A new study by Johns Hopkins University has revealed that an AI-based lifestyle modification app for people with prediabetes was able to reduce the risk of developing diabetes as effectively as traditional programs led by human experts.
The study's findings were published in the journal JAMA. The first randomized, phase 3 clinical trial demonstrated that a fully AI- powered digital program can meet the Centers for Disease Control and Prevention (CDC) criteria for reducing diabetes risk.
"Even outside of diabetes research, there are very few clinical trials that directly compare AI- led interventions with traditional human-led ones, which is what makes our findings particularly significant," said Nestoras Mathiodakis, co-director of the Johns Hopkins University Diabetes Prevention Program and principal investigator of the study.
The study involved 368 people with an average age of 58, who were referred to either a traditional remote human program or an application based on a reinforcement learning algorithm, which provides personalized notifications to help improve eating behavior, physical activity, and weight management.
The results showed that 31 percent of participants in the smart app and 31.9 percent of participants in the human programs were able to achieve the CDC's diabetes prevention criteria after just one year, meaning that the effectiveness was very similar between the two groups.
Benjamin Lalani, co-author and Harvard medical student, explained: "The biggest obstacle to completing a diabetes prevention program is getting started, and logistical hurdles are often the reason. So, we noticed that the accessibility of the digital program made people more willing to participate from the outset."
The study also revealed that program start rates were higher in the group that used artificial intelligence (93.4 percent versus 82.7 percent), as were completion rates (63.9 percent versus 50.3 percent), indicating that accessibility and flexibility increased engagement and commitment.
artificial intelligence technologies in preventive healthcare, especially for groups that face difficulties in attending traditional human programs.
Lalai said: "Unlike human programs, AI-powered programs can be fully automated and available around the clock, which makes them more resilient to challenges such as staff shortages or time constraints."
The research team plans to continue the study on a larger scale, to explore the effectiveness of the application among the less fortunate or those who lack the resources to participate in traditional programs, as well as to study patient preferences and costs associated with digital programs.
These results indicate an important shift in the future of diabetes prevention, where artificial intelligence can play a pivotal role in expanding the scope of personalized healthcare and making it more flexible and inclusive.
.webp)
0 Comments