Researchers at Harvard University developed an AI system to identify the best peer leaders in social networks and promote healthy behaviors. Tested among homeless youth, the AI-selected leaders reduced risky behaviors like unprotected sex faster than traditional methods. This approach could help tackle public health issues such as HIV prevention, substance abuse, and nutrition education.
(Image Source: Harvard University)
Harvard Revolutionizes Public Health Interventions: The COVID-19 pandemic has shown us how crucial accurate information and trusted individuals are when fighting diseases. However, one of the biggest challenges for public health officials is getting this vital information to the right people, especially in vulnerable communities. Now, a new AI system is helping solve that problem by identifying the most effective individuals to spread health messages.
Researchers have tested this system among homeless youth, a group at high risk for HIV infection, and the results are remarkable. The AI-selected peer leaders helped reduce risky behaviors faster and more effectively than traditional approaches.
Youth experiencing homelessness are 10 times more likely to test positive for HIV than young people with stable housing. Social workers and public health officials have used peer change agents—trusted individuals who promote healthy behaviors within their social circles—to encourage HIV testing and condom use.
The idea is simple: if the message comes from a peer, it is more likely to be trusted and acted upon. However, success depends on selecting the right peer leaders. Traditionally, the most "popular" people are chosen to act as leaders, but popularity doesn't always mean someone is well-connected across different groups.
That's where AI comes in.
Researchers from Harvard University, the University of Southern California, and Pennsylvania State University developed an AI system that maps out social networks and identifies the most connected individuals across different groups. These people, even if not the most "popular," have a greater reach within their networks and can spread health information more effectively.
The team tested the system in a field study involving over 700 homeless youth. They partnered with drop-in centers—places where young people experiencing homelessness can access food, clothing, and other resources, including mobile HIV testing.
Working with social workers and participants, the researchers mapped out the social connections within these centers. Using their AI system, they identified peer leaders with the most diverse connections across social groups.
Once selected, these leaders received training on sexual health, HIV prevention, leadership, and communication skills. Their task was to talk to their peers about HIV prevention, encourage regular testing, and promote condom use.
The results were impressive. The AI-assisted strategy, called CHANGE (CompreHensive Adaptive Network sampling for social influence), led to significant reductions in risky behaviors like unprotected sex.
What's even more striking is how quickly these changes happened. Most participants in the CHANGE group adopted safer behaviors within just one month. In contrast, groups where the "popular" youth were chosen as peer leaders only showed improvements after three months.
"The speed at which we saw results in the CHANGE group is critical," said a Harvard graduate student and the study's first author, Bryan Wilder. “These young people often move from place to place, so reaching them quickly is key to reducing HIV transmission.”
For a highly transient population like homeless youth, the ability to see rapid improvements is essential. Many young people may leave the drop-in center within a few weeks, so a slow intervention could cause them to die entirely.
The AI system identifies individuals who act as "bridges" between different groups within a social network. Unlike traditional approaches focusing on popularity, this method ensures that information spreads more widely and quickly.
"Counterintuitively, the most popular people aren't usually the most connected within a network," said Milind Tambe, senior author of the study and a professor at Harvard. “As computer scientists, we can optimize the selection of peer leaders based on who has the most connections to different groups.”
The results prove the importance of this strategy. By choosing leaders who are well-connected across various social clusters, the researchers were able to maximize the spread of information and encourage behavior change faster.
The success of the CHANGE strategy goes beyond HIV prevention. Researchers believe this AI system can be used to address other public health challenges, such as promoting nutrition, preventing substance abuse, or spreading information during public health crises like COVID-19.
"To the best of our knowledge, this is the first demonstration of using AI to optimize health interventions in social networks," Tambe said. “We hope this project shows how AI can be applied to benefit society.”
Wilder added, “This strategy could be used to disseminate information about various health issues in communities that are often overlooked.”
This groundbreaking research was a collaboration between Harvard University, the University of Southern California, and Pennsylvania State University. Social workers and researchers were critical in working directly with participants and implementing the intervention.
The study was funded by the Army Research Office and the California HIV/AIDS Research Program.
By combining AI technology with public health strategies, this research offers a powerful new tool to address health challenges in vulnerable communities. Identifying and training the most effective peer leaders can help public health officials reach more people more quickly and with greater impact.
Also Read: Northwestern University researchers lead $56 Million from ARPA-H to pioneer whole-eye transplants
As this approach continues to be developed, it could play a key role in tackling public health issues worldwide, from HIV prevention to nutrition and substance abuse awareness. With AI's help, trusted messengers can make a bigger difference than ever before. Keep reading at Education Post News for more updates around the globe.
Loading ...
Copyright© educationpost.in 2024 All Rights Reserved.
Designed and Developed by @Pyndertech