Home » African American Health » Conversation assistants could be used to improve the health of African Americans around COVID-19

Conversation assistants could be used to improve the health of African Americans around COVID-19

An African American man wearing a protective face mask to prevent viral infection.  Photo credit: iStock

F&A FACULTY

University of Michigan researchers say hiring an interviewer – think “Hey Siri” or perhaps something less global like “Ask the Doc” – could be a way to fill the gap in health inequalities that are emerging affect the African American community, particularly in the U.S. current COVID-19 pandemic.

Lionel Robert

Lionel Robert

Lionel Robert, associate professor in the School of Information and senior author of a study in the journal Digital Government: Research & Practice, says a technological solution could prevent economic, social problems, suspicion and privacy concerns from blocking African American access to health care.

Can you briefly summarize the problem of health inequalities as a whole and related to COVID-19?

African Americans have at least two problems: 1) They are more likely to have comorbidities such as type 2 diabetes and kidney disease, among others. This puts African Americans at greater risk of dying from COVID-19. 2) Traditionally, African Americans have had less access to health care and less confidence in those who provide health care. As a result, African Americans are less likely to have access to the medical care needed to diagnose and treat COVID-19 quickly. This results in the diagnosis being made much later, which can allow the spread of COVID-19 to continue for much longer, even among family and friends. A lack of trust could also undermine their willingness to seek medical help.

Can you explain what conversational agents are and what different forms they take?

Think of Amazon’s Alexa and Apple’s Siri as popular conversation agents. Most of the conversation agents we use are non-verbal and communicate over text using natural language processing. For example, ITS (UM’s Internet Technology Group) and the UM library each have their own conversation agents on their sites to help users. The core of what constitutes conversational agents is their ability to allow human users to have a natural conversation with the agents as they would with any other human user. Alexa and Siri are much more powerful as they can support conversations across a range of different topics, while the ITS Conversational Agent is narrowly designed to answer only questions about ITS.

Why are these especially helpful with the African American population?

Currently, in many cases, they are not. This is because many conversation agents are not designed specifically for African Americans. These conversation agents are not context specific to any particular group of ethnic users. In most cases, as in the ITS example, this is not a problem. However, when it comes to innate conversations about one’s own health, conversation agents must be designed to understand and communicate the terminology that the user is familiar with. It should also be geared towards countering questions of distrust in health care. If properly designed, conversational agents could be a source of credibility information and advice.

Messaging is only as good as the ability to get it to those who need to hear or see it. How are CAs more visible to African Americans than other forms of communication?

The problem is, they aren’t. In fact, in many ways, they are the least visible to the African American community. This would need to be addressed through working with community stakeholders such as local governments, community health centers, and places of worship such as churches. So we clearly benefit from CAs for the African American community, but we also recognize the obstacles that need to be overcome.

How can CAs be used to provide targeted information about COVID-19?

For example, an African American with type 2 diabetes might wake up each morning and tell the other person how they are feeling, perhaps give information such as temperature. Based on this information, along with information available across the Community (there may be a recent COVID-19 outbreak in the area), the person interviewed could recommend that the person see their doctor. The interlocutor could also be more specific and recommend that the person go to the local COVID-19 testing site near their home and provide a location. The conversation agent could also help address misinformation issues. For example, if the person hears misinformation about COVID-19 and its treatment, that person could go to their interlocutor for accurate information.

More information: