L&S Unites Initiative Funds Study Using Machine Learning to Spot Influence in Global Health Policy
global health
(Ben White/Unsplash)

The UC Davis College of Letters and Science has awarded a major interdisciplinary research grant to a research team studying how U.S. private companies and industry organizations affect global public health policy.

The $133,000 project, funded through the L&S Unites Initiative, is led by UC Davis faculty members in data science, economics, English, political science and statistics. The project, “Uncovering Private Sector Influences on Global Public Health Policy Using Automated Text Analysis,” involves a machine learning approach to analyze thousands of pages of text from the U.S. health industry and to identify potential influence in global health policy. 

“We're generally interested in how private actors, firms and industry groups influence public health policy,” said Lauren Peritz, an associate professor in political science and the study’s principal investigator. “This is important because lobbying and influence have a much broader reach especially for places around the world that don't have the same depth of public health infrastructure as we do in the U.S.” 

“We are incredibly excited that this project is the first we are funding through our L&S Unites initiative to support innovative cross-college, interdisciplinary research,” said Lori Lubin, associate dean for research and graduate studies in the College of Letters and Science. “This project is led by talented colleagues who are building a much deeper understanding of outside influences on global health policy, a pressing issue in our post-pandemic world.”

Connecting pages of text to global health decisions

The relationship between U.S. industry and governance, both domestic and international, is well documented, said Peritz. For example, political scientists and journalists in the early 2000s documented links between pharmaceutical and biotechnology companies and government policies related to intellectual property for medications.

Individual companies and industry organizations build these links by direct government lobbying as well as influence campaigns using news media and other forms of mass communication.

“A lot of these are science-based ideas that help strengthen health standards globally,” said Peritz. “We are just trying to identify the interests on the private sector then looking at documents from public health organizations globally and see what kind of language they're using on similar issues.”

The documents that are the basis for this new project come from Kathryn Russ, professor and chair of the Department of Economics and a collaborator on the project. For an earlier study, she received thousands of global health policy documents through government requests under the Freedom of Information Act (FOIA). Under U.S. law, FOIA gives any person the right to access most unclassified U.S. government records. 

“The methods that we are using in this project will enable us to speak about these patterns in topics and language in a way that's much more supported by the data and that can be replicated by others,” said Russ.

From a public records request to natural language processing

The team, which also includes Xiao Hui Tai, an assistant professor in statistics, and Carl Stahmer, executive director of the UC Davis DataLab, is customizing a natural language processing computer algorithm to pinpoint specific language patterns across these thousands of pages of text. 

Natural language processing combines linguistics, computer science, data science and statistics with computer models of language to understand large collections of text. In this study, the research team will have help from undergraduate research assistants, who will do the initial manual coding that will “train” the algorithm to identify those language patterns in new text.

This approach makes it possible to sift through thousands of pages of text that could otherwise take a lifetime to read and code by hand. The results will also be objective in a way that individual human readings might not be.

“We're asking questions that intersect economics and politics and we're drawing on tools that are inherently statistical in nature,” added Peritz. “I don't think that any one person has all of those depths of knowledge to do the whole project on their own.”

About L&S Unites

L&S Unites is a UC Davis College of Letters and Science research initiative that provides two-year awards of up to $150,000 for research led by interdisciplinary collaborations. The program leverages the strengths in research from across Letters and Science and the broader UC Davis community to address current social, political, economic and environmental grand challenges as well as to explore innovative ideas and build the future of emerging research areas.


Primary Category

Secondary Categories

Science & Technology