New AI Catches Early Mouth Cancer Signs Before They Spread

A team of researchers has developed a new AI system that could help dentists catch early warning signs of mouth cancer. The tool, called M2-OPMDNet, looks at two types of oral images along with a patient's health history and risk factors, like smoking or symptoms. In tests, it correctly identified dangerous mouth lesions about 95% of the time, performing better than AI that only looks at images.

What makes this different is how the AI combines information the way a real doctor would. Instead of relying only on visual clues, it also considers a short questionnaire about the patient's lifestyle and symptoms. This matters because many harmless mouth sores look nearly identical to early cancerous ones. The AI even flagged visually subtle cases that are easy for the human eye, or image-only software, to miss.

The system is also "explainable," meaning it can show which factors influenced its decision, whether that was an abnormal bright spot in the image or a patient's smoking history. That transparency helps doctors trust the technology and understand why it recommends a biopsy or a closer look.

There is still work to do before this becomes a standard dental tool. The study was done under real-world screening conditions, but it has not yet been tested in large clinics or approved for clinical use. Still, the results are promising: an affordable, non-invasive way to screen for oral cancer could lead to earlier treatment and better outcomes for thousands of people.