MPC Embraces Artificial Intelligence to Highlight Human Impact of Medicaid Cuts

Detroit-Based 501(c)4 Releases Data Analysis on Long-Term Study of 59 Michiganders at Risk of Losing Medicaid Benefits

Michigan People's Campaign (MPC), a Detroit-based 501(c)4 organization, is using artificial intelligence to understand the emotion, sentiment, and context in the answers people provided during a long-term study of state residents in danger of losing a pandemic-era benefit. What they learned is that the new technology changes everything about listening to people in studies, surveys, polls, and a lot more.

“Our respondents told us stories of cumbersome barriers to accessing Medicaid, Medicare, and emergency Medicare services," Audrey Gerard, a healthcare organizer at MPC, said. "However, the rollback of Medicaid services due to the decision to end the public health emergency caught them off-guard. More than 75% of those surveyed said they were unaware that Medicaid was ending, demonstrating how state-based insurers failed to inform recipients.”

"Michiganders from urban, suburban, and rural areas across immigration status, race, and ethnicity told us they were generally pleased with the Medicaid and Medicare services they have been receiving in Michigan,” Gerard added. “The fact that NLU [natural language understanding] revealed a gap between their Medicaid Quality of Care score and their feelings about the services provided is something that deserves deeper analysis."

"We used NLU to analyze the stories people told us in the study," Sebastian James, MPC's Data & Technology Manager, said. "One of the first things we could see in their words was how the details of their experiences didn't match up with the quality of care score they eventually gave the healthcare provider. Quality of care is one of Medicaid's most important service benchmarks, and we wanted to look deeper into this gap."

The technology will allow MPC to see how people feel at the most granular levels, and it allows for unique new ways to look at data. For example, as shown in this word cloud, the intensity and relevance of the muted pink "Somewhat Dissatisfied" keywords begs some obvious questions: If people were that dissatisfied, why are there no bright red "Completely Dissatisfied" keywords? Are people pulling their punches?

"A good follow-up would be to re-interview the respondents with the larger keywords and ask them to discuss their experience and reassess their quality score," James added.

NLU is an algorithm that breaks down the words in a phrase, analyzes them to understand purpose and interrelationships, and provides a detailed report on emotion, sentiment, syntax, semantics, and context.

"This study is the beginning," James said. “Using NLU along with a quantitative and qualitative question will give us a simple answer, the person's reasoning and the motivations underlying everything in one place. It will have an impact on how we recruit and organize people."

The study was conducted in 2022-23 in partnership with the Center for Popular Democracy, Make the Road States, and the People's Action Institute. Tap here or on the image below for the data story.