Zika Knowledge and Perception Studies

From Knoesis wiki
Revision as of 15:02, 2 May 2017 by Vaikunth (Talk | contribs) (Topic Modeling)

Jump to: navigation, search

One year after the Ebola outbreak ended, the Zika outbreak has started and is also causing fear and misinformation to spread. In the recent years, citizen sensing has picked up greatly with the rise of mobile device popularity, as well as with the rise in social media sites such as Facebook and Twitter. Big social data eliminate the time lag caused by traditional survey based methods, allowing for studying public opinions on issues while addressing privacy concerns of users by studying collective public behavior on specific issues. In this exploratory study, we use a combination of natural language processing and machine learning techniques to determine what information about Zika symptoms, transmission, prevention, and treatment people are discussing using tweets.

Zika Virus

Zika belongs to the Flaviviridae family of viruses. Prior to 2007 there had only been 14 confirmed cases of Zika; however death due to Zika is extremely rare. The Zika virus usually only causes mild symptoms such as a headache, rash, fever, conjunctivitis, and joint pain which can last from a few days to a week after being infected. These symptoms are similar to Dengue and Chikungunya, which are also spread by the same mosquito as Zika. The incubation period is unknown at this time but suspected to be a few days to a week. But the current outbreak that started in 2015 has sparked significant concern. This is the first outbreak of Zika associated with microcephaly, Guillain-Barre syndrome and maculopathy. In Brazil, the number of infants with microcephaly increased 20 times after the start of the Zika virus epidemic.

Since this is the first outbreak of Zika associated with these defects, management is still an important challenge. There are three main ways to get Zika: (i) being bitten by an infected Aedes mosquito, (ii) through sexual contact, and (iii) from mother to fetus. Mosquitos get the virus by feeding on someone with the virus and then spread it by feeding on other people. An infected mother can pass the virus to a newborn during pregnancy. There have been cases of an infected man spreading the virus to his partner during intercourse.

There is currently no medicine or vaccine to treat the Zika virus. Experts suggest rest, plenty of fluids, and acetaminophen or paracetamol for fever and pain. Sexual transmission can be prevented by abstaining or using condoms. Infection by mosquito bite can be prevented by wearing long-sleeved shirts and long pants, staying in places with air conditioning, staying in places that have door and window screens, sleeping under a mosquito net, and using insect repellants. People returning from places with Zika should prevent being bitten by a mosquito for three weeks to prevent the spread of the virus to uninfected mosquitoes.

Social Media

System Architecture for Zika study

Topic Modeling

We used topic modelling to find the underlying topics in each of the four disease categories to know more about the important issues in each of these categories. Latent Dirichlet Allocation (LDA) is a common method of topic modelling. LDA is a generative probabilistic model for collections of discrete data such as text corpora. It is a popular statistical model for discovering the hidden topics within the data set and helps to unravel more information regarding the data. LDA was developed by David M. Blei, Andrew Y. Ng and Michael I. Jordan in 2003 and since then has seen many areas of application document classification sentiment analysis even bio informatics.

The topic modeling results generate insightful results that allow researchers to understand the citizens’ concerns. Each topic in LDA represents certain properties, which reflects the pattern in the tweets. These topics can be easily interpreted by domain experts which allows us to get deeper into the themes within each category that can allow a more targeted interaction with health organizations. The results for topic modeling on Zika can be found here.

People

Principal Investigators: Tanvi Banerjee
Co-Investigators: William Romine, Amit P. Sheth
Graduate Students: Michele Miller, RoopTeja Muppalla, Ravali Mamidi

Publications

  1. Michele Miller, Tanvi Banerjee, RoopTeja Muppalla, William Romine, Amit P. Sheth, 'What Are People Tweeting about Zika? An Exploratory Study Concerning Symptoms, Treatment, Transmission, and Prevention'. Journal of Medical Internet Research, 2017.
  2. RoopTeja Muppalla, Michele Miller, Tanvi Banerjee and William Romine, 'Discovering Explanatory Models to Identify Relevant Tweets on Zika', Engineering in medicine and biology society(EMBC), 39th annual international conference of the IEEE, Apr 2017.

Contact: Tanvi Banerjee