Coping with Heterogeneity and Uncertainty of COVID-19 ... · COVID-19, CORD-19, WHO COVID-2019, COVID-19 Tweet IDs, US Census, Facebook population density map, Covidtracking.com.

Coping with Heterogeneity and Uncertaintyof COVID-19 Datasets

Issues, Approaches, and Consequences of the COVID-19 Crisis

Ambuj K SinghYu-Xiang Wang

University of CaliforniaSanta Barbara

April 28, 2020

Ambuj Singh, Yu-Xiang Wang (UCSB) COVID-19 seminar April 28, 2020 1 / 22

Seminar series

Previous presentations

(April 14) SARS-CoV2: The Virus and theDisease (Carolina Gonzalez and LynnFitzgibbons)

(April 21) Current Epidemiological Models:Scientific Basis and Evaluation (FrancescoBullo)

Thanks to doctors, nurses, first responders, and everyone in thehospital/healthcare/biomedical ecosystem. And thanks to everyone else foradapting to the new lifestyle.


Lessons from history

(1854) Cholera outbreak in London

(1918–19) Spanish Flu

(2013–16) Western African Ebola virus epidemic

How have data and models informed decision making in past epidemics?


Tracing the origin of a cholera outbreak

Source: Wikipedia

In August of 1854, Soho, a suburb of London, was hit hard by aterrible outbreak of cholera.

Dr. John Snow had long believed that water contaminated by sewagewas the cause of cholera. A challenge to the theories of diseasetransmission then.

Through ‘contact-tracing’ of water (homes, restaurants, coffee shops,breweries), he established that the source was the pump on BroadStreet. Shown is a memorial in his honor.


Interventions in Spanish Flu epidemic of 1918–19

Source: CDC

The most severe pandemic in recent history.

About 500 million people or one-third of the world’s populationbecame infected with this virus. The number of deaths was estimatedto be at least 50 million worldwide with about 675,000 in the US.

Hatchett et al analyzed Non-Pharmaceutical Interventions (NPI) in17 US cities1.

Findings support the hypothesis that rapid implementation of multipleNPIs can significantly reduce disease transmission, but that spreadwill be renewed upon relaxation of such measures.

1Richard J. Hatchett, Carter E. Mecher, and Marc Lipsitch. “Public healthinterventions and epidemic intensity during the 1918 influenza pandemic”. In:Proceedings of the National Academy of Sciences 104.18 (2007), pp. 7582–7587.


Western African Ebola virus epidemic (2013–16)

Source: Wikipedia

CDC model predicted 500K–1.4 million infections. A year later therewere about 25K Ebola cases, including 10.3K deaths. (ScientificAmerican, Dec 8 2014)Prediction was more than 50 times worse. Why?

Errors in deterministic modeling2

Difficult to model behavioral change.Of course, there were many good predictions post-hoc.“It is difficult to make predictions, especially about the future.”

“Ebola defied the prophets of doom. It never went airborne, and itseconomic effects were less painful than expected. Being wrong rarelyfeels this good. But it will be harder to catch the world’s attentionnext time.” Economist, Feb 2015.

2Aaron A. King et al. “Avoidable errors in the modelling of outbreaks of emergingpathogens, with special reference to Ebola”. In: Proc. R. Soc. B. (2015).


Talk outline

Historical context of models and interventions

COVID-19 data

HeterogeneousUncertainDynamic

Measuring the effect of NPIs (Non-Pharmaceutical Interventions)

How do current models use available data?

Contact tracing

Uncertainty and bias in data and models

Case study: fitting SIR model to Santa Barbara Data

Challenges and opportunities


COVID-19 data is heterogeneous, uncertain, and dynamic

Heterogeneous: Risk factors vary at the individual level (presence ofcomorbidity). Different countries (and different regions within acountry) have different outcomes based on geography, culture, andthe human element. Many confounding variables.Uncertain: Testing error of biomedical tests, uncertainty oftransmission fraction, contact rate, recovery rate, ‘asymptomatic’ vs‘presymptomatic’ diagnosis, biased samplingDynamic: Tightening and loosening of interventions, behavioralchanges, disease dynamics at the individual and population level

Observed symptomatic individuals in ahomogeneous population of 450K (SIRmodel), initially with 200 symptomaticand 300 recovered, with γ = 0.2.

More in the presentation by Yu-Xiang


State of COVID-19 disease

US is heterogeneous with respect to the disease.

Disease state can be measured in multiple ways: percent change ininfections, death rate, percent change in positive test rate, percentchange in immunity, ..


Available data and models

Observe and control disease dynamics

High uncertainty

Medium uncertainty

Least certainty

Biomedical

1 Transmissionfraction

2 Incubationperiod

3 Recovery rate

4 Death rate

5 Asymptomaticfraction

6 Immunity tofutureinfections

7 Age effects

8 Seasonal effects

Population data

1 Populationdensity

2 Age distribution

3 # Hospitaladmissions

4 # ICUadmissions

5 # Infections

6 # Recovery

7 # Deaths

8 # Immune

9 # Doubling rate

Contact data

1 Stratifiedcontact rates

2 Humanmobility

3 Highwaytraffic

4 Air traffic

5 Super-spreaders

6 Businessclosures

7 Schoolclosures


Sources of data and knowledge

Data about the epidemic is being made public: nCOv-2019,COVID-19, CORD-19, WHO COVID-2019, COVID-19 Tweet IDs, USCensus, Facebook population density map, Covidtracking.com.

John Hopkins CSSE curates a worldwide dataset with daily numbersdown to the county level3

More in the presentation by Yu-Xiang

Since policy-making depends so much on models, we should alsomake them public.

If more data improve models, so does allowing people to look undertheir bonnets... As well as allowing for expert critique, it is a valuableway of building up public trust. (Economist, April 4, 2020).

Peer review system is under strain

2319 articles (1823 medRxiv, 496 bioRxiv) since January 19; about 24per dayChallenge of getting right information to public quickly

3https://github.com/CSSEGISandData/COVID-19Ambuj Singh, Yu-Xiang Wang (UCSB) COVID-19 seminar April 28, 2020 11 / 22

Measuring the effect of interventions

A number of datasets can be used for measuring the effect of NPIs:

Facebook social connectedness index

Google COVID19 community mobility reports

Streetlightdata (shown above)


Challenge of dealing with uncertainty

After hearing an economist talk about his forecast’s uncertainties andwhy a range of estimates was needed, President Lyndon B. Johnsonreportedly said: “Ranges are for cattle, give me a number.”

“Epidemiology is a science of possibilities and persuasion, not ofcertainties or hard proof” (New Yorker, April 26, 2020)

Models typically rely on multiple uncertain inputs whose interactionsare difficult to analyze.

Behavioral changes are especially hard to anticipate (earlier Ebolaexample).


How do current models use data?

CHIME model4

Institute for Health Metrics and Evaluation model (IHME)5

Imperial College model6

Multi-level models7

4COVID-19 Hospital Impact Model for Epidemics (CHIME).https://penn-chime.phl.io/.

5Christopher Murray. “Forecasting COVID-19 impact on hospital bed-days,ICU-days, ventilator-days and deaths by US state in the next 4 months”. In: medRxiv(2020). doi: 10.1101/2020.03.27.20043752. eprint: https:

//www.medrxiv.org/content/early/2020/03/30/2020.03.27.20043752.full.pdf.url:https://www.medrxiv.org/content/early/2020/03/30/2020.03.27.20043752.

6Neil Ferguson et al. “Report 9: Impact of non-pharmaceutical interventions (NPIs)to reduce COVID19 mortality and healthcare demand”. In: (2020).

7Duygu Balcan et al. “Multiscale mobility networks and the spatial spreading ofinfectious diseases”. In: Proceedings of the National Academy of Sciences 106.51(2009), pp. 21484–21489. doi: 10.1073/pnas.0906910106.


https://penn-chime.phl.io/

https://doi.org/10.1101/2020.03.27.20043752

https://www.medrxiv.org/content/early/2020/03/30/2020.03.27.20043752.full.pdf

https://www.medrxiv.org/content/early/2020/03/30/2020.03.27.20043752.full.pdf

https://www.medrxiv.org/content/early/2020/03/30/2020.03.27.20043752

https://doi.org/10.1073/pnas.0906910106

CHIME model

One of two models endorsed to predict hospital capacity in California.

Based on SIR model

Number of current hospitalizations used to predict the number (andrate) of current infectionsThis and observed doubling rate is used to predict the infection rate.Use empirical estimate of recovery rate

Predict rate of hospital admits, rate of ICU admits, ventilator need

Can change the dynamics by “social distancing,” realized by a percentreduction in infection rate

Assumes homogeneity

Deterministic model


Imperial College model

Influenced the policy of US and UK governments

Stochastic simulation based model based on SIR. Parameters learntfrom observed data.

Age specific variationsA fraction of asymptomatic individuals

Distribute population into age groups

Contact rates: home, work, school, random

Each NPI affects these rates in different ways

Scalability is an issue

Can one learn directly from data?


Institute for Health Metrics and Evaluation (IHME) model

Parameterized matching of time series from Wuhan (and otherlocations that have recovered) based on the number of reporteddeaths and the level of NPI.

Matching adjusted for age distributionConsiders four NPIs: school closures, non-essential business closures,stay-at-home recommendations, and travel restrictions.Three parameters are learned: maximum death rate, inflection point,and a growth parameter.

Spatial resolution does not allow city-level predictions.

Sensitive to reported deaths: can be quite different from actualdeaths8, and can also be affected by the quality of care.

Difficult to model new interventions.

Hard to change predictions quickly (over a few days) and accuracy ofpredictions may not improve as the forecast horizon decreases9.

8“Tracking Covid-19 excess deaths across countries”. In: Economist (Apr. 2020).9Roman Marchant et al. “Learning as We Go: An Examination of the Statistical

Accuracy of COVID19 Daily Death Count Predictions”. In: medRxiv (2020). doi:10.1101/2020.04.11.20062257.


https://doi.org/10.1101/2020.04.11.20062257

Multi-scale mobility models

Hierarchical model most useful at the level of a country or a largeregion10.

Lowest level dynamics based on SIR (or SEIR with an intermediate‘Exposed’ state).

Middle level dynamics based on traffic between cities: learnt based onacutal data or gravitational model.

Top level dynamics based on air traffic.

Successfully modeled the effect of air travel on the transmission fromWuhan11.

Homogeneous model at the lowest level.

10Duygu Balcan et al. “Multiscale mobility networks and the spatial spreading ofinfectious diseases”. In: Proceedings of the National Academy of Sciences 106.51(2009), pp. 21484–21489. doi: 10.1073/pnas.0906910106.

11Matteo Chinazzi et al. “The effect of travel restrictions on the spread of the 2019novel coronavirus (COVID-19) outbreak”. In: Science 368.6489 (2020), pp. 395–400.doi: 10.1126/science.aba9757.


https://doi.org/10.1073/pnas.0906910106

https://doi.org/10.1126/science.aba9757

Contact tracing

A requirement by the CDC for opening up the country12.

Manual implementation: in-depth interviews with those who may beinfected/exposed. Number of public health employees needed:100,000–300,000.

Different combinations of manual and phone-based systems deployedin New Zealand, Taiwan, Singapore (TraceTogether), Australia(COVIDSafe), India (Aarogya Setu), and South Korea.

The number of new cases in New Zealand has dropped to single digits.

Care19 mobile app in North Dakota, South Dakota, Utah.

Plans for extensive phone-based tracing (Google and Apple).

Should contact trace include GPS?Missing data and false positivesPrivacy

12CDC. Contact Tracing : Part of a Multipronged Approach to Fight the COVID-19Pandemic. https://www.cdc.gov/coronavirus/2019-ncov/php/principles-

contact-tracing.html. 2020.Ambuj Singh, Yu-Xiang Wang (UCSB) COVID-19 seminar April 28, 2020 19 / 22

https://www.cdc.gov/coronavirus/2019-ncov/php/principles-contact-tracing.html

https://www.cdc.gov/coronavirus/2019-ncov/php/principles-contact-tracing.html

Can NPIs work on campus?

Only a personal opinion but do seethe opinion piece by Christina Paxson: “College Campuses MustReopen in the Fall. Here’s How We Do It” (NY Times, April 26, 2020).

Automated contact tracing and GPS data by opt-in; manual otherwisePrior estimate of individual’s disease state based on

mobility patterndisease state of contacted individualsdisease state of visited places

Build a contact network of around 35K nodes with attributes ‘at-risk’and ‘disease state’Test individuals on a continuous basis with priority to ‘at-risk’ and‘hub nodes’ in the network

ensure reproduction number < 1 at each nodeBuild capacity for

healthcarealternate instruction plan

Ensure privacy of collected information: More in the presentation byYu-Xiang.


Remainder of the talk

• A closer look at the different types of “uncertainty”

• Case study: fitting SIR model to Santa Barbara data.

• A path towards “re-opening UCSB”

21

Measuring uncertainty with confidence intervals,in the IMHE model, a.k.a., “Murray model”

22

Explanation given for the shaded-region

23

Where are these uncertainties coming from?

• Uncertainty of estimated model parameters• Uncertainty in the natural random process

• Bias in the model itself• Bias from sample selection

• Changes in the distribution• Effects of new interventions

Uncertainty that vanishas datasets get larger

Uncertainty that we could measureby model-checking techniques.

Usually can only gauge in the hindsight

24

The need for localized models

• Different demographics, culture, social norms

• Different industry composition

• Different availability of healthcare

• But we often do not have enough local data• 1% of the cities have 99% percent of the data.

25

The histogram of counties in the US bynumber of cases (as of Apr 25)

Santa Barbara:7 deaths469 cases

New York City:17,126 deaths155,113 cases

26

The number of days since first confirmed case

Santa Barbara:41 days

27

Recall the basic SIR model (three parameters)

�<latexit sha1_base64="g+IRNevPc1g3Nk0ZK/Q3r3J8OUI=">AAAB7HicbZBNS8NAEIYn9avWr6pHL8EieCqJCHosevFYwdRCG8pmO2mXbjZhdyKU0t/gxYMiXv1B3vw3btsctPWFhYd3ZtiZN8qkMOR5305pbX1jc6u8XdnZ3ds/qB4etUyaa44BT2Wq2xEzKIXCgARJbGcaWRJJfIxGt7P64xNqI1L1QOMMw4QNlIgFZ2StoBshsV615tW9udxV8AuoQaFmr/rV7ac8T1ARl8yYju9lFE6YJsElTivd3GDG+IgNsGNRsQRNOJkvO3XPrNN341Tbp8idu78nJiwxZpxEtjNhNDTLtZn5X62TU3wdToTKckLFFx/FuXQpdWeXu32hkZMcW2BcC7ury4dMM042n4oNwV8+eRVaF3Xf8v1lrXFTxFGGEziFc/DhChpwB00IgIOAZ3iFN0c5L86787FoLTnFzDH8kfP5A8PYjqQ=</latexit><latexit sha1_base64="g+IRNevPc1g3Nk0ZK/Q3r3J8OUI=">AAAB7HicbZBNS8NAEIYn9avWr6pHL8EieCqJCHosevFYwdRCG8pmO2mXbjZhdyKU0t/gxYMiXv1B3vw3btsctPWFhYd3ZtiZN8qkMOR5305pbX1jc6u8XdnZ3ds/qB4etUyaa44BT2Wq2xEzKIXCgARJbGcaWRJJfIxGt7P64xNqI1L1QOMMw4QNlIgFZ2StoBshsV615tW9udxV8AuoQaFmr/rV7ac8T1ARl8yYju9lFE6YJsElTivd3GDG+IgNsGNRsQRNOJkvO3XPrNN341Tbp8idu78nJiwxZpxEtjNhNDTLtZn5X62TU3wdToTKckLFFx/FuXQpdWeXu32hkZMcW2BcC7ury4dMM042n4oNwV8+eRVaF3Xf8v1lrXFTxFGGEziFc/DhChpwB00IgIOAZ3iFN0c5L86787FoLTnFzDH8kfP5A8PYjqQ=</latexit><latexit sha1_base64="g+IRNevPc1g3Nk0ZK/Q3r3J8OUI=">AAAB7HicbZBNS8NAEIYn9avWr6pHL8EieCqJCHosevFYwdRCG8pmO2mXbjZhdyKU0t/gxYMiXv1B3vw3btsctPWFhYd3ZtiZN8qkMOR5305pbX1jc6u8XdnZ3ds/qB4etUyaa44BT2Wq2xEzKIXCgARJbGcaWRJJfIxGt7P64xNqI1L1QOMMw4QNlIgFZ2StoBshsV615tW9udxV8AuoQaFmr/rV7ac8T1ARl8yYju9lFE6YJsElTivd3GDG+IgNsGNRsQRNOJkvO3XPrNN341Tbp8idu78nJiwxZpxEtjNhNDTLtZn5X62TU3wdToTKckLFFx/FuXQpdWeXu32hkZMcW2BcC7ury4dMM042n4oNwV8+eRVaF3Xf8v1lrXFTxFGGEziFc/DhChpwB00IgIOAZ3iFN0c5L86787FoLTnFzDH8kfP5A8PYjqQ=</latexit><latexit sha1_base64="g+IRNevPc1g3Nk0ZK/Q3r3J8OUI=">AAAB7HicbZBNS8NAEIYn9avWr6pHL8EieCqJCHosevFYwdRCG8pmO2mXbjZhdyKU0t/gxYMiXv1B3vw3btsctPWFhYd3ZtiZN8qkMOR5305pbX1jc6u8XdnZ3ds/qB4etUyaa44BT2Wq2xEzKIXCgARJbGcaWRJJfIxGt7P64xNqI1L1QOMMw4QNlIgFZ2StoBshsV615tW9udxV8AuoQaFmr/rV7ac8T1ARl8yYju9lFE6YJsElTivd3GDG+IgNsGNRsQRNOJkvO3XPrNN341Tbp8idu78nJiwxZpxEtjNhNDTLtZn5X62TU3wdToTKckLFFx/FuXQpdWeXu32hkZMcW2BcC7ury4dMM042n4oNwV8+eRVaF3Xf8v1lrXFTxFGGEziFc/DhChpwB00IgIOAZ3iFN0c5L86787FoLTnFzDH8kfP5A8PYjqQ=</latexit>

�<latexit sha1_base64="7gxsz35rV55iow5b8W6EqtQy0sk=">AAAB7XicbZBNSwMxEIZn61etX1WPXoJF8FR2RdBj0YvHCvYD2qXMptk2NtksSVYopf/BiwdFvPp/vPlvTNs9aOsLgYd3ZsjMG6WCG+v7315hbX1jc6u4XdrZ3ds/KB8eNY3KNGUNqoTS7QgNEzxhDcutYO1UM5SRYK1odDurt56YNlwlD3acslDiIOExp2id1ewOUErslSt+1Z+LrEKQQwVy1Xvlr25f0UyyxFKBxnQCP7XhBLXlVLBpqZsZliId4YB1HCYomQkn822n5Mw5fRIr7V5iydz9PTFBacxYRq5Toh2a5drM/K/WyWx8HU54kmaWJXTxUZwJYhWZnU76XDNqxdgBUs3droQOUSO1LqCSCyFYPnkVmhfVwPH9ZaV2k8dRhBM4hXMI4ApqcAd1aACFR3iGV3jzlPfivXsfi9aCl88cwx95nz+G7Y8V</latexit><latexit sha1_base64="7gxsz35rV55iow5b8W6EqtQy0sk=">AAAB7XicbZBNSwMxEIZn61etX1WPXoJF8FR2RdBj0YvHCvYD2qXMptk2NtksSVYopf/BiwdFvPp/vPlvTNs9aOsLgYd3ZsjMG6WCG+v7315hbX1jc6u4XdrZ3ds/KB8eNY3KNGUNqoTS7QgNEzxhDcutYO1UM5SRYK1odDurt56YNlwlD3acslDiIOExp2id1ewOUErslSt+1Z+LrEKQQwVy1Xvlr25f0UyyxFKBxnQCP7XhBLXlVLBpqZsZliId4YB1HCYomQkn822n5Mw5fRIr7V5iydz9PTFBacxYRq5Toh2a5drM/K/WyWx8HU54kmaWJXTxUZwJYhWZnU76XDNqxdgBUs3droQOUSO1LqCSCyFYPnkVmhfVwPH9ZaV2k8dRhBM4hXMI4ApqcAd1aACFR3iGV3jzlPfivXsfi9aCl88cwx95nz+G7Y8V</latexit><latexit sha1_base64="7gxsz35rV55iow5b8W6EqtQy0sk=">AAAB7XicbZBNSwMxEIZn61etX1WPXoJF8FR2RdBj0YvHCvYD2qXMptk2NtksSVYopf/BiwdFvPp/vPlvTNs9aOsLgYd3ZsjMG6WCG+v7315hbX1jc6u4XdrZ3ds/KB8eNY3KNGUNqoTS7QgNEzxhDcutYO1UM5SRYK1odDurt56YNlwlD3acslDiIOExp2id1ewOUErslSt+1Z+LrEKQQwVy1Xvlr25f0UyyxFKBxnQCP7XhBLXlVLBpqZsZliId4YB1HCYomQkn822n5Mw5fRIr7V5iydz9PTFBacxYRq5Toh2a5drM/K/WyWx8HU54kmaWJXTxUZwJYhWZnU76XDNqxdgBUs3droQOUSO1LqCSCyFYPnkVmhfVwPH9ZaV2k8dRhBM4hXMI4ApqcAd1aACFR3iGV3jzlPfivXsfi9aCl88cwx95nz+G7Y8V</latexit><latexit sha1_base64="7gxsz35rV55iow5b8W6EqtQy0sk=">AAAB7XicbZBNSwMxEIZn61etX1WPXoJF8FR2RdBj0YvHCvYD2qXMptk2NtksSVYopf/BiwdFvPp/vPlvTNs9aOsLgYd3ZsjMG6WCG+v7315hbX1jc6u4XdrZ3ds/KB8eNY3KNGUNqoTS7QgNEzxhDcutYO1UM5SRYK1odDurt56YNlwlD3acslDiIOExp2id1ewOUErslSt+1Z+LrEKQQwVy1Xvlr25f0UyyxFKBxnQCP7XhBLXlVLBpqZsZliId4YB1HCYomQkn822n5Mw5fRIr7V5iydz9PTFBacxYRq5Toh2a5drM/K/WyWx8HU54kmaWJXTxUZwJYhWZnU76XDNqxdgBUs3droQOUSO1LqCSCyFYPnkVmhfVwPH9ZaV2k8dRhBM4hXMI4ApqcAd1aACFR3iGV3jzlPfivXsfi9aCl88cwx95nz+G7Y8V</latexit>

Expected number of people one patient can affect per day.

Percentage of infectious people recovering (or dying) per day.

+ Z1<latexit sha1_base64="pW53VK0gyeWO3PCYE5WeRL9iyw0=">AAAB8nicdZDLSgMxFIYzXmu9VV26CRZBEEpSxLa7ohuXFewFp0PJZDJtaGYyJhmhDH0MNy4UcevTuPNtzLQVVPSHwMf/n0POOX4iuDYIfThLyyura+uFjeLm1vbObmlvv6NlqihrUymk6vlEM8Fj1jbcCNZLFCORL1jXH1/mefeeKc1lfGMmCfMiMox5yCkx1nJPYf8uJQG8HeBBqYwqCCGMMcwB186RhUajXsV1iPPIqgwWag1K7/1A0jRisaGCaO1ilBgvI8pwKti02E81SwgdkyFzLcYkYtrLZiNP4bF1AhhKZV9s4Mz93pGRSOtJ5NvKiJiR/p3l5l+Zm5qw7mU8TlLDYjr/KEwFNBLm+8OAK0aNmFggVHE7K6Qjogg19kpFe4SvTeH/0KlWsOXrs3LzYnGOAjgER+AEYFADTXAFWqANKJDgATyBZ8c4j86L8zovXXIWPQfgh5y3TxbXkHo=</latexit><latexit sha1_base64="pW53VK0gyeWO3PCYE5WeRL9iyw0=">AAAB8nicdZDLSgMxFIYzXmu9VV26CRZBEEpSxLa7ohuXFewFp0PJZDJtaGYyJhmhDH0MNy4UcevTuPNtzLQVVPSHwMf/n0POOX4iuDYIfThLyyura+uFjeLm1vbObmlvv6NlqihrUymk6vlEM8Fj1jbcCNZLFCORL1jXH1/mefeeKc1lfGMmCfMiMox5yCkx1nJPYf8uJQG8HeBBqYwqCCGMMcwB186RhUajXsV1iPPIqgwWag1K7/1A0jRisaGCaO1ilBgvI8pwKti02E81SwgdkyFzLcYkYtrLZiNP4bF1AhhKZV9s4Mz93pGRSOtJ5NvKiJiR/p3l5l+Zm5qw7mU8TlLDYjr/KEwFNBLm+8OAK0aNmFggVHE7K6Qjogg19kpFe4SvTeH/0KlWsOXrs3LzYnGOAjgER+AEYFADTXAFWqANKJDgATyBZ8c4j86L8zovXXIWPQfgh5y3TxbXkHo=</latexit><latexit sha1_base64="pW53VK0gyeWO3PCYE5WeRL9iyw0=">AAAB8nicdZDLSgMxFIYzXmu9VV26CRZBEEpSxLa7ohuXFewFp0PJZDJtaGYyJhmhDH0MNy4UcevTuPNtzLQVVPSHwMf/n0POOX4iuDYIfThLyyura+uFjeLm1vbObmlvv6NlqihrUymk6vlEM8Fj1jbcCNZLFCORL1jXH1/mefeeKc1lfGMmCfMiMox5yCkx1nJPYf8uJQG8HeBBqYwqCCGMMcwB186RhUajXsV1iPPIqgwWag1K7/1A0jRisaGCaO1ilBgvI8pwKti02E81SwgdkyFzLcYkYtrLZiNP4bF1AhhKZV9s4Mz93pGRSOtJ5NvKiJiR/p3l5l+Zm5qw7mU8TlLDYjr/KEwFNBLm+8OAK0aNmFggVHE7K6Qjogg19kpFe4SvTeH/0KlWsOXrs3LzYnGOAjgER+AEYFADTXAFWqANKJDgATyBZ8c4j86L8zovXXIWPQfgh5y3TxbXkHo=</latexit><latexit sha1_base64="pW53VK0gyeWO3PCYE5WeRL9iyw0=">AAAB8nicdZDLSgMxFIYzXmu9VV26CRZBEEpSxLa7ohuXFewFp0PJZDJtaGYyJhmhDH0MNy4UcevTuPNtzLQVVPSHwMf/n0POOX4iuDYIfThLyyura+uFjeLm1vbObmlvv6NlqihrUymk6vlEM8Fj1jbcCNZLFCORL1jXH1/mefeeKc1lfGMmCfMiMox5yCkx1nJPYf8uJQG8HeBBqYwqCCGMMcwB186RhUajXsV1iPPIqgwWag1K7/1A0jRisaGCaO1ilBgvI8pwKti02E81SwgdkyFzLcYkYtrLZiNP4bF1AhhKZV9s4Mz93pGRSOtJ5NvKiJiR/p3l5l+Zm5qw7mU8TlLDYjr/KEwFNBLm+8OAK0aNmFggVHE7K6Qjogg19kpFe4SvTeH/0KlWsOXrs3LzYnGOAjgER+AEYFADTXAFWqANKJDgATyBZ8c4j86L8zovXXIWPQfgh5y3TxbXkHo=</latexit>

+ Z2<latexit sha1_base64="DJmbNDLX/GtjED/I3XzSYF7f8uc=">AAAB8nicdZDLSgMxFIYzXmu9VV26CRZBEEpSxLa7ohuXFewFp0PJZDJtaCYzJhmhDH0MNy4UcevTuPNtzLQVVPSHwMf/n0POOX4iuDYIfThLyyura+uFjeLm1vbObmlvv6PjVFHWprGIVc8nmgkuWdtwI1gvUYxEvmBdf3yZ5917pjSP5Y2ZJMyLyFDykFNirOWewv5dSgJ4O6gOSmVUQQhhjGEOuHaOLDQa9SquQ5xHVmWwUGtQeu8HMU0jJg0VRGsXo8R4GVGGU8GmxX6qWULomAyZa1GSiGkvm408hcfWCWAYK/ukgTP3e0dGIq0nkW8rI2JG+neWm39lbmrCupdxmaSGSTr/KEwFNDHM94cBV4waMbFAqOJ2VkhHRBFq7JWK9ghfm8L/oVOtYMvXZ+XmxeIcBXAIjsAJwKAGmuAKtEAbUBCDB/AEnh3jPDovzuu8dMlZ9ByAH3LePgEYW5B7</latexit><latexit sha1_base64="DJmbNDLX/GtjED/I3XzSYF7f8uc=">AAAB8nicdZDLSgMxFIYzXmu9VV26CRZBEEpSxLa7ohuXFewFp0PJZDJtaCYzJhmhDH0MNy4UcevTuPNtzLQVVPSHwMf/n0POOX4iuDYIfThLyyura+uFjeLm1vbObmlvv6PjVFHWprGIVc8nmgkuWdtwI1gvUYxEvmBdf3yZ5917pjSP5Y2ZJMyLyFDykFNirOWewv5dSgJ4O6gOSmVUQQhhjGEOuHaOLDQa9SquQ5xHVmWwUGtQeu8HMU0jJg0VRGsXo8R4GVGGU8GmxX6qWULomAyZa1GSiGkvm408hcfWCWAYK/ukgTP3e0dGIq0nkW8rI2JG+neWm39lbmrCupdxmaSGSTr/KEwFNDHM94cBV4waMbFAqOJ2VkhHRBFq7JWK9ghfm8L/oVOtYMvXZ+XmxeIcBXAIjsAJwKAGmuAKtEAbUBCDB/AEnh3jPDovzuu8dMlZ9ByAH3LePgEYW5B7</latexit><latexit sha1_base64="DJmbNDLX/GtjED/I3XzSYF7f8uc=">AAAB8nicdZDLSgMxFIYzXmu9VV26CRZBEEpSxLa7ohuXFewFp0PJZDJtaCYzJhmhDH0MNy4UcevTuPNtzLQVVPSHwMf/n0POOX4iuDYIfThLyyura+uFjeLm1vbObmlvv6PjVFHWprGIVc8nmgkuWdtwI1gvUYxEvmBdf3yZ5917pjSP5Y2ZJMyLyFDykFNirOWewv5dSgJ4O6gOSmVUQQhhjGEOuHaOLDQa9SquQ5xHVmWwUGtQeu8HMU0jJg0VRGsXo8R4GVGGU8GmxX6qWULomAyZa1GSiGkvm408hcfWCWAYK/ukgTP3e0dGIq0nkW8rI2JG+neWm39lbmrCupdxmaSGSTr/KEwFNDHM94cBV4waMbFAqOJ2VkhHRBFq7JWK9ghfm8L/oVOtYMvXZ+XmxeIcBXAIjsAJwKAGmuAKtEAbUBCDB/AEnh3jPDovzuu8dMlZ9ByAH3LePgEYW5B7</latexit><latexit sha1_base64="DJmbNDLX/GtjED/I3XzSYF7f8uc=">AAAB8nicdZDLSgMxFIYzXmu9VV26CRZBEEpSxLa7ohuXFewFp0PJZDJtaCYzJhmhDH0MNy4UcevTuPNtzLQVVPSHwMf/n0POOX4iuDYIfThLyyura+uFjeLm1vbObmlvv6PjVFHWprGIVc8nmgkuWdtwI1gvUYxEvmBdf3yZ5917pjSP5Y2ZJMyLyFDykFNirOWewv5dSgJ4O6gOSmVUQQhhjGEOuHaOLDQa9SquQ5xHVmWwUGtQeu8HMU0jJg0VRGsXo8R4GVGGU8GmxX6qWULomAyZa1GSiGkvm408hcfWCWAYK/ukgTP3e0dGIq0nkW8rI2JG+neWm39lbmrCupdxmaSGSTr/KEwFNDHM94cBV4waMbFAqOJ2VkhHRBFq7JWK9ghfm8L/oVOtYMvXZ+XmxeIcBXAIjsAJwKAGmuAKtEAbUBCDB/AEnh3jPDovzuu8dMlZ9ByAH3LePgEYW5B7</latexit>

We added Z_1, Z_2 to model theuncertainty not captured by the model.�(Z1 + Z2)

<latexit sha1_base64="bl1fmrp7WVEtwwceqkyNnqkrqGM=">AAAB9HicdZDLSgMxFIYz9VbrrerSTbAIFbEkRWy7K7pxWcFeaDsMmTTThmYuJplCGfocblwo4taHcefbmGkrqOiBwMf/n8M5+d1IcKUR+rAyK6tr6xvZzdzW9s7uXn7/oKXCWFLWpKEIZccligkesKbmWrBOJBnxXcHa7vg69dsTJhUPgzs9jZjtk2HAPU6JNpJ9DotdB8Mz2HXKp06+gEoIIYwxTAFXLpGBWq1axlWIU8tUASyr4eTf+4OQxj4LNBVEqR5GkbYTIjWngs1y/VixiNAxGbKewYD4TNnJ/OgZPDHKAHqhNC/QcK5+n0iIr9TUd02nT/RI/fZS8S+vF2uvaic8iGLNArpY5MUC6hCmCcABl4xqMTVAqOTmVkhHRBKqTU45E8LXT+H/0CqXsOHbi0L9ahlHFhyBY1AEGFRAHdyABmgCCu7BA3gCz9bEerRerNdFa8ZazhyCH2W9fQKh8ZAQ</latexit><latexit sha1_base64="bl1fmrp7WVEtwwceqkyNnqkrqGM=">AAAB9HicdZDLSgMxFIYz9VbrrerSTbAIFbEkRWy7K7pxWcFeaDsMmTTThmYuJplCGfocblwo4taHcefbmGkrqOiBwMf/n8M5+d1IcKUR+rAyK6tr6xvZzdzW9s7uXn7/oKXCWFLWpKEIZccligkesKbmWrBOJBnxXcHa7vg69dsTJhUPgzs9jZjtk2HAPU6JNpJ9DotdB8Mz2HXKp06+gEoIIYwxTAFXLpGBWq1axlWIU8tUASyr4eTf+4OQxj4LNBVEqR5GkbYTIjWngs1y/VixiNAxGbKewYD4TNnJ/OgZPDHKAHqhNC/QcK5+n0iIr9TUd02nT/RI/fZS8S+vF2uvaic8iGLNArpY5MUC6hCmCcABl4xqMTVAqOTmVkhHRBKqTU45E8LXT+H/0CqXsOHbi0L9ahlHFhyBY1AEGFRAHdyABmgCCu7BA3gCz9bEerRerNdFa8ZazhyCH2W9fQKh8ZAQ</latexit><latexit sha1_base64="bl1fmrp7WVEtwwceqkyNnqkrqGM=">AAAB9HicdZDLSgMxFIYz9VbrrerSTbAIFbEkRWy7K7pxWcFeaDsMmTTThmYuJplCGfocblwo4taHcefbmGkrqOiBwMf/n8M5+d1IcKUR+rAyK6tr6xvZzdzW9s7uXn7/oKXCWFLWpKEIZccligkesKbmWrBOJBnxXcHa7vg69dsTJhUPgzs9jZjtk2HAPU6JNpJ9DotdB8Mz2HXKp06+gEoIIYwxTAFXLpGBWq1axlWIU8tUASyr4eTf+4OQxj4LNBVEqR5GkbYTIjWngs1y/VixiNAxGbKewYD4TNnJ/OgZPDHKAHqhNC/QcK5+n0iIr9TUd02nT/RI/fZS8S+vF2uvaic8iGLNArpY5MUC6hCmCcABl4xqMTVAqOTmVkhHRBKqTU45E8LXT+H/0CqXsOHbi0L9ahlHFhyBY1AEGFRAHdyABmgCCu7BA3gCz9bEerRerNdFa8ZazhyCH2W9fQKh8ZAQ</latexit><latexit sha1_base64="bl1fmrp7WVEtwwceqkyNnqkrqGM=">AAAB9HicdZDLSgMxFIYz9VbrrerSTbAIFbEkRWy7K7pxWcFeaDsMmTTThmYuJplCGfocblwo4taHcefbmGkrqOiBwMf/n8M5+d1IcKUR+rAyK6tr6xvZzdzW9s7uXn7/oKXCWFLWpKEIZccligkesKbmWrBOJBnxXcHa7vg69dsTJhUPgzs9jZjtk2HAPU6JNpJ9DotdB8Mz2HXKp06+gEoIIYwxTAFXLpGBWq1axlWIU8tUASyr4eTf+4OQxj4LNBVEqR5GkbYTIjWngs1y/VixiNAxGbKewYD4TNnJ/OgZPDHKAHqhNC/QcK5+n0iIr9TUd02nT/RI/fZS8S+vF2uvaic8iGLNArpY5MUC6hCmCcABl4xqMTVAqOTmVkhHRBKqTU45E8LXT+H/0CqXsOHbi0L9ahlHFhyBY1AEGFRAHdyABmgCCu7BA3gCz9bEerRerNdFa8ZazhyCH2W9fQKh8ZAQ</latexit>

N<latexit sha1_base64="RFqBpS+K/+GdhyPN/lvDImg+NFI=">AAAB6HicdVDJSgNBEO2JW4xb1KOXxiB4GnpiHJNb0IsnScAskAyhp1OTtOlZ6O4RQsgXePGgiFc/yZt/Y2cRVPRBweO9Kqrq+YngShPyYWVWVtfWN7Kbua3tnd29/P5BU8WpZNBgsYhl26cKBI+gobkW0E4k0NAX0PJHVzO/dQ9S8Ti61eMEvJAOIh5wRrWR6je9fIHYlQopOS4m9jkhRbdiCDkrll0XOzaZo4CWqPXy791+zNIQIs0EVarjkER7Eyo1ZwKmuW6qIKFsRAfQMTSiIShvMj90ik+M0sdBLE1FGs/V7xMTGio1Dn3TGVI9VL+9mfiX10l1UPYmPEpSDRFbLApSgXWMZ1/jPpfAtBgbQpnk5lbMhlRSpk02ORPC16f4f9Is2o7h9VKhermMI4uO0DE6RQ66QFV0jWqogRgC9ICe0LN1Zz1aL9brojVjLWcO0Q9Yb58Uq40e</latexit><latexit sha1_base64="RFqBpS+K/+GdhyPN/lvDImg+NFI=">AAAB6HicdVDJSgNBEO2JW4xb1KOXxiB4GnpiHJNb0IsnScAskAyhp1OTtOlZ6O4RQsgXePGgiFc/yZt/Y2cRVPRBweO9Kqrq+YngShPyYWVWVtfWN7Kbua3tnd29/P5BU8WpZNBgsYhl26cKBI+gobkW0E4k0NAX0PJHVzO/dQ9S8Ti61eMEvJAOIh5wRrWR6je9fIHYlQopOS4m9jkhRbdiCDkrll0XOzaZo4CWqPXy791+zNIQIs0EVarjkER7Eyo1ZwKmuW6qIKFsRAfQMTSiIShvMj90ik+M0sdBLE1FGs/V7xMTGio1Dn3TGVI9VL+9mfiX10l1UPYmPEpSDRFbLApSgXWMZ1/jPpfAtBgbQpnk5lbMhlRSpk02ORPC16f4f9Is2o7h9VKhermMI4uO0DE6RQ66QFV0jWqogRgC9ICe0LN1Zz1aL9brojVjLWcO0Q9Yb58Uq40e</latexit><latexit sha1_base64="RFqBpS+K/+GdhyPN/lvDImg+NFI=">AAAB6HicdVDJSgNBEO2JW4xb1KOXxiB4GnpiHJNb0IsnScAskAyhp1OTtOlZ6O4RQsgXePGgiFc/yZt/Y2cRVPRBweO9Kqrq+YngShPyYWVWVtfWN7Kbua3tnd29/P5BU8WpZNBgsYhl26cKBI+gobkW0E4k0NAX0PJHVzO/dQ9S8Ti61eMEvJAOIh5wRrWR6je9fIHYlQopOS4m9jkhRbdiCDkrll0XOzaZo4CWqPXy791+zNIQIs0EVarjkER7Eyo1ZwKmuW6qIKFsRAfQMTSiIShvMj90ik+M0sdBLE1FGs/V7xMTGio1Dn3TGVI9VL+9mfiX10l1UPYmPEpSDRFbLApSgXWMZ1/jPpfAtBgbQpnk5lbMhlRSpk02ORPC16f4f9Is2o7h9VKhermMI4uO0DE6RQ66QFV0jWqogRgC9ICe0LN1Zz1aL9brojVjLWcO0Q9Yb58Uq40e</latexit><latexit sha1_base64="RFqBpS+K/+GdhyPN/lvDImg+NFI=">AAAB6HicdVDJSgNBEO2JW4xb1KOXxiB4GnpiHJNb0IsnScAskAyhp1OTtOlZ6O4RQsgXePGgiFc/yZt/Y2cRVPRBweO9Kqrq+YngShPyYWVWVtfWN7Kbua3tnd29/P5BU8WpZNBgsYhl26cKBI+gobkW0E4k0NAX0PJHVzO/dQ9S8Ti61eMEvJAOIh5wRrWR6je9fIHYlQopOS4m9jkhRbdiCDkrll0XOzaZo4CWqPXy791+zNIQIs0EVarjkER7Eyo1ZwKmuW6qIKFsRAfQMTSiIShvMj90ik+M0sdBLE1FGs/V7xMTGio1Dn3TGVI9VL+9mfiX10l1UPYmPEpSDRFbLApSgXWMZ1/jPpfAtBgbQpnk5lbMhlRSpk02ORPC16f4f9Is2o7h9VKhermMI4uO0DE6RQ66QFV0jWqogRgC9ICe0LN1Zz1aL9brojVjLWcO0Q9Yb58Uq40e</latexit>

Size of the susceptible population.28

A (very rough) sketch calculation of theuncertainty… assume the model is correct.• Borrow γ parameter from Italy model.

• Fitting Santa Barbara data to the deterministic / stochastic SIR model:

• Least square fit

• Maximum Likelihood � 2 [0.076, 0.375]<latexit sha1_base64="oo+75CPU3dwwllKyAd85FBTgysI=">AAACAnicdZDLSgMxFIYzXmu9VV2Jm2ARXEhJqvayK7pxWcFeoDOUTJppQzOZIckIpRQ3voobF4q49Snc+TZm2goqeiDw8f/ncHJ+PxZcG4Q+nIXFpeWV1cxadn1jc2s7t7Pb1FGiKGvQSESq7RPNBJesYbgRrB0rRkJfsJY/vEz91i1Tmkfyxoxi5oWkL3nAKTFW6ub2XZ8ZAl0uYQcVULl0AiEqnJbPvW4ubwWEMMYwBVwuIQvVaqWIKxCnlq08mFe9m3t3exFNQiYNFUTrDkax8cZEGU4Fm2TdRLOY0CHps45FSUKmvfH0hAk8skoPBpGyTxo4Vb9PjEmo9Sj0bWdIzED/9lLxL6+TmKDijbmME8MknS0KEgFNBNM8YI8rRo0YWSBUcftXSAdEEWpsalkbwtel8H9oFgvY8vVZvnYxjyMDDsAhOAYYlEENXIE6aAAK7sADeALPzr3z6Lw4r7PWBWc+swd+lPP2CayJlGI=</latexit><latexit sha1_base64="oo+75CPU3dwwllKyAd85FBTgysI=">AAACAnicdZDLSgMxFIYzXmu9VV2Jm2ARXEhJqvayK7pxWcFeoDOUTJppQzOZIckIpRQ3voobF4q49Snc+TZm2goqeiDw8f/ncHJ+PxZcG4Q+nIXFpeWV1cxadn1jc2s7t7Pb1FGiKGvQSESq7RPNBJesYbgRrB0rRkJfsJY/vEz91i1Tmkfyxoxi5oWkL3nAKTFW6ub2XZ8ZAl0uYQcVULl0AiEqnJbPvW4ubwWEMMYwBVwuIQvVaqWIKxCnlq08mFe9m3t3exFNQiYNFUTrDkax8cZEGU4Fm2TdRLOY0CHps45FSUKmvfH0hAk8skoPBpGyTxo4Vb9PjEmo9Sj0bWdIzED/9lLxL6+TmKDijbmME8MknS0KEgFNBNM8YI8rRo0YWSBUcftXSAdEEWpsalkbwtel8H9oFgvY8vVZvnYxjyMDDsAhOAYYlEENXIE6aAAK7sADeALPzr3z6Lw4r7PWBWc+swd+lPP2CayJlGI=</latexit><latexit sha1_base64="oo+75CPU3dwwllKyAd85FBTgysI=">AAACAnicdZDLSgMxFIYzXmu9VV2Jm2ARXEhJqvayK7pxWcFeoDOUTJppQzOZIckIpRQ3voobF4q49Snc+TZm2goqeiDw8f/ncHJ+PxZcG4Q+nIXFpeWV1cxadn1jc2s7t7Pb1FGiKGvQSESq7RPNBJesYbgRrB0rRkJfsJY/vEz91i1Tmkfyxoxi5oWkL3nAKTFW6ub2XZ8ZAl0uYQcVULl0AiEqnJbPvW4ubwWEMMYwBVwuIQvVaqWIKxCnlq08mFe9m3t3exFNQiYNFUTrDkax8cZEGU4Fm2TdRLOY0CHps45FSUKmvfH0hAk8skoPBpGyTxo4Vb9PjEmo9Sj0bWdIzED/9lLxL6+TmKDijbmME8MknS0KEgFNBNM8YI8rRo0YWSBUcftXSAdEEWpsalkbwtel8H9oFgvY8vVZvnYxjyMDDsAhOAYYlEENXIE6aAAK7sADeALPzr3z6Lw4r7PWBWc+swd+lPP2CayJlGI=</latexit><latexit sha1_base64="oo+75CPU3dwwllKyAd85FBTgysI=">AAACAnicdZDLSgMxFIYzXmu9VV2Jm2ARXEhJqvayK7pxWcFeoDOUTJppQzOZIckIpRQ3voobF4q49Snc+TZm2goqeiDw8f/ncHJ+PxZcG4Q+nIXFpeWV1cxadn1jc2s7t7Pb1FGiKGvQSESq7RPNBJesYbgRrB0rRkJfsJY/vEz91i1Tmkfyxoxi5oWkL3nAKTFW6ub2XZ8ZAl0uYQcVULl0AiEqnJbPvW4ubwWEMMYwBVwuIQvVaqWIKxCnlq08mFe9m3t3exFNQiYNFUTrDkax8cZEGU4Fm2TdRLOY0CHps45FSUKmvfH0hAk8skoPBpGyTxo4Vb9PjEmo9Sj0bWdIzED/9lLxL6+TmKDijbmME8MknS0KEgFNBNM8YI8rRo0YWSBUcftXSAdEEWpsalkbwtel8H9oFgvY8vVZvnYxjyMDDsAhOAYYlEENXIE6aAAK7sADeALPzr3z6Lw4r7PWBWc+swd+lPP2CayJlGI=</latexit>

� 2 [0.276, 0.299]<latexit sha1_base64="s28cCxyPlg+NAlZMucqTckorodM=">AAACAnicdVDLSgMxFM3UV62vUVfiJlgEF1IyY512dkU3LivYB7RDyaRpG5rJDElGKKW48VfcuFDErV/hzr8xfQgqeuByD+fcS3JPmHCmNEIfVmZpeWV1Lbue29jc2t6xd/fqKk4loTUS81g2Q6woZ4LWNNOcNhNJcRRy2giHl1O/cUulYrG40aOEBhHuC9ZjBGsjdeyDdkg1hm0mYAsV3JJ3CqHpvh907Dwq+D4qOp5RzhFyPd8QdOaWPQ86BTRDHixQ7djv7W5M0ogKTThWquWgRAdjLDUjnE5y7VTRBJMh7tOWoQJHVAXj2QkTeGyULuzF0pTQcKZ+3xjjSKlRFJrJCOuB+u1Nxb+8Vqp75WDMRJJqKsj8oV7KoY7hNA/YZZISzUeGYCKZ+SskAywx0Sa1nAnh61L4P6m7Bcfw62K+crGIIwsOwRE4AQ4ogQq4AlVQAwTcgQfwBJ6te+vRerFe56MZa7GzD37AevsE2FOUfw==</latexit><latexit sha1_base64="s28cCxyPlg+NAlZMucqTckorodM=">AAACAnicdVDLSgMxFM3UV62vUVfiJlgEF1IyY512dkU3LivYB7RDyaRpG5rJDElGKKW48VfcuFDErV/hzr8xfQgqeuByD+fcS3JPmHCmNEIfVmZpeWV1Lbue29jc2t6xd/fqKk4loTUS81g2Q6woZ4LWNNOcNhNJcRRy2giHl1O/cUulYrG40aOEBhHuC9ZjBGsjdeyDdkg1hm0mYAsV3JJ3CqHpvh907Dwq+D4qOp5RzhFyPd8QdOaWPQ86BTRDHixQ7djv7W5M0ogKTThWquWgRAdjLDUjnE5y7VTRBJMh7tOWoQJHVAXj2QkTeGyULuzF0pTQcKZ+3xjjSKlRFJrJCOuB+u1Nxb+8Vqp75WDMRJJqKsj8oV7KoY7hNA/YZZISzUeGYCKZ+SskAywx0Sa1nAnh61L4P6m7Bcfw62K+crGIIwsOwRE4AQ4ogQq4AlVQAwTcgQfwBJ6te+vRerFe56MZa7GzD37AevsE2FOUfw==</latexit><latexit sha1_base64="s28cCxyPlg+NAlZMucqTckorodM=">AAACAnicdVDLSgMxFM3UV62vUVfiJlgEF1IyY512dkU3LivYB7RDyaRpG5rJDElGKKW48VfcuFDErV/hzr8xfQgqeuByD+fcS3JPmHCmNEIfVmZpeWV1Lbue29jc2t6xd/fqKk4loTUS81g2Q6woZ4LWNNOcNhNJcRRy2giHl1O/cUulYrG40aOEBhHuC9ZjBGsjdeyDdkg1hm0mYAsV3JJ3CqHpvh907Dwq+D4qOp5RzhFyPd8QdOaWPQ86BTRDHixQ7djv7W5M0ogKTThWquWgRAdjLDUjnE5y7VTRBJMh7tOWoQJHVAXj2QkTeGyULuzF0pTQcKZ+3xjjSKlRFJrJCOuB+u1Nxb+8Vqp75WDMRJJqKsj8oV7KoY7hNA/YZZISzUeGYCKZ+SskAywx0Sa1nAnh61L4P6m7Bcfw62K+crGIIwsOwRE4AQ4ogQq4AlVQAwTcgQfwBJ6te+vRerFe56MZa7GzD37AevsE2FOUfw==</latexit><latexit sha1_base64="s28cCxyPlg+NAlZMucqTckorodM=">AAACAnicdVDLSgMxFM3UV62vUVfiJlgEF1IyY512dkU3LivYB7RDyaRpG5rJDElGKKW48VfcuFDErV/hzr8xfQgqeuByD+fcS3JPmHCmNEIfVmZpeWV1Lbue29jc2t6xd/fqKk4loTUS81g2Q6woZ4LWNNOcNhNJcRRy2giHl1O/cUulYrG40aOEBhHuC9ZjBGsjdeyDdkg1hm0mYAsV3JJ3CqHpvh907Dwq+D4qOp5RzhFyPd8QdOaWPQ86BTRDHixQ7djv7W5M0ogKTThWquWgRAdjLDUjnE5y7VTRBJMh7tOWoQJHVAXj2QkTeGyULuzF0pTQcKZ+3xjjSKlRFJrJCOuB+u1Nxb+8Vqp75WDMRJJqKsj8oV7KoY7hNA/YZZISzUeGYCKZ+SskAywx0Sa1nAnh61L4P6m7Bcfw62K+crGIIwsOwRE4AQ4ogQq4AlVQAwTcgQfwBJ6te+vRerFe56MZa7GzD37AevsE2FOUfw==</latexit>

29

N 2 [467, 499]<latexit sha1_base64="nKqnCwuNPgmh683lgSbQnCAD9cY=">AAAB+nicdVBNSwMxEM3Wr1q/Wj16CRbBg5RsrdvurejFk1SwrbBdSjZN29BsdkmySqn9KV48KOLVX+LNf2O2raCiDwYe780wMy+IOVMaoQ8rs7S8srqWXc9tbG5t7+QLuy0VJZLQJol4JG8CrChngjY105zexJLiMOC0HYzOU799S6VikbjW45j6IR4I1mcEayN184VL2GECehWnegwrruvnuvkiKrkuqtgORKVThMqOawg6KdccB9olNEMRLNDo5t87vYgkIRWacKyUZ6NY+xMsNSOcTnOdRNEYkxEeUM9QgUOq/Mns9Ck8NEoP9iNpSmg4U79PTHCo1DgMTGeI9VD99lLxL89LdL/mT5iIE00FmS/qJxzqCKY5wB6TlGg+NgQTycytkAyxxESbtNIQvj6F/5NWuWQbflUp1s8WcWTBPjgAR8AGVVAHF6ABmoCAO/AAnsCzdW89Wi/W67w1Yy1m9sAPWG+fu5CRtw==</latexit><latexit sha1_base64="nKqnCwuNPgmh683lgSbQnCAD9cY=">AAAB+nicdVBNSwMxEM3Wr1q/Wj16CRbBg5RsrdvurejFk1SwrbBdSjZN29BsdkmySqn9KV48KOLVX+LNf2O2raCiDwYe780wMy+IOVMaoQ8rs7S8srqWXc9tbG5t7+QLuy0VJZLQJol4JG8CrChngjY105zexJLiMOC0HYzOU799S6VikbjW45j6IR4I1mcEayN184VL2GECehWnegwrruvnuvkiKrkuqtgORKVThMqOawg6KdccB9olNEMRLNDo5t87vYgkIRWacKyUZ6NY+xMsNSOcTnOdRNEYkxEeUM9QgUOq/Mns9Ck8NEoP9iNpSmg4U79PTHCo1DgMTGeI9VD99lLxL89LdL/mT5iIE00FmS/qJxzqCKY5wB6TlGg+NgQTycytkAyxxESbtNIQvj6F/5NWuWQbflUp1s8WcWTBPjgAR8AGVVAHF6ABmoCAO/AAnsCzdW89Wi/W67w1Yy1m9sAPWG+fu5CRtw==</latexit><latexit sha1_base64="nKqnCwuNPgmh683lgSbQnCAD9cY=">AAAB+nicdVBNSwMxEM3Wr1q/Wj16CRbBg5RsrdvurejFk1SwrbBdSjZN29BsdkmySqn9KV48KOLVX+LNf2O2raCiDwYe780wMy+IOVMaoQ8rs7S8srqWXc9tbG5t7+QLuy0VJZLQJol4JG8CrChngjY105zexJLiMOC0HYzOU799S6VikbjW45j6IR4I1mcEayN184VL2GECehWnegwrruvnuvkiKrkuqtgORKVThMqOawg6KdccB9olNEMRLNDo5t87vYgkIRWacKyUZ6NY+xMsNSOcTnOdRNEYkxEeUM9QgUOq/Mns9Ck8NEoP9iNpSmg4U79PTHCo1DgMTGeI9VD99lLxL89LdL/mT5iIE00FmS/qJxzqCKY5wB6TlGg+NgQTycytkAyxxESbtNIQvj6F/5NWuWQbflUp1s8WcWTBPjgAR8AGVVAHF6ABmoCAO/AAnsCzdW89Wi/W67w1Yy1m9sAPWG+fu5CRtw==</latexit><latexit sha1_base64="nKqnCwuNPgmh683lgSbQnCAD9cY=">AAAB+nicdVBNSwMxEM3Wr1q/Wj16CRbBg5RsrdvurejFk1SwrbBdSjZN29BsdkmySqn9KV48KOLVX+LNf2O2raCiDwYe780wMy+IOVMaoQ8rs7S8srqWXc9tbG5t7+QLuy0VJZLQJol4JG8CrChngjY105zexJLiMOC0HYzOU799S6VikbjW45j6IR4I1mcEayN184VL2GECehWnegwrruvnuvkiKrkuqtgORKVThMqOawg6KdccB9olNEMRLNDo5t87vYgkIRWacKyUZ6NY+xMsNSOcTnOdRNEYkxEeUM9QgUOq/Mns9Ck8NEoP9iNpSmg4U79PTHCo1DgMTGeI9VD99lLxL89LdL/mT5iIE00FmS/qJxzqCKY5wB6TlGg+NgQTycytkAyxxESbtNIQvj6F/5NWuWQbflUp1s8WcWTBPjgAR8AGVVAHF6ABmoCAO/AAnsCzdW89Wi/W67w1Yy1m9sAPWG+fu5CRtw==</latexit>

Predictions of SB cases with fitted SIR model

- Uncertainty bounds of the least square fit ofdeterministic SIR model is narrow, but wrong.

- Confidence bounds of the MLE of stochastic SIRmodel are more conservative, but too wide to be useful.

- Neither model is correct.

30

The parameters of SIR model should changeover time as a response to the interventions

31

lock down

Timeline of events in Santa Barbara County and UCSB

• Mar 10: UCSB transitions to remote instruction

• Mar 16: First case reported in SB County

• Mar 19: State-wide “Stay-at-home” order

• Mar 31: Two cases confirmed within UCSB Community.

• Many more events before / after / in between…

32

The intervention is complex and multi-dimensional…• “Social distancing recommendation“• “Lock down order”• CDC Recommends “Don’t wear masks”• CDC Recommends “Wear masks”• The evolution of “Fear factor”• How much are people following these orders?• Types of businesses to resume

33

Receptions of the interventions can be objectivelymeasured using phone-tracking data!

34

Google’s COVID-19 Community Mobility Report: Santa Barbara County

35

We have to transfer knowledge from elsewhere, but make local adjustments…• Describing interventions by “features”

• Describe local communities by “features”

• Measure effects of interventions objectively.

• Model the data-observation process so we can infer the unseen.

36

Allow us to make inferencesin new cases

A potential path towards “reopening UCSB”?

• Build a refined model on a local level.

• Describe individuals by “features”• Class schedule, mobility patterns on campus, social activities• Contact tracing

• Targeted testing and re-testing of “high risk” individuals.

• Implement measures that are known to work, and learn from feedback.

37

Privacy concerns creep in even as we publishonly aggregate statistics….

38

“The movements of soldiers within Bagram air base - the largest US military facility in Afghanistan”

“All models are wrong, but some are useful.”

• How useful a model is can often be tested

• Quantify uncertainty with “ranges”

• Be transparent on the assumptions and potential pitfalls

39

George Box1919 - 2013

NSF Funded project on COVID-19

• “Interventional COVID-19 Response Forecasting in Local Communities Using Neural Domain Adaptation Models”• We cannot do it alone!

40

In collaboration with:Dr. Richard Beswick, Dr. Lynn FitzgibbonsMs. Jeanne Li

Coping with Heterogeneity and Uncertainty of COVID-19 ... · COVID-19, CORD-19, WHO COVID-2019, COVID-19 Tweet IDs, US Census, Facebook population density map, Covidtracking.com.

Documents

Coping with Heterogeneity and Uncertainty of COVID-19 ... · COVID-19, CORD-19, WHO COVID-2019, COVID-19 Tweet IDs, US Census, Facebook population density map, Covidtracking.com.