"Merlin" Assistant Diagnostic CT imaging could change the future of medicine A team of researchers from the US National Institutes of Health (NIH) has unveiled a new product that may revolutionize medical technology, an AI model called Merlin which can analyze 3D computed tomography and accurately predict disease risk.

A team of researchers from Stanford University used a database of more than 15,000 images, combined with almost 1 million radiation reports and diagnostic codes to train Merlin on the relationship between imaging and medical messages. The results were very impressive: Merlin was able to predict disease codes 81% accurately over other specific models and 90% accuracy in some cases.

In addition, using CT imaging to predict the risk of chronic diseases such as diabetes, heart disease or stroke is 75% accurate which suggests that AI can detect early signs of a condition that humans may not see interestingly Merlin still works well even with CT images of previously unseen organs like the spleen demonstrates its ability to learn across data limits researchers believe this technology will help diagnose illnesses faster, reduce medical radiation exposure and open doors for a new era of prescription medicine.Merlin could be used as a base for information and specialized training tools in future models

Akshay Chaudhari, Ph.D., team leader concluded that Merlin is the foundation for medical AI community from this point onwards, and sky is the limit This research was published in Nature magazine on March 4, 2026 under the title Merlin: A Computed Tomography Vision Language Foundation Model and Dataset

Composite image: Ai Grok