An AI learned to spot rabies in a microscope slide — and it is right 99.6% of the time

The University of Surrey's Institute for People-Centred AI and the Animal and Plant Health Agency have built the first AI workflow that detects rabies infection in cell cultures and measures protective antibody levels automatically, replacing slow manual review of microscope images. Trained on more than 2,300 fluorescence images and 36,000 annotated sections, the best model classified 99.6% correctly, in a peer-reviewed paper in Nature Scientific Reports. Rabies still kills about 59,000 people a year and is nearly always fatal once symptoms appear — so faster detection saves lives.