A study of disproportionately affected populations by race/ethnicity during the SARS-CoV-2 pandemic using multi-population SEIR modeling and ensemble data assimilation

Emmanuel Fleurantin; Christian Sampson; Daniel Paul Maes; Justin Bennett; Tayler Fernandes-Nunez; Sophia Marx; Geir Evensen

doi:10.3934/fods.2021022

Article Contents

2021, Volume 3, Issue 3: 479-541. Doi: 10.3934/fods.2021022

This issue Previous Article An international initiative of predicting the SARS-CoV-2 pandemic using ensemble data assimilation Next Article Feedback particle filter for collective inference

A study of disproportionately affected populations by race/ethnicity during the SARS-CoV-2 pandemic using multi-population SEIR modeling and ensemble data assimilation

1.
Florida Atlantic University, 777 Glades Rd., Boca Raton, FL 33431, USA
2.
University of North Carolina, 130 Mason Farm Road Chapel Hill, NC 27599, USA
3.
University of Michigan, 530 Church St, Ann Arbor, MI 48109, USA
4.
Arizona State University, 1151 S Forest Ave, Tempe, AZ 85281, USA
5.
Smith College, Northampton, MA 01063, USA
6.
NORCE Norwegian Research Centre AS, Nygårdsporten 112, 5008 Bergen, Norway

Received: January 2021

Revised: August 2021

Early access: September 2021

Published: September 2021

The second author (CSS) is supported by the US Office of Naval Research under grant N00014-18-1-2204. The third author (DPM) is supported by the NSF GRFP DGE 1256260. The seventh author (GE) was supported by internal funding from NORCE

Abstract / Introduction Full Text(HTML) Figure(45) / Table(9) Related Papers Cited by

Abstract

The disparity in the impact of COVID-19 on minority populations in the United States has been well established in the available data on deaths, case counts, and adverse outcomes. However, critical metrics used by public health officials and epidemiologists, such as a time dependent viral reproductive number ($ R_t $), can be hard to calculate from this data especially for individual populations. Furthermore, disparities in the availability of testing, record keeping infrastructure, or government funding in disadvantaged populations can produce incomplete data sets. In this work, we apply ensemble data assimilation techniques which optimally combine model and data to produce a more complete data set providing better estimates of the critical metrics used by public health officials and epidemiologists. We employ a multi-population SEIR (Susceptible, Exposed, Infected and Recovered) model with a time dependent reproductive number and age stratified contact rate matrix for each population. We assimilate the daily death data for populations separated by ethnic/racial groupings using a technique called Ensemble Smoothing with Multiple Data Assimilation (ESMDA) to estimate model parameters and produce an $R_t(n)$ for the $n^{th}$ population. We do this with three distinct approaches, (1) using the same contact matrices and prior $R_t(n)$ for each population, (2) assigning contact matrices with increased contact rates for working age and older adults to populations experiencing disparity and (3) as in (2) but with a time-continuous update to $R_t(n)$. We make a study of 9 U.S. states and the District of Columbia providing a complete time series of the pandemic in each and, in some cases, identifying disparities not otherwise evident in the aggregate statistics.

Keywords:

Mathematics Subject Classification: 91F99, 65K10, 65P99, 65K99.

Citation:

Full Text(HTML)

Figure 1. The general form of the $ R_{ij}^A(n) $ contact matrix elements for contact rates between age groups $ i $ and $ j $ in a given sub-population. The contact matrix is then subdivided into different blocks where parameters $ \alpha, \beta, \eta, \gamma, \varepsilon, \zeta, \delta_1, \delta_2, \delta_3, \xi_1, \xi_2, \xi_3 $ and $ \xi_4 $ control the contact rates between different age groups which generate similar patterns for spreading the disease. In particular, we define $ \gamma, \delta_2, \xi_2 $ and $ \xi_3 $ to be the parameters for the contact rates of the working class age groups

Download: Full-size image PowerPoint slide

Figure 2. $ \mathbf{R}^{\rm{A}} $ for the NDAP $ (i) $ and DAP $ (ii) $ in our runs without DA for each intervention period along with the $ \mathbf{R}^{\rm{C}} $ matrix $ (iii) $ used in these simulations. We also used the $ \mathbf{R}^{\rm{A}} $'s in $ (i) $ and $ (ii) $ for the DA runs

Download: Full-size image PowerPoint slide

Figure 3. Results for non-DA runs with DAP and NDAP at about the same population

Download: Full-size image PowerPoint slide

Figure 4. Results for non-DA runs with NDAP at $ 70\% $ of the total population

Download: Full-size image PowerPoint slide

Figure 5. Examples of analysis runs for various values of $ R^C_{nm} $ from the District of Columbia (DC). Top: left $ R^C_{nm} = 0 $, right $ R^C_{nm} = 10^{-3} $. Bottom: left $ R^C_{nm} = 10^{-1} $, right $ R^C_{nm} = 6\times10^{-1} $

Download: Full-size image PowerPoint slide

Figure 6. Analysis results for the District of Columbia (DC) with $ R^A_{ij}=1 $ and piece-wise updates to $ R_t(n) $

Download: Full-size image PowerPoint slide

Figure 7. Analysis results for the District of Columbia (DC) with age stratified $ R^A $ and piece-wise updates to $ R_t(n) $

Download: Full-size image PowerPoint slide

Figure 8. Analysis results for the continuous update case for the state of AK

Download: Full-size image PowerPoint slide

Figure 9. Analysis results for the continuous update case for the state of CA

Download: Full-size image PowerPoint slide

Figure 10. Analysis results for the continuous update case for the state of CT

Download: Full-size image PowerPoint slide

Figure 11. Analysis results for the continuous update case for the state of DE

Download: Full-size image PowerPoint slide

Figure 12. Analysis results for the continuous update case for the District of Columbia

Download: Full-size image PowerPoint slide

Figure 13. Analysis results for the continuous update case for the state of HI

Download: Full-size image PowerPoint slide

Figure 14. Analysis results for the continuous update case for the state of MD

Download: Full-size image PowerPoint slide

Figure 15. Analysis results for the continuous update case for the state of MI

Download: Full-size image PowerPoint slide

Figure 16. Analysis results for the continuous update case for the state of UT

Download: Full-size image PowerPoint slide

Figure 17. Analysis results for the continuous update case for the state of WA

Download: Full-size image PowerPoint slide

Figure 18. Analysis results when using piece-wise updates to $ R(t) $ for AK. Top row: $ R^A $ with entries of all ones. Bottom Row: $ R^A $ for DAPs and NDAPs

Download: Full-size image PowerPoint slide

Figure 19. Analysis results when using piecewise updates to $ R(t) $ for CA. Top row: $ R^A $ with entries of all ones. Bottom Row: $ R^A $ for DAPs and NDAPs

Download: Full-size image PowerPoint slide

Figure 20. Analysis results when using piecewise updates to $ R(t) $ for CT. Top row: $ R^A $ with entries of all ones. Bottom Row: $ R^A $ for DAPs and NDAPs

Download: Full-size image PowerPoint slide

Figure 21. Analysis results when using piecewise updates to $ R(t) $ for DE. Top row: $ R^A $ with entries of all ones. Bottom Row: $ R^A $ for DAPs and NDAPs

Download: Full-size image PowerPoint slide

Figure 22. Analysis results when using piecewise updates to $ R(t) $ for DC. Top row: $ R^A $ with entries of all ones. Bottom Row: $ R^A $ for DAPs and NDAPs

Download: Full-size image PowerPoint slide

Figure 23. Analysis results when using piecewise updates to $ R(t) $ for HI. Top row: $ R^A $ with entries of all ones. Bottom Row: $ R^A $ for DAPs and NDAPs

Download: Full-size image PowerPoint slide

Figure 24. Analysis results when using piecewise updates to $ R(t) $ for MD. Top row: $ R^A $ with entries of all ones. Bottom Row: $ R^A $ for DAPs and NDAPs

Download: Full-size image PowerPoint slide

Figure 25. Analysis results when using piecewise updates to $ R(t) $ for MI. Top row: $ R^A $ with entries of all ones. Bottom Row: $ R^A $ for DAPs and NDAPs

Download: Full-size image PowerPoint slide

Figure 26. Analysis results when using piecewise updates to $ R(t) $ for UT. Top row: $ R^A $ with entries of all ones. Bottom Row: $ R^A $ for DAPs and NDAPs

Download: Full-size image PowerPoint slide

Figure 27. Analysis results when using piecewise updates to $ R(t) $ for WA. Top row: $ R^A $ with entries of all ones. Bottom Row: $ R^A $ for DAPs and NDAPs

Download: Full-size image PowerPoint slide

Figure 28. Analysis results with a CFR prior of $ 0.009 $ ($ \sigma_{CFR} = 0.05 $) and the same $ R_t(n) $ ($ \sigma_{R(t)} = 1.5 $) prior for all populations. This figure corresponds to columns 2 and 3 in Table 5

Download: Full-size image PowerPoint slide

Figure 29. Analysis results with a CFR prior of $ 0.020 $ ($ \sigma_{CFR} = 0.05 $) and the same $ R_t(n) $ ($ \sigma_{R(t)} = 1.5 $) prior for all populations. This figure corresponds to columns 4 and 5 in Table 5

Download: Full-size image PowerPoint slide

Figure 30. Analysis results with a CFR prior of $ 0.001 $ ($ \sigma_{CFR} = 0.03 $) and the same $ R_t(n) $ ($ \sigma_{R(t)} = 1.5 $) prior for all populations. This figure corresponds to columns 6 and 7 in Table 5

Download: Full-size image PowerPoint slide

Figure 31. Analysis results with a CFR prior of $ 0.020 $ ($ \sigma_{CFR} = 0.05 $) and prior $ R_t(n) $ ($ \sigma_{R(t)} = 0.5 $) curves coming from initial piece-wise assimilation also assuming a $ 0.020 $ CFR prior. This figure corresponds to columns 2 and 3 in Table 6

Download: Full-size image PowerPoint slide

Figure 32. Analysis results with a CFR prior of $ 0.020 $ ($ \sigma_{CFR} = 0.002 $) and prior $ R_t(n) $ ($ \sigma_{R(t)} = 0.5 $) curves coming from initial piece-wise assimilation also assuming a $ 0.020 $ CFR prior. This figure corresponds to columns 4 and 5 in Table 6

Download: Full-size image PowerPoint slide

Figure 33. Analysis results with a CFR prior of $ 0.020 $ ($ \sigma_{CFR} = 0.05 $) for the Black population only and a CFR Prior of $ 0.009 $ ($ \sigma_{CFR} = 0.05 $) for all other populations. The prior $ R_t(n) $ ($ \sigma_{R(t)} = 1.5 $) curves are the same for all populations. This figure corresponds to columns 2 and 3 in Table 7

Download: Full-size image PowerPoint slide

Figure 34. Analysis results with a CFR prior of $ 0.020 $ ($ \sigma_{CFR} = 0.002 $) for the Black population only and a CFR Prior of $ 0.009 $ ($ \sigma_{CFR} = 0.0009 $) for all other populations. The prior $ R_t(n) $ ($ \sigma_{R(t)} = 0.5 $) curves coming from initial piece-wise assimilation also assuming a $ 0.020 $ CFR prior. This figure corresponds to columns 4 and 5 in Table 7 and is the only case where Black infections can be lower than that of the LatinX population and still have more deaths

Download: Full-size image PowerPoint slide

Figure 35. Analysis results with a CFR prior of $ 0.020 $ ($ \sigma_{CFR} = 0.05 $) for the Black population only and a CFR Prior of $ 0.009 $ ($ \sigma_{CFR} = 0.05 $) for all other populations. The prior $ R_t(n) $ ($ \sigma_{R(t)} = 0.5 $) curves coming from initial piece-wise assimilation also assuming a $ 0.020 $ CFR prior. This figure corresponds to columns 6 and 7 in Table 7

Download: Full-size image PowerPoint slide

Figure 36. Analysis results for the continuous update case for the state of AK

Download: Full-size image PowerPoint slide

Figure 37. Analysis results for the continuous update case for the state of CA

Download: Full-size image PowerPoint slide

Figure 38. Analysis results for the continuous update case for the state of CT

Download: Full-size image PowerPoint slide

Figure 39. Analysis results for the continuous update case for the state of DC

Download: Full-size image PowerPoint slide

Figure 40. Analysis results for the continuous update case for the state of DE

Download: Full-size image PowerPoint slide

Figure 41. Analysis results for the continuous update case for the state of HI

Download: Full-size image PowerPoint slide

Figure 42. Analysis results for the continuous update case for the state of MD

Download: Full-size image PowerPoint slide

Figure 43. Analysis results for the continuous update case for the state of MI

Download: Full-size image PowerPoint slide

Figure 44. Analysis results for the continuous update case for the state of UT

Download: Full-size image PowerPoint slide

Figure 45. Analysis results for the continuous update case for the state of WA

Download: Full-size image PowerPoint slide

Table 1. Matrix scaling parameters for DAP and NDAP workers in lockdown and post-lockdown time periods

Scalings	Matrix Parameters
Scalings	$\alpha$	$\beta$	$\gamma$	$\eta$	$\epsilon$	$\zeta$	$\delta_1$	$\delta_2$	$\delta_3$	$\xi_1$	$\xi_2$	$\xi_3$	$\xi_4$
DAP lock	0.5	0.7	0.55	0.25	0.25	0.35	0.3	0.7	0.7	0.4	0.65	0.55	0.6
DAP post-lock	0.7	0.8	0.7	0.3	0.3	0.4	0.6	0.85	0.7	0.85	0.7	0.65	0.65
NDAP lock	0.5	0.7	0.5	0.2	0.2	0.3	0.3	0.6	0.7	0.4	0.6	0.5	0.6
NDAP post-lock	0.7	0.8	0.7	0.25	0.25	0.35	0.6	0.7	0.75	0.7	0.7	0.65	0.65

| Show Table

DownLoad: CSV

Table 2. Date breakdown by intervention periods for all states and the District of Columbia

Interventions	Information on Intervention Periods by State and the District of Columbia
Interventions	AK	CA	CT	DC	DE	HI	MD	MI	UT	WA
Start date	3/8/20	2/25/20	2/27/20	2/27/20	3/1/20	2/28/20	2/25/20	2/21/20	2/29/20	1/9/20
1st Phase	3/19/20	3/19/20	3/19/20	3/17/20	3/17/20	3/17/20	3/17/20	3/17/20	3/17/20	3/17/20
2nd Phase	5/27/20	5/27/20	5/27/20	5/29/20	5/31/20	5/15/20	5/15/20	5/19/20	5/5/20	5/28/20

| Show Table

DownLoad: CSV

Table 3. Demographic breakdown by race/ethnicity (where data is available) for all states and the District of Columbia. Groups that meet the criteria to be a DAP are in bold. AIAN = American Indian and Alaska Native, NHPI = Native Hawaiian and Pacific Islander

Race/Ethnicity	Percentage of Population per State and the District of Columbia
Race/Ethnicity	AK	CA	CT	DC	DE	HI	MD	MI	UT	WA
AIAN	0.15	0.076	X	X	X	X	X	0.005	0.023	0.01
Asian	0.06	0.14	0.04	0.0435	0.045	0.38	0.06	0.035	0.038	0.08
Black	0.03	0.06	0.1	0.4453	0.22	0.02	0.29	0.14	0.021	0.04
LatinX	X	0.39	0.16	0.113	0.09	X	0.1	X	0.142	0.13
Multi	X	X	0.02	X	0.02	X	X	X	X	0.05
NHPI	0.01	0.039	X	X	X	0.1	X	X	0.016	0.008
Other	0.08	X	0.01	0.01	X	0.24	X	0.03	0.01	0.005
White	0.65	0.37	0.67	0.4196	0.62	0.25	0.51	0.78	0.78	0.69

| Show Table

DownLoad: CSV

Table 5. Results for the CFR using the same $ R_t(n) $ prior for all populations with $ \sigma_{R(t)} = 1.5 $

Population	Prior, $ \sigma=0.05 $	Post, $ \chi^2=112 $	Prior, $ \sigma=0.05 $	Post, $ \chi^2=92 $	Prior, $ \sigma=0.03 $	Post, $ \chi^2=87 $
Asian	0.009	0.0074	0.020	0.0093	0.001	0.0070
Black	0.009	0.0126	0.020	0.0157	0.001	0.0084
LatinX	0.009	0.0213	0.020	0.0246	0.001	0.0147
Multi	0.009	0.0103	0.020	0.0104	0.001	0.0086
Other	0.009	0.0065	0.020	0.0042	0.001	0.0065
White	0.009	0.0229	0.020	0.0271	0.001	0.0134

| Show Table

DownLoad: CSV

Table 6. Results for the CFR using $ R_t(n) $ priors taken from piece-wise assimilation first assuming a CFR of 2% for all populations

Population	Prior, $ \sigma=0.05 $	Post, $ \chi^2=89 $	Prior, $ \sigma=0.002 $	Post, $ \chi^2=41 $
Asian	0.020	0.0282	0.020	0.0199
Black	0.020	0.0337	0.020	0.0199
LatinX	0.020	0.0317	0.020	0.0200
Multi	0.020	0.0316	0.020	0.0202
Other	0.020	0.0299	0.020	0.0204
White	0.020	0.0259	0.020	0.0195

| Show Table

DownLoad: CSV

Table 7. Results for the CFR assuming a CFR of 2% for only the Black Group

Population	Prior, $ \sigma=0.05 $	Post, $ \chi^2=302 $	Prior, $ \sigma=.1x $	Post, $ \chi^2=38 $	Prior, $ \sigma=0.05 $	$ \chi^2=83 $
Asian	0.0090	0.0101	0.0090	0.0091	0.0090	0.0249
Black	0.0200	0.0226	0.0200	0.0201	0.0200	0.0371
LatinX	0.0090	0.0211	0.0090	0.0092	0.0090	0.0299
Multi	0.0090	0.0097	0.0090	0.0092	0.0090	0.0285
Other	0.0090	0.0053	0.0090	0.0093	0.0090	0.0234
White	0.0090	0.0248	0.0090	0.0090	0.0090	0.0256

| Show Table

DownLoad: CSV

Table 4. CFR estimates coming from the run presented in Section 4.2.6

Population	Prior, $ \sigma=0.009 $	Post, $ \chi^2=31 $
Asian	0.0900	0.0089
Black	0.0900	0.0100
LatinX	0.0900	0.0090
Multi	0.0900	0.0093
Other	0.0900	0.0094
White	0.0900	0.0094

| Show Table

DownLoad: CSV

Table 8. The table gives a set of first-guess model parameters. As we could not find scientific estimates of these parameters, we set their values based on available information from the internet and initial model-tuning experiments. We leave it to the data assimilation system to fine-tune the parameter values

Parameter	First guess	Description
$ \tau_ \rm{inc} $	5.5	Incubation period
$ \tau_ \rm{inf} $	3.8	Infection time
$ \tau_ \rm{recm} $	14.0	Recovery time mild cases
$ \tau_ \rm{recs} $	5.0	Recovery time severe cases
$ \tau_ \rm{hosp} $	6.0	Time until hospitalization
$ \tau_ \rm{death} $	16.0	Time until death
$ p_ \mathrm{f} $	0.009	Case fatality rate
$ p_ \mathrm{s} $	0.039	Hospitalization rate (severe cases)
$ p_ \mathrm{h} $	0.4	Fraction of fatally ill going to hospital

| Show Table

DownLoad: CSV

Table 9. The $ p $-numbers indicate the fraction of sick people in an age group ending up with mild symptoms, severe symptoms (hospitalized), and fatal infection

Age group	1	2	3	4	5	6	7	8	9	10	11
Age range	0–5	6–12	13–19	20–29	30–39	40–49	50–59	60–69	70–79	80–89	90–105
Population	351159	451246	446344	711752	730547	723663	703830	582495	435834	185480	45230
p–mild	1.0000	1.0000	0.9998	0.9913	0.9759	0.9686	0.9369	0.9008	0.8465	0.8183	0.8183
p–severe	0.0000	0.0000	0.0002	0.0078	0.0232	0.0295	0.0570	0.0823	0.1160	0.1160	0.1160
p–fatal	0.0000	0.0000	0.0000	0.0009	0.0009	0.0019	0.0061	0.0169	0.0375	0.0656	0.0656

| Show Table

DownLoad: CSV

Related Papers

Cited by

References

[1]	Cyberstates 2020: The definitive guide to the U.S. tech industry and tech wrokforce, URL https://www.cyberstates.org, Last accessed 2021-04-13.
[2]	Disparities in Wealth by Race and Ethnicity in the 2019 Survey of Consumer Finances, URL https://www.federalreserve.gov/econres/notes/feds-notes/disparities-in-wealth-by-race-and-ethnicity-in-the-2019-survey-of-consumer-finances-20200928.htm, Last accessed 2021-04-13.
[3]	Diversity in high tech, URL https://www.eeoc.gov/special-report/diversity-high-tech, Last accessed 2021-04-13.
[4]	Economy at a Glance: California, URL https://data.bls.gov/timeseries/LASST060000000000006?, Last accessed 2021-04-13.
[5]	IHME COVID-19 estimates, URL http://www.healthdata.org/covid/data-downloads, Last accessed 2021-04-13.
[6]	Options to Reduce Quarantine for Contacts of Persons with SARS-CoV-2 Infection Using Symptom Monitoring and Diagnostic Testing, URL https://www.cdc.gov/coronavirus/2019-ncov/more/scientific-brief-options-to-reduce-quarantine.html, Last accessed 2021-04-13.
[7]	Racial Data Dashboard, 2021, URL https://covidtracking.com/race/dashboard, Last accessed 2021-04-13.
[8]	Risk for covid-19 infection, hospitalization, and death by race/ethnicity, URL https://www.cdc.gov/coronavirus/2019-ncov/covid-data/investigations-discovery/hospitalization-death-by-race-ethnicity.html, Last accessed 2021-08-03.
[9]	Rt COVID-19, URL https://rt.live/, Last accessed 2021-04-13.
[10]	Statistics and Church Facts \| Total Church Membership, URL http://newsroom.churchofjesuschrist.org/facts-and-statistics/state/utah, Last accessed 2021-07-31.
[11]	E. Armstrong, M. Runge and J. Gerardin, Identifying the measurements required to estimate rates of COVID-19 transmission, infection, and detection, using variational data assimilation, Infectious Disease Modelling.
[12]	M. Asch, M. Bocquet and M. Nodet, Data Assimilation: Methods, Algorithms, and Applications, SIAM, Society for Industrial and Applied Mathematics, 2016. doi: 10.1137/1.9781611974546.pt1.
[13]	L. M. A. Bettencourt, R. M. Ribeiro, G. Chowell, T. Lant and C. Castillo-Chavez, Towards real time epidemiology: Data assimilation, modeling and anomaly detection of health surveillance data streams, Lecture Notes in Computer Science Intelligence and Security Informatics: Biosurveillance, 79–90. doi: 10.1007/978-3-540-72608-1_8.
[14]	M. Bocquet and P. Sakov, An iterative ensemble Kalman smoother, Quarterly Journal of the Royal Meteorological Society, 140 (2013), 1521-1535. doi: 10.1002/qj.2236.
[15]	A. Carrassi, M. Bocquet, L. Bertino and G. Evensen, Data assimilation in the geosciences: An overview on methods issues and perspectives, WCC, 9 2018. doi: 10.1002/wcc.535.
[16]	A. A. Emerick and A. C. Reynolds, Ensemble smoother with multiple data assimilation, Computers & Geosciences, 55 (2013), 3-15. doi: 10.1016/j.cageo.2012.03.011.
[17]	G. Evensen, Analysis of iterative ensemble smoothers for solving inverse problems, Computational Geosciences, 22 (2018), 885-908. doi: 10.1007/s10596-018-9731-y.
[18]	G. Evensen, J. Amezcua, M. Bocquet, A. Carrassi, A. Farchi, A. Fowler, P. L. Houtekamer, C. K. Jones, R. J. de Moraes, M. Pulido, C. Sampson and F. C. Vossepoel, An international initiative of predicting the SARS-CoV-2 pandemic using ensemble data assimilation, Foundations of Data Science, (2020). doi: 10.3934/fods.2021001.
[19]	J. R. Eyre, S. J. English and M. Forsythe, Assimilation of satellite data in numerical weather prediction. part i: The early years, Quarterly Journal of the Royal Meteorological Society, 146 (2019), 49-68. doi: 10.1002/qj.3654.
[20]	A. L. Garcia-Basteiro, G. Moncunill, M. Tortajada, M. Vidal, C. Guinovart, A. Jiménez, R. Santano, S. Sanz, S. Méndez, A. Llupià, R. Aguilar, S. Alonso, D. Barrios, C. Carolis, P. Cisteró, E. Chóliz, A. Cruz, S. Fochs, C. Jairoce, J. Hecht, M. Lamoglia, M. J. Martínez, R. A. Mitchell, N. Ortega, N. Pey, L. Puyol, M. Ribes, N. Rosell, P. Sotomayor, S. Torres, S. Williams, S. Barroso, A. Vilella, J. Muñoz, A. Trilla, P. Varela, A. Mayor and C. Dobaño, Seroprevalence of antibodies against SARS-CoV-2 among health care workers in a large spanish reference hospital, Nature Communications, 11 (2020), Article number: 3500. doi: 10.1038/s41467-020-17318-x.
[21]	C. G. Grijalva, M. A. Rolfes, Y. Zhu, H. Q. McLean, K. E. Hanson, E. A. Belongia, N. B. Halasa, A. Kim, C. Reed, A. M. Fry and H. K. Talbot, Transmission of SARS-COV-2 infections in households - Tennessee and Wisconsin, April-September 2020, MMWR. Morbidity and Mortality Weekly Report, 69 (2020), 1631–1634. doi: 10.15585/mmwr.mm6944e1.
[22]	P. L. Houtekamer and F. Zhang, Review of the ensemble kalman filter for atmospheric data assimilation, Monthly Weather Review, 144 (2016), 4489-4532. doi: 10.1175/MWR-D-15-0440.1.
[23]	J. P. A. Ioannidis, Infection fatality rate of COVID-19 inferred from seroprevalence data, Bulletin of the World Health Organization, 99 (2020), 19–33F. doi: 10.2471/blt.20.265892.
[24]	J. Jeppesen, Fact sheet: Reanalysis, URL https://www.ecmwf.int/en/about/media-centre/focus/2020/fact-sheet-reanalysis, 2020, Last accessed 2021-07-31.
[25]	E. J. Kostelich, Y. Kuang, J. M. Mcdaniel, N. Z. Moore, N. L. Martirosyan and M. C. Preul, Accurate state estimation from uncertain data and models: An application of data assimilation to mathematical models of human brain tumors, Biology Direct, 6 (2011), 64. doi: 10.1186/1745-6150-6-64.
[26]	W. Lieberman-Cribbin, S. Tuminello, R. M. Flores and E. Taioli, Disparities in COVID-19 testing and positivity in new york city, American Journal of Preventive Medicine, 59 (2020), 326-332. doi: 10.1016/j.amepre.2020.06.005.
[27]	N. Narea, Immigrants have helped keep essential services running. But those without legal status have no financial safety net, URL https://www.vox.com/2020/5/5/21244630/undocumented-immigrants-coronavirus-relief-cares-act, 2020, Last accessed 2021-07-31.
[28]	I. Pathak, Y. Choi, D. Jiao, D. Yeung and L. Liu, Racial-ethnic disparities in case fatality ratio narrowed after age standardization: A call for race-ethnicity-specific age distributions in state covid-19 data, MedRxiv, (2020). doi: 10.1101/2020.10.01.20205377.
[29]	J. Skjervheim, G. Evensen, J. Hove and J. G. Vabø, An ensemble smoother for assisted history matching, SPE, (2011), 141929. doi: 10.2118/141929-MS.
[30]	A. S. Stordal and A. H. Elsheikh, Iterative ensemble smoothers in the annealed importance sampling framework, Advances in Water Resources, 86 (2015), 231-239. doi: 10.1016/j.advwatres.2015.09.030.
[31]	G. Vernieres, A. Anis, R. N. Miller and L. L. Ehret, Generalized inversion of thermistor-chain data and a layer model of lake kinneret, Ocean Modelling, 12 (2006), 112-139. doi: 10.1016/j.ocemod.2005.04.004.
[32]	Z. Wu, T. Phan, J. Baez, Y. Kuang and E. J. Kostelich, Predictability and identifiability assessment of models for prostate cancer under androgen suppression therapy, Mathematical Biosciences and Engineering, 16 (2019), 3512-3536. doi: 10.3934/mbe.2019176.