Will you catch a cold this year? Big Data tell you!

"Ding ding" mobile phone rang, you pick up a look, the screen red typing flashing "The area has 35 people suffering from influenza, please proceed cautiously, do the necessary protective measures!" "Such intelligent predictions may no longer be dreams. Would you like to know when you catch a cold this winter? Twitterhealth can tell you!

Twitterhealth is a research project initiated by the University of Rochester in New York, which can analyze the user's twitterhttp://www.aliyun.com/zixun/aggregation/7686.html "> Weibo content to predict whether users will catch a cold. Note that they only analyze the content of the microblog!

The project was originally intended for use by researchers in data mining and machine learning systems, which was later issued. Henry Kautz, director of computer science at the University of Rochester, said: "Twitterhealth is a project that specializes in analyzing different geographic information, such as GPS information for smartphones." ”

Kautz said: "We find that more and more social media are starting to participate in the function of geographical positioning, people in the Twitter microblogging will be in the passing of their geographical information." Our research team downloads and studies such information from the website. ”

Kautz students set up a computer network that is designed to download the location of the big city's Twitter microblog. Then they started picking up the available data from a huge amount of large data sources.

"We find that people often mention their health in Twitter Weibo. For example, ' I have a runny nose ', ' I have a cold ', ' I feel uncomfortable ' and so on. Let's assume that we can track seasonal flu based on these words. ”

So the research team began to humble handwriting machine learning algorithm, in hundreds of experimental microblog selected "cold microblog."

Finally, the team's algorithm has been able to reach 99% accuracy in selecting a "cold microblog", almost consistent with the human brain's analysis of the text, and "cold" analysis faster than the largest centers for Disease control in the United States.

"From the large data we can find seasonal influenza distribution and transmission, we measure and predict the accuracy of the disease Control center completely." "Kautz said.

Twitterhealth's great success has also prompted many students to start businesses in big data and data mining. In addition to disease-spreading trends, they want to track more trends.

Kautz said: "In addition to health forecasts, you can also do more business-related applications, such as tracking the latest fashion trends, tracking the hottest network singers, network jokes. ”

But Kautz also highlighted the great potential of large data in the healthcare industry. "In the past, it was slow and expensive to collect health data through questionnaires. He also mentions that Twitterhealth is helpful in fighting depression and suicide, and can warn you before a suicide occurs.

Kautz said: "By analyzing large data sources, we can find out whether users go through a disease-prone neighbourhood or eat in a restaurant where the flu patient has eaten, because there is a risk of increasing the chance of getting the flu." ”

Twitter Weibo's content is a good source of data analysis, but social networking sites such as Facebook can see more privacy, but also because of privacy issues that make data access problematic. If Facebook can be persuaded to use services such as twitterhealth, or to get the status and articles of Facebook users, Twitterhealth can serve more people and reduce the incidence of disease.

Tech2ipo: This method can be completely transplanted to China. Many Chinese people like to send their stories online, such as "I had 32 meals today", "I have a cold, so lonely," incredibly pregnant "," LV bag is not as good as Gucci "," a teacher is a good school "," seven Stone for Wish "and so on content, can do diet, health, population, consumption, education , emotional analysis, the potential is still very large. But also take into account a lot of false content, such as a girl in someone else's car microblogging said "or a Cadillac sitting comfortably" such words, will be the data analysis caused great interference.

Article Source: InformationWeek

(Responsible editor: The good of the Legacy)