The user enters a query for an event: <medical condition, location, time, normalization>

1

http://meco.l3s.uni-hannover.de:8081/timemed/

Supporting Temporal Analytics for Health-Related Events in Microblogs Nattiya Kanhabua, Avaré Stewart,

Wolfgang Nejdl L3S Research Center

Leibniz Universität, Hannover, Germany {kanhabua, stewart, nejdl}@L3S.de

Sara Romano Dipartimento di Informatica e Sistemistica

Federico II University, Naples, Italy sara.romano@unina.it

APPROACH: Temporal analytics tool for supporting a temporal, retrospective analysis of infectious disease outbreaks mentioned in Twitter. Our tool will help medical professionals to analyze disease outbreaks with real-time, social media data. In addition, we provide a means of comparing the temporal development of an outbreak event mentioned in social media against official outbreak reports.

The functionalities of our temporal analytics tool include: 1) Automatically extract outbreak events from official health

reports from World Health Organization and ProMED-mail

2) Generate time series data of Twitter for corresponding real-world outbreak events

3) Visualize/correlate the time series of Twitter vs. official sources in different temporal granularities (daily, weekly, monthly) and location granularities (country, continent, latitude, worldwide)

CHALLENGES: • Automatically detecting public health-related events is crucially important for early warning, which helps health authorities to prevent and/or mitigate public health threats.

• Twitter messages (or tweets) can be used to infer the existence and magnitude of real-world health-related events, for example: (a) I have the mumps...am I alone?; or (2) #Cholera breaks out in #Dadaab refugee camp in #Kenya http://t.co/....

• None of these previous work focused on an temporal analysis of Twitter data for general diseases that are not only seasonal, but also sporadic diseases that occur in low tweet-density areas like Kenya or Bangladesh, as we perform in this work.

Tweets

WHO and ProMED-Mail

Reports

Information Extraction

Twitter Index

Event Index

Text Pre-processing

Event Aggregation

Location Extraction

Relevance Filtering

Event Extraction

Twitter Processing

Display Results

2 The system retrieves and displays results related to the event.

Summary of the event: including estimated dates and victims/cases

3

Time series visualization for different locations

4

Cross correlation results of Twitter and official health report data

5

The system returns the list of all documents related to the event

6

Contact info: Nattiya Kanhabua L3S Research Center Appelstrasse 9a, 30167 Hannover, Germany Email: kanhabua@L3S.de