A collaborative research team led by Khoa Luu, associate professor in the Department of Electrical Engineering and Computer Science at the University of Arkansas, has been awarded a $1 million grant from the National Science Foundation to investigate how tobacco promotion on social media affects youth addiction.

The researchers are, from the U of A, Khoa Luu and Page Dobbs; from Mississippi State University, Asad Waqar Malik; from Kansas State University, Samee Khan; and from the U of A Division of Agriculture, Brandon McFadden.

The four-year project, titled "SCH: Large Scale Multi Modality Learning System to Identify Tobacco Addiction and Predictive Analytics via Social Media Platforms," will examine the growing amount of tobacco-related content shared online and its potential impact on public health. Researchers have pointed to a strong link between social media promotion and rising tobacco use among young people, though the connection has not been thoroughly studied. The proposed project is inherently interdisciplinary, spanning AI and machine learning (Luu), public health (Dobbs), privacy-preserving machine learning (Khan and Malik) and societal impact assessment specialists (McFadden).

The team will design artificial intelligence tools to detect tobacco promotion and assess its influence. Their approach focuses on fairness, privacy and adaptability. Plans include building multimodal AI models, utilizing vision-language transformers and applying debiased learning methods to ensure accurate results across diverse communities.

To support the work, the project will create a large-scale dataset of tobacco-related content and introduce a privacy-preserving learning system that uses differential privacy and secure multi-party computation. The team will also develop a continual learning system incorporating human oversight to meet ethical standards while adjusting to new online trends.

All software developed will be deployed through cloud services, offering researchers and public health officials practical and effective tools. The results are expected to include more accurate detection systems and portable AI applications that can decrease youth exposure to tobacco promotion nationwide.