The June 3, 2020 7-Day Projections and Long-Term Projections for
NCR COVID-19 Cases; and Medical Demand Projections for Hospitalized, ICU, and
Ventilator Cases are shown below.
The Long-Term Projections go out to 21 days. Please understand that with the longer projections, the level of uncertainty also increases.
Individual county-level projection graphs are available in the attachment.
=========================================
IEM’s AI Modeling:
Short-term COVID-19 Projections For MWCOG
NEW: Long-range COVID-19
Projections
Leveraging over 15 years of support to HHS
for medical consequence modeling and our proprietary artificial intelligence
(AI) models, IEM believes that our Coronavirus model outputs can be used to
assist localities and their medical facilities to better prepare for an
increase in hospitalizations, to better plan for and locate drive-through
testing facilities, and to determine where increased levels of transmission may
be occurring.
We have been refining our AI model over the past month and are confident
in its ability to provide accurate 7-Day projections of confirmed cases that
can be used for operational and logistical planning. Additionally, we also provide projections for
medical demand – hospitalizations, intensive care unit (ICU) care, and mechanical
ventilation. New long-range projections have been
developed to provide projections for up to 3 weeks. These long-range projections naturally have higher uncertainty than the short-term
(7-day) projections.
AI-based Model Background
IEM is currently using an AI model to fit
data from various sources and project new cases of COVID-19. We do not
assume the average number of secondary infections (R-value) stays the same over
time. IEM’s AI model finds the best
R-value over time to evaluate how it changes over the course of the
outbreak. The IEM modeling team is
running ~11 million simulations to fit each jurisdiction’s data for actual
confirmed cases and using the best fit for the R-value to project new cases
over the next 7 days. Any changes in the R-value are purely based on the
confirmed case data and are not attributable to specific factors such as social
distancing (although social distancing is a major factor). The AI models are executed on a daily basis
to evaluate the changing dynamics of the COVID-19 pandemic. Our projections
have typically been within 20%, and are often within 10%, of actual confirmed
cases.
The projections shown in this document
are based on data pulled in as of 6/3/20 11 a.m.
Please provide any feedback or send any
questions that you might have to us. We
are continually updating and improving the model, so your feedback is
critical.
Also, if you have more current or refined
data for your jurisdiction that you would like IEM to factor in, please let us
know.
IEM’s Modeling Lead
Dr. Prasith “Sid” Baccam is a Computational Epidemiologist expert at
IEM with more than 20 years of
experience in medical consequence modeling and simulation of disease outbreaks
and medical consequences following hypothetical attacks with biological agents
or emerging infectious diseases. He develops key simulation models and decision
support tools at IEM, specializing in public health, disaster response, and
medical countermeasures (MCM) to enhance data-driven decision making and
improve modeling assumptions.
Upon receiving his Ph.D. in Applied Mathematics and Immunobiology at Iowa State
University, Dr. Baccam worked as a Postdoctoral Research Associate at Los
Alamos National Laboratory where he focused on researching viral and
immunological modeling. He has worked with state and local public health
officials as well as Federal agencies, including HHS, the Centers for Disease Control and Prevention (CDC), and the Department of Homeland Security (DHS). Dr. Baccam has published numerous papers on public health
response models and implications on policy and has been invited to participate
in workshops and symposiums held by the Institute of Medicine (now the National
Academy of Health). His modeling results have been briefed to the Executive Office of the President and
informed two presidential policy actions.
Actual Confirmed Cases: 6/2/20 Projected
Confirmed Cases: 6/9/20
Actual Confirmed Cases On:
|
Projected Cases For:
|
|||||||||||
5/31
|
6/1
|
6/2
|
6/3
|
6/4
|
6/5
|
6/6
|
6/7
|
6/8
|
6/9
|
6/16
|
6/23
|
|
Washington DC
|
8,801
|
8,886
|
9,016
|
9,101
|
9,183
|
9,263
|
9,341
|
9,416
|
9,488
|
9,559
|
9,998
|
10,355
|
Charles
Co., MD
|
1,120
|
1,133
|
1,152
|
1,167
|
1,182
|
1,197
|
1,212
|
1,227
|
1,241
|
1,256
|
1,357
|
1,454
|
Frederick
Co., MD
|
1,939
|
1,961
|
2,013
|
2,042
|
2,071
|
2,101
|
2,131
|
2,162
|
2,193
|
2,464
|
2,733
|
|
Montgomery Co., MD
|
11,476
|
11,731
|
11,924
|
12,138
|
12,350
|
12,562
|
12,773
|
12,982
|
13,192
|
13,400
|
14,836
|
16,236
|
Prince George's Co., MD
|
15,808
|
15,968
|
16,116
|
16,339
|
16,558
|
16,772
|
16,983
|
17,190
|
17,394
|
17,593
|
18,898
|
20,052
|
Arlington
Co., VA
|
2,123
|
2,133
|
2,142
|
2,160
|
2,177
|
2,193
|
2,209
|
2,224
|
2,239
|
2,254
|
2,344
|
2,416
|
Fairfax
Co., VA
|
11,341
|
11,548
|
11,718
|
11,935
|
12,150
|
12,362
|
12,573
|
12,782
|
12,989
|
13,195
|
14,590
|
15,911
|
Loudoun
Co., VA
|
2,611
|
2,636
|
2,659
|
2,719
|
2,779
|
2,841
|
2,905
|
2,969
|
3,035
|
3,102
|
3,612
|
4,197
|
Prince William Co., VA
|
7,085
|
7,186
|
7,323
|
7,494
|
7,665
|
7,838
|
8,011
|
8,185
|
8,360
|
8,536
|
9,795
|
11,101
|
Alexandria, VA
|
1,981
|
2,000
|
2,016
|
2,039
|
2,061
|
2,083
|
2,104
|
2,124
|
2,144
|
2,163
|
2,283
|
2,382
|
NCR
|
64,285
|
65,182
|
66,079
|
67,134
|
68,176
|
69,212
|
70,242
|
71,261
|
72,275
|
73,283
|
80,177
|
86,837
|
Projection Graphs:
Maryland County Projection Graphs:
Maryland County Projection Graphs:
Virginia County Projection Graphs:
Virginia County Projection Graphs:
Virginia City Projection Graphs:
Some
recipients of our daily COVID-19 short-term (7 day) projections have requested
projections of demand for: hospital bed, intensive care unit (ICU) beds, and
mechanical ventilation. We realize that different states and localities will
have different characteristics for hospital demand of COVID-19 cases, and we
are presenting the best assumptions we could find for those medical demands
based on scientific literature and health data reporting. Specifically:
•
Beds: For
hospitalization, we use a range of 10% and 20% of cases require hospitalization
based on CDC’s report (MMWR, March 18, 2020) and
state reports of COVID-19 cases.
•
ICU: The CDC report
found that 24% of hospitalized cases require ICU care.
•
Ventilators: Based on
clinical data from China and state reports, we assume that 50% of ICU cases
require a ventilator.
If
you have other estimates for these assumptions, please share them with us as we
work to refine our modeling, assumptions, and data on a daily basis.
The
medical demands shown in the table assume 20% of cumulative confirmed cases require hospitalization. To get the
medical demand for the assumption that 10% of confirmed cases require
hospitalization, simply divide the demand by 2.
If you have your own jurisdiction-specific percentages, those can be
applied to the projected cases to derive your own medical demands. Likewise, you can apply your own
jurisdiction’s demographics information to the projected cases to estimate
impacts on subpopulations within your community (such as people over the age of
65 years).
Medical Demand Projections:
Actual Confirmed Cases On:
|
Projected Cases (Hospitalized) [ICU]
{Ventilator} For:
|
||||
5/31 6/1
|
6/2
|
6/9
|
6/16
|
6/23
|
|
Washington
DC
|
8,801 8,886
|
9,016
|
9,559 (1,912)
[459] {229}
|
9,998 (2,000)
[480] {240}
|
10,355 (2,071)
[497] {249}
|
Charles Co., MD
|
1,120 1,133
|
1,152
|
1,256 (251)
[60] {30}
|
1,357 (271)
[65] {33}
|
1,454 (291)
[70] {35}
|
Frederick Co., MD
|
1,939 1,961
|
2,013
|
2,225 (445)
[107] {53}
|
2,464 (493)
[118] {59}
|
2,733 (547)
[131] {66}
|
Montgomery Co., MD
|
11,476 11,731
|
11,924
|
13,400 (2,680)
[643] {322}
|
14,836 (2,967)
[712] {356}
|
16,236 (3,247)
[779] {390}
|
Prince George's Co., MD
|
15,808 15,968
|
16,116
|
17,593 (3,519)
[844] {422}
|
18,898 (3,780)
[907] {454}
|
20,052 (4,010)
[962] {481}
|
Arlington Co., VA
|
2,123 2,133
|
2,142
|
2,254 (451)
[108] {54}
|
2,344 (469)
[112] {56}
|
2,416 (483)
[116] {58}
|
Fairfax Co., VA
|
11,341 11,548
|
11,718
|
13,195 (2,639)
[633] {317}
|
14,590 (2,918)
[700] {350}
|
15,911 (3,182)
[764] {382}
|
Loudoun Co., VA
|
2,611 2,636
|
2,659
|
3,102 (620)
[149] {74}
|
3,612 (722)
[173] {87}
|
4,197 (839)
[201] {101}
|
Prince William Co., VA
|
7,085 7,186
|
7,323
|
8,536 (1,707)
[410] {205}
|
9,795 (1,959)
[470] {235}
|
11,101 (2,220)
[533] {266}
|
Alexandria,
VA
|
1,981 2,000
|
2,016
|
2,163 (433)
[104] {52}
|
2,283 (457)
[110] {55}
|
2,382 (476)
[114] {57}
|
NCR
|
64,285 65,182
|
66,079
|
73,283 (14,657) [3,518]
{1,759}
|
80,177
(16,035) [3,848] {1,924}
|
86,837
(17,367) [4,168] {2,084}
|
For additional information from IEM,
please contact Bryan Koon, Vice President of Emergency Management and Homeland
Security at bryan.koon@iem.com
or 850-519-7966 or Stephanie Tennyson at stephanie.tennyson@iem.com or 202-309-4257.
No comments:
Post a Comment