Estimating epidemiological delay distributions for infectious diseases

I summarise our research on refining delay distribution estimates in epidemic modeling, a journey prompted by my general confusion. We explore and compare various approaches, drawing insights from simulations and Ebola virus disease outbreak data, and suggest paths for future improvements. Also featuring Poppy, the puppy, who is currently gnawing on my hand and occasionally archiving my emails.

Sam Abbott

Note this was written at 2am whilst sitting up with Poppy the puppy who was dealing with the trauma of her first outside the house walk.

Poppy the puppy

Warning: One five year-old described this paper as double extra boring - these comments have been noted and we are working on it (we still think it is both interesting and important though).

Where can I read it?

You can read the paper here. If interested in the code repository you can check it out here. The in-progress package that generalises the paper code is available here (contributions welcome!).

What motivated this work?

My involvement in this work grew out of my general confusion about how to best estimate delay distributions during outbreaks, and in particular, how the various methods that are used in practice compare and interact. I have been grappling with this problem for a while, but never to my complete satisfaction. This all came to a head when I was asked to help write an editorial for Ward et al. 2022, and I realized I needed some help.

As is often the case when deep thinking is required, I reached out to Sang Woo Park for assistance. This began a rather long conversation where my main contribution was repeatedly admitting, “I don’t understand.” After a while, we arrived at this gist, which finally made things click for me.

In the course of this back-and-forth, we read a lot of the related literature and realized that there was generally a lot of confusion about how to estimate delay distributions and how to compare methods. We also recognized that there was significant confusion about the various biases that can be introduced and how they interact. Obviously, this was a substantial topic, so we needed the help of some very tolerant and talented co-authors to make a dent in it. Time to light the beacons and call for help!

This collaboration led to a very enjoyable and productive period of tackling some of the challenges in estimating delay distributions. I am deeply grateful to all the co-authors for their patience and hard work, especially given the long gestation period of this paper due to its complexity and it being largely a side project for all involved. I’m also looking forward to continuing to work on this topic and hopefully making progress on the many open questions that remain.

What is the paper about?


What exactly did we do?

Key Findings

Read the paper for more details (there is a summary at the beginning and in every section to get you going - we know it is long)!

Future work

Poppy the puppy

If I entrapped you with cute puppy pictures then I apologise. Here is another picture of Poppy the puppy to make up for it. She is a 13 week old Lab/Border Collie mix from near Stonehenge. She is very smart, cute, and a serial killer (so we get on well).

Poppy the puppy just being totally normal and not planning a sustained campaign of guerrilla insurgency


If you see mistakes or want to suggest changes, please create an issue on the source repository.


For attribution, please cite this work as

Abbott (2024, Jan. 15). Sam Abbott: Estimating epidemiological delay distributions for infectious diseases. Retrieved from

BibTeX citation

  author = {Abbott, Sam},
  title = {Sam Abbott: Estimating epidemiological delay distributions for infectious diseases},
  url = {},
  year = {2024}