A new study in Nature Biomedical Engineering suggests that an AI-powered model developed at the University of Michigan can “read a brain MRI and diagnose a person in seconds.”
The model “detected neurological conditions with up to 97.5% accuracy and predicted how urgently a patient required treatment.” The invention, called Prima, was tested on “more than 30,000 MRI studies over the course of a year.” Prima outperformed other AI models on diagnostic performance tested against over 50 radiologic diagnoses. The model also succeeded in “determining which cases should take higher priority.”
Prima can “automatically alert providers so rapid action can be taken” in cases where immediate medical attention is required. It simultaneously processes video, images, and text, and it was trained on “every MRI…taken since radiology digitization began [at] University of Michigan Health decades ago.”
The device shows promise not just for brain MRIs but for “mammograms, chest X-rays, and ultrasounds” as well.
