An agent-based epidemics simulation to compare and explain screening and vaccination prioritisation strategies

2025-04-30 0 0 1.21MB 14 页 10玖币

侵权投诉

An agent-based epidemics simulation to

compare and explain screening and

vaccination prioritisation strategies

Carole Adam (Univ. Grenoble-Alpes, LIG, Grenoble, France)

and H´

el`

ene Arduin (UMR IDEES, CNRS, Rouen, France)

Abstract

This paper describes an agent-based model of epidemics dynamics. This model is willingly simpliﬁed, as its goal is not

to predict the evolution of the epidemics, but to explain the underlying mechanisms in an interactive way. This model

allows to compare screening prioritisation strategies, as well as vaccination priority strategies, on a virtual population.

The model is implemented in Netlogo in different simulators, published online to let people experiment with them. This

paper reports on the model design, implementation, and experimentations. In particular we have compared screening

strategies to evaluate the epidemics vs control it by quarantining infectious people; and we have compared vaccinating

older people with more risk factors, vs younger people with more social contacts.

Note: This paper is an extended version of a conference paper presented at ISCRAM 2022 [1].

Keywords

Agent-based modelling and simulation, epidemics modelling, screening, vaccination, contacts, scientiﬁc mediation

Introduction

The COVID-19 epidemics has now been lasting for over 2

years since the ﬁrst cases in late 2019. The only control

strategy for the ﬁrst year was general lockdowns, but was

hard to maintain longterm for economical and mental health

reasons. In a second phase, when tests became available,

the new strategy was to start large screening campaigns and

to isolate (detected) sick individuals. Vaccines have then

become available in December 2020, and have allowed to

lift most constraining sanitary measures.

However, with the epidemics lasting longer than expected

and its end being hard to predict, people are tiring out,

sanitary measures are not always well accepted, trust goes

down [33]. Fake news circulate with deadly consequences

[27], such as refusing or hesitating to get the vaccine [12,22].

We believe that it is very important to inform the population

and explain the mechanisms of the epidemics and the reasons

behind all measures [28]. Indeed, understanding constraints

will improve their acceptability.

We claim that an interactive simulator is a good tool to

explain mechanisms by letting users play a role and learn by

exploring what-if scenarios. We have therefore designed an

agent-based model that allows to simulate various screening

and vaccination strategies on a virtual population, and to

compare these strategies in order to discover insight about

their optimal parameters. It is simple and interactive, and

users from the general population can play with it in order to

understand the complex mechanisms behind the epidemics

and the reasons for the various sanitary measures. This work

is part of a larger initiative aiming at answering questions

from the general public about the COVID-19 epidemics,

through interactive simulators along with explaining texts

written by researchers from various disciplines [11].

Challenges of screening

This section introduces some useful background about

screening, quality features of tests, and possible prioritisation

strategies for allocating a limited number of tests.

Lockdowns and screening

When lifting the lockdown after the ﬁrst COVID-19 wave in

spring 2020, the main goals of many countries around the

world were to get back to a less restricted way of living

while still maintaining the epidemic under control to avoid

a ”second wave”. Indeed, due to the low circulation of the

virus during lockdown, herd immunity was still insufﬁcient

to prevent a rebound and new wave. For instance, it was

estimated*that only 3 to 7 % of the French population had

been exposed to the virus (and was therefore immunised)

when exiting the ﬁrst French lockdown in May 2020. And

as time has since proved, not only a second wave, but several

more epidemics waves appeared, pushing some countries to

enforce other lockdowns.

But the general lockdown is hard to maintain on the

longterm for both psychological and economic reasons [3],

and is hard to lift without creating a new wave since herd

immunity does not develop during lockdown. One solution

is to selectively isolate only sick individuals. However, this

strategy is hard to implement efﬁciently. Indeed, the COVID-

19 incubation time is long (a week on average, but sometimes

up to 20 days), so infected people have time to infect their

contacts before they are detected and quarantined. Besides,

the share of asymptomatic cases was still mostly unknown,

but estimated to be around 30% [35], meaning many infected

∗by Pasteur Institute: https://www.pasteur.fr/fr/espace-presse/documents-

presse/modelisation-indique-qu-entre-3-7-francais-ont-ete-infectes

Pre-print - 21 October 2022

arXiv:2210.13089v1 [cs.MA] 24 Oct 2022

2Pre-print 2022(10-22)

people could unknowingly spread the virus among their

contacts.

This implies that governments could not rely entirely

on symptomatic displays to isolate infected people, but

needed to test their population broadly by running large

scale screening campaigns. This is precisely the strategy

recommended by the World Health Organisation (WHO), as

early as the 16 March 2020†: to test any suspicious case

to conﬁrm potentially infected individuals; to trace their

contacts in order to identify chains of contamination; and

isolate only (potentially) infectious people. But it took time

to develop reliable tests and start this strategy.

Quality of tests

There exists different types of tests to detect the SARS-

COV-2 virus responsible for COVID-19, in particular PCR

(polymerase chain reaction) tests, serological tests, antigenic

tests, and auto-tests that one can realise at home. These tests

have different levels of quality, depending on 2 factors:

•Sensitivity of a test indicates the probability that the

test is positive when the tested person is really sick.

A 100% sensitive test applied to a sick individual

will always return positive; therefore a negative test

gives absolute certainty that the tested individual is

indeed not sick. In other words, there are no false

negatives with a 100% sensitive test; so sensitivity is

the proportion of true negatives.

•Speciﬁcity of a test indicates the probability that the

test is negative when the tested person is really not

sick. A 100% speciﬁc test applied to a non sick

individual will always return negative; therefore a

positive test gives absolute certainty that the tested

individual is indeed sick. In other words, there are no

false positives with a 100% speciﬁc test, so speciﬁcity

is the proportion of true positives.

However, it is impossible to design tests that are perfect on

both criteria (or even on a single one). Screening tests always

have an error margin. In particular, screening tests cannot be

both highly speciﬁc and highly sensitive, so a compromise

must be found between two opposites:

• Very sensitive tests are more likely to be positive

with sick individuals: this reduces the rate of false

negatives, so prevents missing infected people who

keep moving around instead of being quarantined;

• Very speciﬁc tests are less likely to be positive when

the individual is not sick: this reduces the rate of false

positives, to prevent from quarantining healthy people.

The ﬁrst screening tests designed for COVID-19 were

relatively speciﬁc (in the range of 95 to 98% of true positives)

but still little sensitive (sometimes up to 30 to 40% of false

negatives, sick but not detected by the test). As a result, it

was sometimes necessary to do a second test to conﬁrm a

negative test result.

Screening objectives

Time was needed to develop reliable quality tests and

increase testing capacity. As a result, testing kits were rare at

the start of the epidemics, forcing governments to prioritise

who should be tested ﬁrst to optimise the impact of the

testing campaign. Even nowadays, when testing kits are

widely available, and as new variants of the virus circulate

very fast, the number of daily tests has exploded, posing a

new issue of ﬁnancing those tests. Some countries therefore

again choose to restrict tests to some categories of people,

for instance, the elderly who are more at risk of serious

forms, or people with symptoms. Other countries require non

vaccinated people to pay for the tests, also in order to limit

the number of tests performed.

Screening tests actually pursue two main (partly contra-

dictory) goals.

• The ﬁrst one is to control the epidemics, by

spotting infected people and isolating them to break

contamination chains.

• The second one is to know the epidemic, i.e. evaluate

as precisely as possible the total number of people

infected at a given time, and deduce the actual case

fatality rate.

These goals involve different screening strategies: in order to

best control the epidemics, one should test in priority people

who are more likely to carry the virus, but this leads to an

over-estimation of the global circulation; to best know the

epidemics, one should randomly test a representative sample

of the global population, but this will lead to a large number

of negative tests, failing to isolate many infected people. The

best screening strategy is therefore not intuitive.

Screening prioritisation strategies

Under the constraint that testing kits are in limited supply,

governments want to prioritise wisely who should be tested,

in order to reach both goals with the minimum amount of

tests. For instance, France started testing late and slowly‡: it

took some time to design reliable tests, and the small number

of such available tests was thus limited to healthcare workers

and people at risk. Nowadays, tests are widely available and

are the most cost-effective mitigating measure [30], but some

countries start restricting them again in order to limit the

ﬁnancial cost for society, for instance, by reserving them to

elderly people, or by asking non-vaccinated individuals to

pay for the tests.

The various possible targeting strategies have different

impacts on both goals stated above:

•Random targeting consists in choosing randomly

people who should be tested. This is a more

representative sample of the population, and provides

better knowledge of the current state of the epidemics.

But when the incidence of the virus is very small

(as it was after the ﬁrst lockdown), the proportion of

people infected is very low, so most tests will return

negative. There is therefore a risk of ”wasting” many

tests, i.e. the chances of ﬁnding infected people to

isolate them and control the epidemics are low.

†https://www.who.int/dg/speeches/detail/

who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---16-march-2020

‡https://www.usinenouvelle.com/article/

en-retard-la-france-monte-en-puissance-pour-les-tests-de-diagnostic-du-covid-19.

N945261

Prepared using sagearxiv.cls

Adam and Arduin 3

• A solution is to target suspicious cases (the

symptomatic ones), but this strategy is insufﬁcient to

control the epidemics since it ignores all the (also

contagious) asymptomatic cases. Besides, the sample

is not representative of the general population, and the

high proportion of positive tests in the sample might

lead to overestimate the global circulation of the virus.

• Another strategy consists in targeting people who

work outside of home, since they are more likely

to get infected and/or infect others. For instance, at

the beginning of the epidemic, healthcare workers

were tested in priority, since they were the most

exposed to the virus; in order to reopen schools, there

was also a focus on testing teachers, school workers,

and now all the children from the same class as an

infected pupil. This strategy focuses on controlling the

epidemics while allowing for economic activity, but

it ignores contaminations that happen outside of work

(shopping, leisure...).

• Finally a last interesting strategy consists in targeting

high-risk people. Their proﬁle is now better known, in

particular elderly people or people with comorbidities

[23]. The goal of this strategy is to detect infection

soon and treat them early to prevent serious

complications. But the results would then not be

representative of the global circulation of the epidemic

in the general population.

The choice is not intuitive, and we claim that simulation

can help compare different strategies in order to draw

interesting insight. Indeed, simulation allows to run the exact

same scenario with only the parameters of the screening

campaign varying, which is impossible in reality, and to

compare estimations with the ”real” epidemic curve, which

can be known only in a simulation.

Challenges of vaccination

Vaccines became available in December 2020, and many

different ones are now available. This section provides some

useful background related to our model.

Vaccination

The goal of vaccination is to provide people with some level

of immunity against the virus, that will protect them from

infection, and ultimately to create collective immunity at

the level of the population to stop further propagation of

the virus. The impact of the vaccination campaign can be

measured on different indicators: the height of the epidemic

peak (how many people are sick at the same time at the peak),

the duration of the epidemics wave (how long it takes before

nobody is sick anymore), the total number of people who

got sick, the total number of serious cases, the number of

people in hospitals, or the number of deaths. The efﬁciency

of a vaccine can be deﬁned on different terms [19], such as

reducing the risk of infection, the severity of the disease, or

the duration of infectivity. A vaccine can have different levels

of efﬁciency on these different aspects. It was initially hard

to know exactly which factor was impacted by the vaccine:

its effect on infection was tested by trials before release,

but it was not clear if it was also effective on transmission

[26], on asymptomatic forms [8] or on the risk of serious

forms. Governments sometimes made simpliﬁed statements

to encourage people to get the vaccine§. The efﬁciency of

the vaccine has also decreased against the new variants of

the virus [7].

Individual vulnerability

Literature has shown that infections and serious forms are

more frequent in people with risk factors, in particular

those aged above 60 [31] and/or having comorbidities (often

associated with age) such as diabetes or chronic illnesses

[14,6,15].

Literature also shows that younger people have more

contacts in average than older people. For instance [21]

studied inﬂuenza in Japan and showed difference in contact

patterns based on age, gender, as well as during the week

vs holiday. [5] also study the role of contacts with people of

different age groups; they conclude that the case growth rate

increases when there is more contacts with elderly people,

but decreases with contacts among the same age group.

Vaccine prioritisation strategies

The ﬁrst vaccine against COVID-19 was released in

December 2020. Developed countries invested a lot of

money to secure enough doses for their population, but the

production and injection of millions of doses takes time.

Furthermore, the immunity conferred by the vaccine only

lasts for a few months [13], so it is necessary to inject

boosters regularly. As a result, governments are faced with

a choice about who should be vaccinated ﬁrst:

• Vaccinating in priority people with comorbidities

(elderly people or people with other health factors

increasing their vulnerability), in order to protect them

from serious forms of the illness;

• Vaccinating in priority people with more contacts

(such as health workers who are in contact with many

sick or vulnerable people, teachers, or younger people

who have more social contacts), in order to reduce the

global transmission of the virus.

• Randomly vaccinating the entire population, without

any order of priority.

Similar to screening, the optimal strategy is not intuitive

since different strategies improve different indicators. One

might reduce the number of serious forms but leave many

younger people exposed, or reduce the total number of

contaminations to prevent saturation of the healthcare system

but take the risk of more serious forms among vulnerable

people. Therefore, once again simulation could help in

testing strategies by varying their parameters in isolation.

Agent-based simulation of epidemics

Goal of this work

One can see from the background above that ﬁnding the

best (screening or vaccination) strategy on all accounts is

§https://www.liberation.fr/checknews/

transmission-du-covid-19-les-autorites-ont-elles-menti-sur-lefficacite-du-vaccin-pour-justifier-les-pass-sanitaire-et-vaccinal-20221014_

JEAR5KU73RFNTPRDWQCLHSNZQE/

Prepared using sagearxiv.cls

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

Anagent-basedepidemicssimulationtocompareandexplainscreeningandvaccinationprioritisationstrategiesCaroleAdam(Univ.Grenoble-Alpes,LIG,Grenoble,France)andH´eleneArduin(UMRIDEES,CNRS,Rouen,France)AbstractThispaperdescribesanagent-basedmodelofepidemicsdynamics.Thismodeliswillinglysimplied,asitsgoalisn...

展开>> 收起<<

An agent-based epidemics simulation to compare and explain screening and vaccination prioritisation strategies.pdf

共14页,预览3页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

An agent-based epidemics simulation to compare and explain screening and vaccination prioritisation strategies

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: