Tony Tsai

May the force of science be with you

May 23, 2017 - 3 minute read - Comments - R

Narrative Outline of My Talk at the 10th China R Conference


The 10th China R Conference was held in Tsinghua University during May 19 - 21. I initialized the session of R in Public Health. Though it was my first time to organize a session, to invite speakers, and to host it, the session was successful. The number of audiences was more than 55 persons and beyond my expectation, and the discussions were enthusiastic. I believe that via my talk more Chinese have known the R Epidemics Consortium (RECON) and they may try to use RECON packages to facilitate their epidemics research.

Since materials on RECON of my slides were partly provided by Dr. Thibaut Jombart, the founder of RECON, I had sent my slides and narrative outline to Dr. Jombart for reviewing before my talk. The narrative outline of my talk is as follows:

“On May 21, 2017 I will give a talk on R Epidemics Consortium and Using Its Packages to Analyze Influenza Data (see slide 1) during the R in Public Health session of the 10th China R conference. The talk consists of four parts (see slide 2).

During the talk, I will firstly introduce the precursor of RECON — Hackout 3 (see slide 3), then RECON (see slides 4 and 5) and its forum (see slide 6). Of course, I will call on audiences to join RECON.

In the second part of RECON packages, I will introduce the aims of RECON packages (see slide 7) and then list all packages that RECON has (see slides 8 and 9), among which I will focus on three packages — outbreaks, incidence and EpiEstim as I will use them to demostrate how to use RECON packages to analyze influenza data in the following parts.

The slides of first two parts mainly refer to Dr. Jombart’s talk - Webinar_2017 and I will give thanks to Dr. Jombart during the talk.

In the Epidemic Curve and incidence package part, I will illustrate the concept of Epidemic Curve (see slide 9) and show that the ISOweek-based weekly stacked bar plot is ubiquitous plot in epidemiological reports by such a figure of China Influenza Laboratory Surveillance Information from WHO GISRS (see slide 10). In the following slides 12 - 14, I will demostrate how tedious to plot daily, ISOweek-based epidemic curve of influenza A H7N9 in China, 2013 with group of gender. The dataset used for plots is from outbreaks package, which will be firstly introduced in slide 12. Then incidence package will go to the stage (see slide 15) and I will demotrate how easy to plot those figures using incidence package and emphasize on the feature of ISOweek support that I contributed to incidence package (see slides 16 - 18).

In the last part of Time-varying Reproduction Number $R_t$ and EpiEstim package, I will first introduce the concept of reproduction number R and its significant implication (see slide 19) and illustrate how it relates to epidemic curve (see slide 20). Then I will introduce EpiEstim package developed by Dr. Anne Cori and the paper behind EpiEstim package (see slide 21). In the last R code demo (see slides 22 and 23), I will use EpiEstim package to estimate the instantaneous reproduction number for laboratory confirmed pandemic 2009 influenza A (H1N1) in Beijing and plot the estimation with holidays overlaying to explore the holidays effect on the transmission of pandemic 2009 influenza A (H1N1). You will see in the demo incidence package will also be used. By using EpiEstim package to analyze my own pandemic 2009 influenza A (H1N1) in China, I found a bug of EpiEstim package and reported it to Dr. Cori. This is why I would like to introduce EpiEstim package during this talk.”