Extracting health-related causality from twitter messages using natural language processing.

Area of Science:

• Computational Linguistics
• Public Health Informatics
• Social Media Analysis

Background:

• Twitter serves as a rich source for understanding daily health-related topics and concerns.
• Analyzing health-related tweets can provide valuable insights into public health conditions.
• This research explores automated methods for extracting information from social media.

Purpose of the Study:

• To evaluate an approach for extracting causal relationships from health-related tweets.
• To leverage natural language processing (NLP) techniques for analyzing social media data.
• To identify expressions related to specific health topics such as stress, insomnia, and headache.

Main Methods:

• Utilized lexico-syntactic patterns derived from dependency parser outputs for causality extraction.
• Focused on three prevalent health topics: stress, insomnia, and headache.
• Employed a large dataset comprising 24 million tweets for analysis.

Main Results:

• The proposed NLP approach demonstrated high performance, achieving an average precision ranging from 74.59% to 92.27%.
• Results were validated through comparison with human annotations, indicating the method's reliability.
• The precision rates suggest the effectiveness of the automated extraction technique.

Conclusions:

• Manual analysis of extracted causalities revealed significant findings regarding user-expressed health concerns.
• The study highlights the potential of NLP in uncovering public health trends from social media.
• Findings contribute to a better understanding of how individuals discuss health issues online.