Mars-Avril-2013_a1248.html, 2013, Accessed March
14 7, 2023.
15 43. P. Montero and M. Loukid, Croissance des enfants de la ville de marrakech (maroc) : analyse
16 comparative avec des enfants espagnols, Bulletins et mémoires de la Société d’anthropologie
17 de Paris, 3 (1991), 211–224, URL https://www.persee.fr/doc/bmsap_0037-8984_1991_
18 num_3_3_1783.
19 44. N. Mellouki, Vieillissement et mode alimentaire : étude anthropologique à Marrakech, PhD thesis,
20 Universite Cadi Ayyad, Marrakesh, 2007, URL http://wd.fmpm.uca.ma/biblio/theses/
21 annee-htm/FT/2007/these45-07.pdf.
22 45. Ministère de la Santé, Available online: https://www.sante.gov.ma/Documents/2019/05/
23 Rapport%20de%20l%20enqu%C3%AAte%20Stepwise.pdf, 2019, Accessed March 7, 2023.
24 46. T. Kretz, J. Lohmiller and P. Sukennik, Some indications on how to calibrate the social force
25 model of pedestrian dynamics, Transportation Research Record, 2672 (2018), 228–238.
26 47. J. Noh and G. Danuser, Estimation of the fraction of covid-19 infected people in u.s. states
27 and countries worldwide, PLOS ONE, 16 (2021), 1–10, URL https://doi.org/10.1371/
28 journal.pone.0246772.
29 48. The World Health Organization, Available online: https://covid19.who.int/region/emro/
30 country/ma, 2020, Accessed March 7, 2023.
31 49. World Health Organization, Available online: http:
32 //www.who.int/news-room/questions-and-answers/item/
33 coronavirus-disease-covid-19-how-is-it-transmitted, 2021, Accessed March
34 7, 2023.
35 50. C. Tadmon and S. Foko, A transmission dynamics model of covid-19: Case of cameroon, Infec-
36 tious Disease Modelling, 7 (2022), 211–249.
37 51. A. Li, Y. Wang, P. Cong and X. Zou, Re-examination of the impact of some non-pharmaceutical
38 interventions and media coverage on the covid-19 outbreak in wuhan, Infectious Disease Mod-
39 elling, 6 (2021), 975–987.

Mathematical Biosciences and Engineering Volume , Issue ,

 19

1 52. B. Gorbunov, Aerosol particles generated by coughing and sneezing of a SARS-CoV-2 (COVID-
2 19) host travel over 30 m distance, Aerosol and Air Quality Research, 21 (2021), 200468, URL
3 https://doi.org/10.4209%2Faaqr.200468.
4 53. A. Popa, J.-W. Genger, M. D. Nicholson, T. Penz, D. Schmid, S. W. Aberle et al., Genomic
5 epidemiology of superspreading events in austria reveals mutational dynamics and transmission
6 properties of sars-cov-2, Science Translational Medicine, 12 (2020), eabe2555, URL https:
7 //www.science.org/doi/abs/10.1126/scitranslmed.abe2555.
8 54. Y. Liu, Z. Ning, Y. Chen, M. Guo, Y. Liu, N. K. Gali et al., Aerodynamic analysis of sars-cov-2
9 in two wuhan hospitals, Nature, 582 (2020), 557—560, URL https://doi.org/10.1038/
10 s41586-020-2271-3.
11 55. A. P. S. Munro and S. N. Faust, Children are not covid-19 super spreaders: time to go back to
12 school, Archives of Disease in Childhood, 105 (2020), 618–619, URL https://adc.bmj.com/
13 content/105/7/618.
14 56. Moroccan Ministry of Health, Available online: https://www.sante.gov.ma/
15 Publications/Pages/Bullten_pidmiologique.aspx, 2020, Accessed March 7, 2023.
16 57. W. Sawadogo, M. Tsegaye, A. Gizaw and T. Adera, Overweight and obesity as risk factors for
17 covid-19-associated hospitalisations and death: systematic review and meta-analysis, BMJ Nutri-
18 tion, Prevention & Health, 5 (2022), 10–18, URL https://nutrition.bmj.com/content/5/
19 1/10.
20 58. S. Kwok, S. Adam, J. H. Ho, Z. Iqbal, P. Turkington, S. Razvi et al., Obesity: A critical risk factor
21 in the covid-19 pandemic, Clinical Obesity, 10 (2020), e12403, URL https://onlinelibrary.
22 wiley.com/doi/abs/10.1111/cob.12403.
23 59. M. Imai, K. Iwatsuki-Horimoto, M. Hatta, S. Loeber, P. J. Halfmann, N. Nakajima et al.,
24 Syrian hamsters as a small animal model for sars-cov-2 infection and countermeasure devel-
25 opment, Proceedings of the National Academy of Sciences, 117 (2020), 16587–16595, URL
26 https://www.pnas.org/doi/abs/10.1073/pnas.2009799117.
27 60. C. K. Yinda, J. R. Port, T. Bushmaker, I. Offei Owusu, J. N. Purushotham, V. A. Avanzato et al.,
28 K18-hace2 mice develop respiratory disease resembling severe covid-19, PLOS Pathogens, 17
29 (2021), 1–21, URL https://doi.org/10.1371/journal.ppat.1009195.
30 61. K. A. Ryan, K. R. Bewley, S. A. Fotheringham, G. S. Slack, P. Brown, Y. Hall et al., Dose-
31 dependent response to infection with sars-cov-2 in the ferret model and evidence of protec-
32 tive immunity, Nature Communications, 12 (2021), 81, URL https://doi.org/10.1038/
33 s41467-020-20439-y.
34 62. Y. J. Baek, T. Lee, Y. Cho, J. H. Hyun, M. H. Kim, Y. Sohn et al., A mathematical model of
35 covid-19 transmission in a tertiary hospital and assessment of the effects of different intervention
36 strategies, PLOS ONE, 15 (2020), 1–16, URL https://doi.org/10.1371/journal.pone.
37 0241169.
38 63. Z. Shen, F. Ning, W. Zhou, X. He, C. Lin, D. P. Chin et al., Superspreading sars events, beijing,
39 2003, Emerging infectious diseases, 10 (2004), 256.

Mathematical Biosciences and Engineering Volume , Issue ,

 20

1 64. S. X. Wang, Y. M. LI, B. C. Sun, S. W. Zhang, W. H. Zhao, M. T. Wei et al., The sars outbreak in
2 a general hospital in tianjin, china – the case of super-spreader, Epidemiology and Infection, 134
3 (2006), 786–791.
4 65. D. A. Edwards, D. Ausiello, J. Salzman, T. Devlin, R. Langer, B. J. Beddingfield et al., Exhaled
5 aerosol increases with covid-19 infection, age, and obesity, Proceedings of the National Academy
6 of Sciences, 118 (2021), e2021830118.
7 66. Y. Zhu, C. J. Bloxham, K. D. Hulme, J. E. Sinclair, Z. W. M. Tong, L. E. Steele et al., A
8 Meta-analysis on the Role of Children in Severe Acute Respiratory Syndrome Coronavirus 2 in
9 Household Transmission Clusters, Clinical Infectious Diseases, 72 (2020), e1146–e1153, URL
10 https://doi.org/10.1093/cid/ciaa1825.
11 67. A. Eykelbosh, Available online: https://ncceh.ca/documents/evidence-review/
12 contextualizing-risks-indirect-covid-19-transmission-multi-unit, 2021, Ac-
13 cessed March 7, 2023.
14 68. F. Johansson, A. Peterson and A. Tapani, Waiting pedestrians in the social force model, Phys-
15 ica A: Statistical Mechanics and its Applications, 419 (2015), 95–107, URL https://www.
16 sciencedirect.com/science/article/pii/S0378437114008413.
17 69. G. H. Subramanian, N. Choubey and A. Verma, Modelling and simulating serpentine group
18 behaviour in crowds using modified social force model, Physica A: Statistical Mechanics and
19 its Applications, 604 (2022), 127674, URL https://www.sciencedirect.com/science/
20 article/pii/S0378437122004514.
21 70. P. Cheng, K. Luo, S. Xiao, H. Yang, J. Hang, C. Ou, B. J. Cowling, H.-L. Yen, D. S. Hui,
22 S. Hu and Y. Li, Predominant airborne transmission and insignificant fomite transmission of
23 sars-cov-2 in a two-bus covid-19 outbreak originating from the same pre-symptomatic index case,
24 Journal of Hazardous Materials, 425 (2022), 128051, URL https://www.sciencedirect.
25 com/science/article/pii/S030438942103020X.
26 71. K. K.-W. To, O. T.-Y. Tsang, C. C.-Y. Yip, K.-H. Chan, T.-C. Wu, J. M.-C. Chan, W.-S. Le-
27 ung, T. S.-H. Chik, C. Y.-C. Choi, D. H. Kandamby et al., Consistent detection of 2019 novel
28 coronavirus in saliva, Clinical Infectious Diseases, 71 (2020), 841–843.
29 72. M. Riediker and D.-H. Tsai, Estimation of viral aerosol emissions from simulated individuals with
30 asymptomatic to moderate coronavirus disease 2019, JAMA network open, 3 (2020), e2013807–
31 e2013807.

© 2023 the Author(s), licensee AIMS Press. This

is an open access article distributed under the
32
terms of the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/4.0)

Mathematical Biosciences and Engineering Volume , Issue ,