Ebola in West Africa, 2014-15

I recently read a story, Ebola's not done with West Africa. It's sad to see that the epidemic is still killing people. I've been constructing a model of the disease spread. Unfortunately, it forecasts the epidemic won't be officially over for another year:

On 2016-10-19 ± 22 days
(probably between 2016-09-27 and 2016-11-10).

(With an error margin of many days it's clearly ridiculous for the countdown timer to show hours, minutes, and seconds. But including them reflects the sense of urgency I feel — I look forward eagerly to the end of this epidemic and hope it comes much sooner than my forecast!)

But take that with a grain of salt — extrapolations like this are notoriously unreliable. Watch a history of past predictions since January 2015 as I refined my model and acquired new data:

 Past predictions

What may be more meaningful and interesting are some other insights from the model. In order to get it to fit the data well, I had to make a couple surprising assumptions:

1. Basic reproduction number

The basic reproduction number, $R_0$, measures how fast an infection is spreading. When it's greater than one the disease is growing; when it is less than one it will die out. According to my model, $R_0$ was dropping in early 2014 until it suddenly surged up between April and July. I wonder how the ebola epidemic escaped containment between April and July 2014?

 Basic reproduction number

Read more about how I calculate the basic reproduction number.

2. Case fatality rate

The case fatality rate, $CFR$, measures the risk of an infected person dying. My model suggests that for hospitalized patients, $CFR_H$ dropped in half from July to November 2014. What changed between July and November last year to improve $CFR_H$ so much?

Then $CFR_H$ dropped again in mid 2015. Again, why?

 Case fatality rate

Read more about how I calculate the case fatality rate.



  • science/ebola2014/start.txt
  • Last modified: 2015-11-19 23:00 (6 years ago)
  • by Rik Blok