Reconstructing historic and modern potato late blight outbreaks using text analytics.

Area of Science:

• Plant pathology
• Computational linguistics
• Digital humanities

Background:

• Phytophthora infestans, the pathogen responsible for late blight, has caused significant crop losses and historical famines.
• Archival and modern documents contain valuable information on disease outbreaks, but this data is often unstructured and inaccessible.
• Understanding the historical spread and management of plant diseases is crucial for addressing current and future food security challenges.

Purpose of the Study:

• To utilize text analytics on historical reports (1843-1845) to map early US late blight outbreaks.
• To characterize historical theories regarding the pathogen's origin and control strategies.
• To develop modern late blight intensity maps using social media data (Twitter).

Main Methods:

• Text analytics applied to unstructured historical documents (1843-1845).
• Mapping of disease spread across US states and Canadian provinces.
• Analysis of Twitter feeds for contemporary late blight discussions and intensity mapping.
• Topic modeling to identify themes in historical and modern data.

Main Results:

• The study mapped the spread of late blight from 5 to 17 states/provinces in the US and Canada between 1843-1845.
• Historical documents discussed crop losses, potential Andean origins, and various control methods.
• Modern Twitter data provided near-global and local disease observations, with topic modeling revealing information on disease, research, and outbreaks.

Conclusions:

• Text analytics and social media mapping are powerful tools for exploring and visualizing historical and current pandemics.
• This approach enhances the accessibility of unstructured data for tracking plant diseases.
• The methodology can aid researchers in understanding disease dynamics and informing food security strategies.