CQUniversity research tackles health misinformation in real time
Artificial intelligence is making users second-guess what they see across social media, but CQUniversity researchers are using AI to tackle health misinformation as it spreads through live-streamed content.
CQU Master of Research candidate Nethara Athukorala is developing a multimodal AI system that analyses video, audio and text together to detect and flag potentially misleading health and wellness claims in real time.
Her innovative approach is gaining global recognition, thanks to CQU initiatives to provide platforms for student research.
When health misinformation goes live
Live-streamed health content can reach audiences instantly, leaving little time for platforms to assess whether claims are accurate before viewers act on them.
“Health misinformation is increasingly spreading through live-stream media, where persuasive content can influence consumer decisions in real time,” Nethara said.
“Influencers, or anyone really, can broadcast health claims out live, faster than any moderator can review them.”
Nethara said misleading claims about health products could be particularly concerning when products are promoted as alternatives to established treatments without credible evidence.
“Products marketed as medical ‘dupes’, or duplicates of known products, may have no clinical backing, and vulnerable viewers can be influenced to buy them in real time, trusting the claim because it looks credible.”
Teaching AI to spot the signs
Nethara’s research brings together video, audio and text to help AI identify potentially misleading health information while a live stream is still running.
Her system reads three types of input from each video: Visual content, using the ResNet50 model; audio, using MFCC (a technique for capturing tone and vocal patterns) alongside a speech-to-text model called Whisper; and the transcribed text, processed using a Sentence Transformer model to help the system understand meaning, not just words.
"We're working to improve accuracy using more advanced models such as VideoMAE, alongside other transformer and CNN-based architectures... and with any model like this, it's still being built and refined, so this reflects ongoing, active research rather than a finished system," she explained.
The approach could help online platforms provide timely warnings to viewers and strengthen consumer protection against misleading information.
The research is supervised by Dr Mohammad Saiedur Rahaman, Associate Professor Jahan Hassan and Dr Arjun Neupane from CQU’s School of Engineering and Technology.
Dr Rahaman said responsible and trustworthy AI was central to the research.
“We are developing the technology to help platforms identify potentially misleading content in real time and provide timely warnings to viewers,” Dr Rahaman said.
The research has already gained international recognition, with Nethara winning the Best Overall Presentation Award at the Joint International Higher Degree Research Symposium 2026, hosted by CQU in collaboration with De La Salle University in the Philippines and Changwon University in South Korea.
“It was a pleasure to present alongside such high-quality research from across all three universities, and I look forward to continuing work that contributes to building safer and more trustworthy online information ecosystems,” Nethara said.
The enthusiastic student also recently won the 3-Minute Flash Pitch competition across the School of Engineering and Technology.
Associate Professor Hassan said Nethara’s research is tackling an important real-world challenge.
“Health misinformation is a growing challenge online, and Nethara’s research is exploring how AI can help identify misleading claims before they influence people’s decisions,” she said.
Explore research pathways at CQU’s Higher Degrees by Research website.
