Berks County Medical Society Medical Record Summer 2020 - 41

How effective is population testing? What should be
the daily testing capacity? What is the effect of test
accuracy?
The models formulated in [6, 13] showed that millions of
individuals would have already been infected with the virus by
the time the first death is observed, indicating the importance of
testing the population. Our results indicated that population testing
without epidemic suppression strategies was a nonviable solution to
control the spread of the virus in our virtual town. For the baseline
case R0=2.4, increasing the daily testing capacity from 100 to 500
had a small impact on the peak of the infection. On the other hand,
population testing had a significant effect when applied with social
distancing measures. For example, the infection curve with an
average of 10 contacts became significantly flattened by increasing
the daily testing capacity from 100 to 500 (Figure 3).
In order to investigate how a more accurate test would affect
the containment of the virus, we also used the true-positive rate
of 95%. The results showed a significant flattening of the curve
for the scenario with 10 contacts but not for the other scenarios,
which suggest the importance of making informed decisions about
identifying individuals to test.

What should be ICU capacity?
A major concern about the impact of COVID-19 on healthcare
providers is the availability of intensive-care units (ICU). The
Imperial College London model predicted that the surge of patients
would exceed Great-Britain's ICU bed capacity by 8-fold [4], which
was one of the factors for UK government officials' pivoting to social
distancing measures [5]. Considering the ICU bed capacity of 3
per 10,000 people on average, our simulation results indicated that
the ICU-bed capacity would be marginally adequate to cope with
the peak demand only for the strict social distancing/high testing
scenario in our experiments (Figure 4). For other scenarios, the ICU
bed capacity should be 8- to 15-fold higher to cope with the peak
demand, which is in line with the Imperial College London model's
predictions [4]. When we used the true-positive rate of 95% for
testing, the ICU bed capacity was only adequate for the strict social
distancing/high testing scenario.

Figure 4. Simulation results demonstrating the demand for ICU beds per
day under the scenarios studied with the true-positive rate of 70%.

Discussions
Although our ABS model of a virtual town is only for
demonstration purposes, it produced results remarkably similar
to the patterns that we have seen with COVID-19 data. Like any

modeling tools, ABS models have strengths and weaknesses. Firstly,
their accuracy depends on the availability and quality of the data.
The predictions made in the very early stages of the pandemic were
shown to be off because of the lack of high-quality data about
COVID-19, but recent predictions improved significantly as more
data became available. Therefore, it is essential that countries and
entities share their data openly and promptly. Secondly, all models
are an abstraction of reality. In this article, we frequently use the
term "assume" because it is impossible to represent every detail
of the real-world as-is. In our model, all agents act the same way,
they have the same level of social interactions, and they are all rule
obeying citizens. It is possible to incorporate more granular agent
behaviors by considering different age groups, working conditions,
actual locations, personal connections, ignorance of the rules, etc.
For example, using individuals' actual interaction and movement
data tracked by social media and mobile companies can improve
the accuracy of ABS models. However, the detailed models are
also computationally difficult to simulate. A good ABS model that
captures the essence of the reality with reasonable assumptions can
perform as good as a very detailed model.
Despite their weaknesses and flaws, ABS models have helped
scientists gain insights and offer suggestions to the policymakers
and the public in the case of COVID-19. They are powerful
modeling tools for "what-if " analysis. Even with incomplete and
imperfect data, ABS models can compare alternative scenarios and
make projections for the real-world. Currently, we are working on
improving our model to incorporate more granular agent behaviors
such as ignoring intervention measures and the behavior of different
age groups.
References
[1] K. Dietz and J. Heesterbeek, "Daniel Bernoulli's epidemiological model
revisited," Mathematical biosciences, vol. 180, pp. 1-21, 2002.
[2] W. O. Kermack and A. G. McKendrick, "A contribution to the
mathematical theory of epidemics," Proceedings of the royal society of London.
Series A, Containing papers of a mathematical and physical character, vol. 115,
pp. 700-721, 1927.
[3] C. I. Siettos and L. Russo, "Mathematical modeling of infectious disease
dynamics," Virulence, vol. 4, pp. 295-306, 2013.
[4] N. Ferguson, D. Laydon, G. Nedjati Gilani, N. Imai, K. Ainslie, M.
Baguelin, et al., "Report 9: Impact of non-pharmaceutical interventions (NPIs)
to reduce COVID19 mortality and healthcare demand," 2020.
[5] D. Adam, "Special report: The simulations driving the world's response to
COVID-19," Nature, vol. 580, p. 316, 2020.
[6] J. Lourenco, R. Paton, M. Ghafari, M. Kraemer, C. Thompson, P.
Simmonds, et al., "Fundamental principles of epidemic spread highlight the
immediate need for large-scale serological surveys to assess the stage of the SARSCoV-2 epidemic," medRxiv, p. 2020.03.24.20042291, 2020.
[7] CDC. (2020, 5/28/2020). COVID-19 Pandemic Planning Scenarios.
Available: https://www.cdc.gov/coronavirus/2019-ncov/hcp/planning-scenarios.
html
[8] HMS. (2020, 5/25/2020). If you've been exposed to the coronavirus.
Available: https://www.health.harvard.edu/diseases-and-conditions/if-youvebeen-exposed-to-the-coronavirus
[9] J. Watson, P. F. Whiting, and J. E. Brush, "Interpreting a covid-19 test
result," Bmj, vol. 369, pp. 1-7, 2020.
[10] J. Rothwell. (2020, April 21, 2020). Americans' Social Contacts During
the COVID-19 Pandemic. Available: https://news.gallup.com/opinion/
gallup/308444/americans-social-contacts-during-covid-pandemic.aspx
[11] J. Zhang, M. Litvinova, Y. Liang, Y. Wang, W. Wang, S. Zhao, et al.,
"Age profile of susceptibility, mixing, and social distancing shape the dynamics
of the novel coronavirus disease 2019 outbreak in China," medRxiv, p.
2020.03.19.20039107, 2020.
[12] S. Pei, S. Kandula, and J. Shaman, "Differential Effects of Intervention
Timing on COVID-19 Spread in the United States," medRxiv, p.
2020.05.15.20103655, 2020.
[13] T. Jombart, K. van Zandvoort, T. Russell, C. Jarvis, A. Gimma, S. Abbott,
et al., "Inferring the number of COVID-19 cases from recently reported deaths,"
medRxiv, p. 2020.03.10.20033761, 2020.
SUMMER 2020

Berks County Medical Society Medical Record Summer 2020

Table of Contents for the Digital Edition of Berks County Medical Society Medical Record Summer 2020