Back

COVID-19: Tracking the Pandemic with A Simple Curve Approximation Tool (SCAT)

Courtney, J.

2020-04-11 public and global health
10.1101/2020.04.06.20055467 medRxiv
Show abstract

In the current COVID-19 pandemic, much focus is put on flattening the curve. This epidemiological curve refers to the cases versus time graph, which shows the rise of a disease to its peak before descending. The aim in a pandemic is to flatten this curve by reducing the peak and spreading out the timeline. However, the models used to predict this curve are often not clearly outlined, no model parameters are given, and models are not tested against real data. This lack of detail makes it difficult to recreate the curve. What is much needed is a simple tool for approximating the curve to allow ideas to be tested and comparisons made. This work presents a Simple Curve Approximation Tool (SCAT) which can be used by anyone. This tool allows the user to approximate and draw the curve and allows testing of assumptions, trajectories and the wildly varying figures reported in the media. The mathematics behind SCAT is clearly outlined here but understanding of this is not required. SCAT is provided online as a downloadable MS Excel workbook with some sample cases shown. Throughout this work, the parameters used are specified so that all results can be easily reproduced. Although not intended as a prediction tool, SCAT has achieved less than 0.5 % error in short-term forward prediction. It also shows a very significant improvement on the pandemic exponential approximations found throughout media reporting. As a comparison tool, it highlights obvious differences between COVID-19 and other diseases, such as influenza, and between countries at different stages of the pandemic (China, Italy and the UK are used here for demonstration purposes). SCAT allows for quick approximation of the curve and creates meaningful comparisons and understandable visualisations for COVID-19 and other diseases.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.