Early childhood caries (ECC) is the most common chronic disease in children worldwide. It often affects specific teeth more than others, a puzzle that has remained unsolved—until now. A team of researchers from the University of Hong Kong (HKU) Faculty of Dentistry, the Chinese Academy of Sciences (CAS-QIBEBT), Qingdao Stomatological Hospital, and Qingdao Women and Children’s Hospital has made a breakthrough. They created the world’s first artificial intelligence (AI) system that predicts the risk of tooth decay in individual teeth based on microbial data. This system achieves more than 90% accuracy. Their findings were published in the journal Cell Host & Microbe.
The project was led by Professor Shi Huang, Assistant Professor of Microbiology at HKU’s Faculty of Dentistry. Other key team members include PhD student Yufeng Zhang (HKU), Professor Jian Xu (CAS-QIBEBT), Dr. Fei Teng (Qingdao Stomatological Hospital), and Dr. Fang Yang (Qingdao Women and Children’s Hospital).
The team conducted the largest study to date on the microbial communities found on specific teeth in children aged 3 to 5 years. They used advanced methods, combining 16S rRNA sequencing with shotgun metagenomics to analyze both the composition and function of microbes. Over nearly a year, they collected 2,504 plaque samples from 89 preschoolers. Their work revealed distinct microbial patterns that signal early signs of tooth decay.
A key discovery was a natural gradient of microbes from the front to the back of the mouth. Healthy front teeth (incisors) have different bacteria than back teeth (molars). This spatial pattern is shaped by factors like saliva flow and tooth shape.
However, this gradient breaks down as cavities start to develop. The researchers saw that certain bacteria shift locations, moving from front to back teeth or vice versa, well before cavities become visible.
Building on this, the researchers developed Spatial-MiC, an AI system that predicts cavity risk in individual teeth by analyzing complex microbial patterns. Spatial-MiC looks at a tooth’s microbes and those of its neighboring teeth to assess risk. The system detected existing cavities with 98% accuracy. It predicted cavities two months before clinical signs appeared with 93% accuracy. This is a significant improvement over current whole-mouth evaluation methods, which often fail to detect early decay.
The impact of this technology could be huge for children’s dental health. In China, over 70% of five-year-olds suffer from ECC, and it remains the most common chronic disease in children worldwide. Today’s prevention methods treat all teeth the same, ignoring that some teeth are more vulnerable. This research opens the door to precision dentistry, where care is targeted to the teeth at highest risk before damage occurs.
“These findings change how we understand tooth decay,” Professor Huang said. “We are shifting from seeing cavities as unavoidable to being able to predict and prevent them tooth by tooth at the microbial level.”
The team hopes to test Spatial-MiC in other populations and develop clinical tools for dental clinics globally. Dr. Fang Yang, first author of the study, said, “This work is about more than better dental care. It’s about giving children healthier lives by preventing pain, infection, and developmental problems caused by severe tooth decay—more precisely and effectively.”