California State University, Long Beach
Inside CSULB Logo

Early-Warning System For Drugs

Published: August 17, 2015

Melody Kiang, a professor in the Department of Information Systems, has found a new early-warning system for adverse drug reactions (ADR) in social media.

In an article co-authored with Central University of Finance and Economics’ Ming Yang and Academy of Mathematics and Systems Science’s Wei Shang, published in the Journal of Biomedical Informatics (April, 2015), Kiang points to the internet as a way to preview what can go wrong with new pharmaceuticals.

Adverse drug reactions are believed to be a leading cause of death in the world and pharmacovigilance systems are aimed at the early detection of ADRs, Kiang wrote.

“With the popularity of social media, Web forums and discussion boards they become important sources of data for consumers to share their drug use experience and as a result may provide useful information on drugs and their adverse reactions,” she said.

In the study, Kiang proposed an automated ADR-related post-filtering mechanism using text classification methods.

“In real-life settings ADR-related messages are highly distributed in social media while non-ADR-related messages are unspecific and topically diverse,” she said. “It is expensive to manually label a large amount of ADR-related messages (positive examples) and non-ADR-related messages (negative examples) to train classification systems. To mitigate this challenge, we examine the use of a partially supervised learning classification method to automate the process.”

In her new approach, Kiang proposed a novel pharmacovigilance system leveraging called a Later Dirichlet Allocation modeling module.

“We select drugs with more than 500 threads of discussion and collect all the original posts and comments of these drugs using an automatic Web spidering program as the text corpus,” she said. “Various classifiers were trained by varying the number of positive examples and the number of topics. The trained classifiers were applied to 3,600 posts published over 30 days. Top-ranked posts from each classifier were posted and the resulting set of 300 posts published over 30 days. Top-ranked posts from each classifier were pooled and the resulting set of 300 posts was reviewed by a domain expert to evaluate the classifiers.”

Kiang feels her new design provides satisfactory performance in identifying ADR-related posts for post-marketing drug surveillance. She also noted that the overall design of their system points out a big potentially fruitful direction for building other early warning systems that need to filter big data from social media networks.”

One key to her research is a fundamental misunderstanding between patients and their doctors.

“First comes the term used by medical professionals,” she said. “My method extracts the new words used by consumers. The people who need care can’t seem to communicate with those who provide care. There has been a split in vocabularies.”

That split is the foundation of the Journal of Biomedical Informatics article, she said.

“It deals with filtering big data from social media and building an early warning system for adverse drug reactions,” she explained. “Our goal is to build an early warning system for adverse drug reactions by monitoring the consumer health-related forums. Medical professionals need to know what consumers are saying about adverse drug reactions.”

X
Melody Kiang

Currently, the problem with different ways of reporting adverse drug reactions is a time lag since patients talk about their drug reactions before they report those reactions to health professionals.

“The health professionals wait until a lot of people report these adverse drug reactions before they report them to officials,” she said. “We tried to monitor these web flows and see if we could identify any problems with a specific drug before that problem was reported to officials.”

Kiang is an associate editor of Decision Support Systems and editor in chief of the Journal of Electronic Commerce Research. She came to CSULB from Arizona State University where she worked as an associate professor from 1996-99. She earned her BBA from National Chengchi University in 1984 with a major in business administration, her M.S. from the University of Wisconsin, Madison (1987) and her Ph.D. from the University of Texas, Austin (1991).

Kiang believes her new internet survey represents the first attempt to monitor adverse drug reactions through social media.

“It all began with a student dissertation and monitoring terrorist messages on the web,” she said. “We realized we could apply the same concept to monitor other types of problems. Then we came up with this. What didn’t exist 20 years ago now is everywhere. Social media allows scientists and researchers to collect data in a timelier manner and to collect a lot of data. Social media allows more people to have access. Anyone with a computer can input their comments and ideas. It is up to us now to collect that information. It gives us a much wider database.”

Her research on the language of health care has made her more aware of what words she uses when she talks to her own doctor.

“We will be able to create a kind of dictionary that helps the communication between the two sides of the medical professional and the medical consumer,” she said. “Hopefully, our research will have some impact on that conversation.”