© 2014 IBM Corporation STEM Ebola Model - Sensitivity of Ebola to Interventions Kun Hu, Simone Bianco, Stefan Edlund, James Kaufman IBM Almaden Research Center
© 2014 IBM Corporation
STEM Ebola Model
- Sensitivity of Ebola to Interventions
Kun Hu, Simone Bianco, Stefan Edlund, James Kaufman
IBM Almaden Research Center
© 2014 IBM Corporation
Introduction
Ebola virus disease (EVD)
– RNA virus of the family Filoviridae, genus Ebolavirusin
2014 Outbreak in West Africa, largest in history
– Importation to US, Spain
>10,000 infectious cases, ~5000 death
Incubation: 2~21 days
Transmission:
– Direct contact with infectious blood or body fluids
– Postmortem transmission: 2 to 5 times high
– Sexual transmission through semen
Symptoms
Basic Reproductive Number (R0): 1.3 ~ 1.8
CDC predicts > 1M cases by Jan 2015
2 10/29/2014
© 2014 IBM Corporation
Model
3 10/29/2014
© 2014 IBM Corporation
Model Calibration
Nelder Mead Simplex (NMS) algorithm
– normalized mean square error (NMSE)
Fitted 2 parameters
– infectious transmission rate (𝛽𝑖)
– postmortem transmission rate (𝛽𝑑)
4 10/29/2014
© 2014 IBM Corporation
Model Parameters
3. CDC (2014) http://www.cdc.gov/vhf/ebola/
6. Towers S, et.al., (2014) Temporal Variations in the Effective Reproduction Number of the 2014 West Africa Ebola Outbreak. . PLOS
Currents Outbreaks 2014 Sep 18.
9. Althaus C (2014) Estimating the Reproduction Number of Ebola Virus (EBOV) During the 2014 Outbreak in West Africa. PLOS Currents
Outbreaks 2014 Sep 2.
11. Gomes, et al. (2014) Assessing the International Spreading Risk Associated with the 2014 West African Ebola Outbreak. . PLOS
Currents Outbreaks 2014 Sep 2
5 10/29/2014
© 2014 IBM Corporation
Simulation
Solver:
– Runge-Kutta Cash-Karp deterministic solver
Initialization and Map:
– Guinean (2), Liberia (5) and Sierra Leone (5)
– Administrate level 0
Experimental Parameters
– Hospital admission rate (𝜏): [0,1], w/ step: 0.1
– Burial rate (𝛿): [0,1], w/ step: 0.1
– Infectious transmission rate (𝛽𝑖): [0,2] w/ step: 0.2
– Postmortem transmission rate (𝛽𝑑): [0,2] w/ step: 0.2
6 10/29/2014
© 2014 IBM Corporation
Sensitivity Analysis (1)
7 10/29/2014
© 2014 IBM Corporation
Sensitivity Analysis (2)
8 10/29/2014
© 2014 IBM Corporation
Sensitivity Analysis (3)
9 10/29/2014
© 2014 IBM Corporation
Backup Slide
10 10/29/2014
© 2014 IBM Corporation
Modeling an Ebola outbreak response: STEM
Outbreaks need prompt response
Mathematical modeling of alternative scenarios
Consider air and land travel
Use reliable denominator data
Integrate datasets (humidity, temperature, wind, etc.)
Make reliable predictions at both local and global level
Free and open source (IBM to Eclipse foundation)
11 10/29/2014
© 2014 IBM Corporation
African Model
12 10/29/2014
© 2014 IBM Corporation
Global threat of the West Africa Ebola outbreak
13 10/29/2014