This repository collects figures generated from the BeOutbreakPrepared nCoV 2019 public line list.
The figures are rendered with the following conventions:
- Time within the outbreak is rendered as the x-axis.
- Cases are rendered along the y-axis and labelled with their geographic location on the left side of the graph.
- Each figure shows the imported COVID-19 cases for a given non-China country sorted by cluster_ID, then case ID.
- Arrows for each case signify the flow of time between the first and last recorded event for a given case.
- cluster_ID is assigned via a combination of manual curation and inference where the cluster is named after the ID of the index case.
- cluster_IDs are manually curated by volunteers via a review of press articles regarding the case origin, seeking evidence of autochthonous transmission.
- For cases for whom no travel is recorded, cluster_ID is assigned as the ID of the first individual at their given named location who travelled to China and was infected.
- cluster_IDs are represented by both a row color assignment (yellow for earlier clusters) and a right axes label constructed as such:
- '{ID} {"Inf" if inferred} - Cl {cluster_ID}'
- Imported cases which appear to have originated in China and which are not the index case of an autochthonous cluster are rendered with an uncolored background and a right axes label constructed as such:
- '{ID} Imp' (for imported)
- Cases which do not appear to have originated in China and which do not appear connected to any cluster by inference or curation are rendered with an uncolored background and a right axes label constructed as such:
- '{ID} Unk' (for unknown)
- Inferred clusters' row backgrounds are rendered with lower saturation than manually curated clusters to convey a reduced confidence level.
- Within this data set, temporal data is divided between singular events and ranges which occur between specific events.
- Events include date_travel, date_onset_symptoms, date_admission_hospital, date_death, date_discharge, date_confirmation, and date_missing if no valid date data is present.
- Ranges include time spans between all temporally proximal events excluding date_confirmation.
- Ranges are determined by event occurence and presence in the source data set and are not uniformly present or ordered across cases.
- Events are rendered as unfilled, rectangular glyphs over a singular day in a darkened shade of the event color.
- For days with multiple events, the traditionally later event is rendered as an expanded rectangle over the traditionally earlier.
- Ranges are rendered as filled rectangular spans between two events on the x axis.
- For events with multiple dates, the first date is taken.
- For cases with no date data, the date_missing date is assigned to one day prior to the first event in the data set.
- date_death and date_discharge are calculated from the date_death_or_discharge and outcome fields of the original data.
- Dates which are misformatted, occur in the future, are more than 30 days apart from their expected preceding event, or are before December 15th are set to NaT.