covid-19 forecasting modelmethodology • reported distributions of covid deaths and cases across...

21
COVID - 19 Forecasting Model

Upload: others

Post on 05-Sep-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

COVID-19 Forecasting Model

Page 2: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

Disclaimers

• The following forecast models …

• … are predicated on the persistence of current behaviors.

• … do not account for outbreaks in specific settings (e.g., nursing homes).

• … are much like weather forecasts; that is, proximate predictions are more stable

than distant predictions.

• … include the Emergency Preparedness regions with a sufficient number of

patients to produce reliable results.

Page 3: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

Methodology

• Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions.

• Nonlinear regression models were applied to observed daily counts of patients for Alabama, UAB, and Emergency Preparedness districts. (Forecast models for districts with small numbers of patients are not shown due to the instability of the results.)

• The regression models were informed by global time to peak estimates and the shape of the temporal distribution of deaths/cases from geographic areas that demonstrated optimal resemblance to Alabama, UAB, and Emergency Preparedness districts.

Page 4: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

State Emergency

Preparedness Districts

4

Page 5: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

Alabama COVID Patient Forecast

Page 6: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

0

200

400

600

800

1000

1200

1400

2/24/2020 3/15/2020 4/4/2020 4/24/2020 5/14/2020 6/3/2020 6/23/2020 7/13/2020 8/2/2020 8/22/2020

Nu

mb

er

of

CO

VID

Pati

en

ts

Observed and Predicted Number of COVID Inpatients, Alabama Hospitals

Page 7: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

0

2

4

6

8

10

12

14

16

18

20

Num

ber

of

Death

sDaily Number of Deaths Among COVID+ Hospitalized Patients, Alabama

Page 8: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

UAB COVID Patient Forecast

Page 9: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

0

10

20

30

40

50

60

70

80

90

3/1/2020 3/21/2020 4/10/2020 4/30/2020 5/20/2020 6/9/2020 6/29/2020 7/19/2020 8/8/2020

Num

ber

of

CO

VID

Patients

Observed and Predicted Number of COVID Inpatients, UAB Hospital

Page 10: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

Jefferson District COVID Patient Forecast

Page 11: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

0

50

100

150

200

250

300

2/24/2020 3/15/2020 4/4/2020 4/24/2020 5/14/2020 6/3/2020 6/23/2020 7/13/2020 8/2/2020

Nu

mb

er

of

CO

VID

Pati

en

tsObserved and Predicted Number of COVID Inpatients, Jefferson County Hospitals

Page 12: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

East Central District COVID Patient Forecast

Page 13: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

0

50

100

150

200

250

300

2/24/2020 3/15/2020 4/4/2020 4/24/2020 5/14/2020 6/3/2020 6/23/2020 7/13/2020 8/2/2020

Nu

mb

er

of

CO

VID

Pati

en

tsObserved and Predicted Number of COVID Inpatients, East Central District Hospitals

Page 14: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

Mobile District COVID Patient Forecast

Page 15: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

0

20

40

60

80

100

120

140

160

2/24/2020 3/15/2020 4/4/2020 4/24/2020 5/14/2020 6/3/2020 6/23/2020 7/13/2020 8/2/2020

Nu

mb

er

of

CO

VID

Pati

en

tsObserved and Predicted Number of COVID Inpatients, Mobile District Hospitals

Page 16: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

West Central District COVID Patient Forecast

Page 17: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

0

20

40

60

80

100

120

140

3/20/2020 4/9/2020 4/29/2020 5/19/2020 6/8/2020 6/28/2020 7/18/2020 8/7/2020

Nu

mb

er

of

CO

VID

Pati

en

tsObserved and Predicted Number of COVID Inpatients, West Central District Hospitals

Page 18: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

North Alabama District COVID Patient Forecast

Page 19: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

0

50

100

150

200

250

2/24/2020 3/15/2020 4/4/2020 4/24/2020 5/14/2020 6/3/2020 6/23/2020 7/13/2020 8/2/2020

Nu

mb

er

of

CO

VID

Pati

en

tsObserved and Predicted Number of COVID Inpatients, North Alabama District Hospitals

Page 20: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

Other Districts COVID Patient Trajectories

20

Page 21: COVID-19 Forecasting ModelMethodology • Reported distributions of COVID deaths and cases across the globe were used to model the estimated time to peak and the shape of the distributions

0

50

100

150

200

2503

/24

/20

20

3/2

6/2

02

0

3/2

8/2

02

0

3/3

0/2

02

0

4/1

/20

20

4/3

/20

20

4/5

/20

20

4/7

/20

20

4/9

/20

20

4/1

1/2

02

0

4/1

3/2

02

0

4/1

5/2

02

0

4/1

7/2

02

0

4/1

9/2

02

0

4/2

1/2

02

0

4/2

3/2

02

0

4/2

5/2

02

0

4/2

7/2

02

0

4/2

9/2

02

0

5/1

/20

20

5/3

/20

20

5/5

/20

20

5/7

/20

20

5/9

/20

20

5/1

1/2

02

0

5/1

3/2

02

0

5/1

5/2

02

0

5/1

7/2

02

0

5/1

9/2

02

0

5/2

1/2

02

0

5/2

3/2

02

0

5/2

5/2

02

0

5/2

7/2

02

0

5/2

9/2

02

0

5/3

1/2

02

0

6/2

/20

20

6/4

/20

20

6/6

/20

20

6/8

/20

20

6/1

0/2

02

0

6/1

2/2

02

0

6/1

4/2

02

0

6/1

6/2

02

0

6/1

8/2

02

0

6/2

0/2

02

0

6/2

2/2

02

0

6/2

4/2

02

0

6/2

6/2

02

0

6/2

8/2

02

0

6/3

0/2

02

0

7/2

/20

20

7/4

/20

20

7/6

/20

20

7/8

/20

20

Nu

mb

er o

f C

on

firm

ed C

ases

Number of In-House COVID-19 Patients by Day and Region

North Alabama Northeastern Shoals Southeastern Southwestern West Central