A Study to analyze the Public Sentiment of Zika Outbreak

 A Study to analyze Public Sentiment of Zika Outbreak

Machine learning techniques need to be leveraged to categorize tweets into 3 sentiment categories: negative, neutral, and positive. By analyzing the type of sentiment from tweets, public health agencies could be informed in the negative sentiment category to ascertain public negative sentiment could help inform the CDC on the tone needed in their messages. For example, if tweets concerning vaccines are associated with a negative sentiment, the CDC can share information regarding the benefits of vaccinations. This could encourage some people who were not previously vaccinated to get vaccinated, helping prevent the spread of disease.

Topic Models 

Topics from the Positive Sentiment Tweets:

Table details the ten topics for the positive sentiment tweets, lists the keywords for each topic, and provides an example tweet for each topic. Within positive topics, topic #1 includes tweets about a new model researchers are developing to study Zika pathogenesis, topic #2 is about 3d-printed mini brains used for understanding the Zika virus, topic #3 is regarding the discovery of Zika stunting the development of the fetus, topic #4 includes tweets discussing how the sweat-emitting Brazilian billboards kills the Zika carrying mosquitoes, topic #5 was about the development of vaccines to treat Zika, topic #6 was about different types of tests for identifying Zika infection, topic #7 was about the detection of Zika using fetal tissue, topic #8 was about the IBM magic bullet to destroy all killer viruses, topic #9 was detection of Zika accumulations in the brain, and topic #10 was about the ways to kill mosquitoes.

Here is the link to Positive Sentiment LDAvis results:  http://symptoms-positive.bitballoon.com/

Here is the link to Positive Sentiment Wordcloud results: http://ravali-mamidi.info/sentiment-topics/positive-sentiment-wordcloud

Topics from the Negative Sentiment Tweets:

The topic model for negative sentiment are Topic #1 was about the brain defects caused by Zika, topic #2 includes tweets about the neurological system defects due to Zika infection, topic #3 was the initial reports of Zika-related cases, topic #4 describes the impact of Zika infection, topic #5 was regarding the fetal effects of Zika virus infection, topic #6 was about the abnormalities caused by Zika, topic #7 was about how ultrasounds were not able to identify fetal Zika infection, topic #8 was regarding Guillain-Barré syndrome, topic #9 was about how Zika affects children and adults, and finally, topic #10 was the association between dengue antibodies and the Zika virus.

Here is the link to Negative Sentiment LDAvis results: http://symptoms-negative.bitballoon.com/

Here is the link to Negative Sentiment Wordcloud results: http://ravali-mamidi.info/sentiment-topics/negative-sentiment-wordcloud

Topics from the Neutral Sentiment Tweets:

Table provides the topic, keywords in that topic, and an example tweet for the neutral sentiment topic modeling results. From table 6, topic #1 was how scientists are trying to unravel the Zika mystery, topic #2 was about the dangers of being bitten by infected Aedes mosquitoes to pregnant mothers, topic #3 was about how Zika is a mosquito borne illness, topic #4 was about the laws regarding birth control and abortion, topic #5 was about fighting the mosquitoes, topic #6 was regarding the officials warning the public to be careful not to be bitten by a mosquito at work, topic #7 tells about the knowledge gaps concerning Zika, topic #8 is regarding Zika symptoms, topic #9 shows the comparison between dengue and Zika, and topic #10 describes fetal brain damage caused by Zika infection.

Here is the link to Neutral Sentiment LDAvis results: http://symptoms-neutral.bitballoon.com/

Here is the link to Neutral Sentiment Wordcloud results: http://ravali-mamidi.info/sentiment-topics/neutral-sentiment-wordcloud