Crowdbreaks

Crowdbreaks tracks health trends in real time by analyzing social media data using continuous crowdsourced labeling and machine learning to support public health research.

Key Features:

• Algorithmic and data transparency: Provides transparent access to algorithms and data practices to support reproducible analyses and collaborative research.
• Continuous crowdsourced labeling: Uses ongoing crowdsourcing to label social media content so training data adapts to evolving online discourse.
• Automated workflow: Automates data collection, filtering, labeling, and machine learning classifier training to accelerate analysis of social media streams.
• Real-time flexible monitoring: Enables assessment of multiple evolving health situations simultaneously for timely detection of trends.
• Machine learning classifiers: Trains classifiers on labeled social media data to identify patterns and trends in public health discourse.
• Social media data collection: Collects large-scale social media content from multiple platforms to capture broad health-related discussions.

Scientific Applications:

• Epidemiological surveillance: Tracking mentions of symptoms, outbreaks, and preventive measures to monitor disease spread and public concern.
• Public health interventions: Informing intervention strategies by analyzing current social media discourse on health behaviors and attitudes.
• Behavioral studies: Characterizing public perceptions and behaviors regarding health issues to guide communication and policy decisions.

Methodology:

Data collection from social media platforms; filtering to isolate relevant health-related content; crowdsourced labeling of filtered data; training machine learning classifiers on labeled data.