A Q&A with a UCIPT Postdoctoral Researcher
Using sentiment analysis, is it possible to predict a human being’s longevity? Do insurance companies check for applicant Twitter handles?
I am not aware of any insurance companies that are currently collecting applicants’ Twitter handles. However, users often post suggestions about things that they did, food they ate, or the way that they felt. Therefore, I think social media data can provide valuable information about a person’s health and possibly their longevity. I believe various techniques such as sentiment analysis, topic modeling, and social network analysis would be needed to understand a person’s health profile.
Should young people be concerned about how their social media usage may affect their ability to procure life insurance later?
I believe it is possible to predict people’s health using social media data. However, more research is needed to better understand the amount of useful information that can be extracted from social media data. Our research studies indicate that many social media posts may not be relevant indicators of one’s health. Thus, researchers need to separate out the useful information from the “noise.” I think this area of research is growing rapidly. We are currently developing some very interesting social media analysis methods (e.g., machine learning, deep learning) at UCIPT that will help to extract relevant information.
Looking at UCLA student tweets, what risky behaviors do they engage in? Do you think this could be correlated with their driving records or car insurance policies?
We have not seen many risky behavior tweets based on the data that have been collected so far. However, we have only tracked 200 UCLA students, so our sample size is fairly small. I think we might be able to see more risky behavior–related tweets if we tracked more students, and it might be possible to find a relationship between risky behavior and driving records. We could build a better prediction model if we were able to determine whether students use social media/a smartphone while driving.
When UCLA students experience real-world difficulties, say a flat tire or a flooded apartment, can they use Twitter or other social media tools to assist each other? Are there free ways to use social media as a safety net in times of difficulty or despair?
I think students can use social media to assist each other. For example, I have seen people ask for school or job advice on social media. We are currently conducting a study called HOPE that evaluates the effectiveness of using social media to improve health outcomes. Our aim is to figure out the most effective ways to use social media platforms to assist each other when people are faced with various challenges.
Are actuaries taking note of your research? In the future, will we see actuarial scientists take note of sentiment analysis?
I think it’s possible that actuarial scientists will use social media data. Sentiment analysis could be very useful in that line of work. A recent study reported that sentiment from tweets can be used to predict stock prices, but other techniques such as topic modeling or social network analysis can also be useful. It really depends on the research question. Since this is a relatively young field, I think research scientists will come up with new methods of analysis that are more accurate and effective as time goes by.