A New Ally for Doctors: AI That Detects Brain Emergencies In Seconds
- Feb 16
- 3 min read
A new artificial intelligence called Prima can analyze brain MRI scans in seconds. It identifies neurological diseases, recognizes emergencies, and helps doctors act faster. The new technology can reduce the burden on healthcare systems and significantly improve the care of neurological patients.
The analysis of brain images, such as MRI scans, is essential for diagnosing neurological diseases. However, the number of these exams grows every year, overloading hospitals, increasing waiting times for results, and exhausting doctors. This problem is even more serious in resource-poor regions, where there are fewer specialists available to interpret the exams quickly.
To address this challenge, researchers at the University of Michigan developed an artificial intelligence called Prima, capable of analyzing brain MRI scans in just a few seconds.
The system was trained with a huge amount of real data, including more than two hundred thousand exams, which allowed it to learn general and reliable patterns about brain function and diseases.
Prima was tested for a year in a large healthcare system, analyzing over thirty thousand exams. During this period, the artificial intelligence was able to identify more than fifty different types of important neurological disorders. Its performance was impressive, surpassing other modern artificial intelligence systems and achieving very high levels of accuracy.
In addition to identifying possible diagnoses, Prima can also assess the severity of cases. This means it can help quickly identify emergency situations, such as strokes or brain hemorrhages, that require immediate attention. In these cases, every minute can make the difference between life and death, or between recovery and permanent sequelae.
One of the system's major advantages is its ability to automatically alert the right professionals. If the exam indicates a serious problem, Prima can quickly notify a neurologist, neurosurgeon, or other appropriate specialist, allowing treatment to begin as soon as possible, often while the patient is still in the hospital.
Prima's operation is similar to the reasoning of an experienced radiologist. It doesn't just analyze images in isolation, but also considers the patient's medical history and the reason the exam was requested. This combination of information helps to form a more complete picture of the examined person's health status.
Unlike previous systems, which were trained only for specific tasks, Prima was developed to handle a wide variety of clinical situations. It learned from millions of images and real-world data accumulated over years, making it more flexible and closer to everyday medical practice.
Researchers also highlight that the system showed balanced results across different patient groups, reducing the risk of bias. This reinforces Prima's potential as a reliable tool to support physicians, improve workflow in hospitals, and make neurological care faster, more accurate, and more accessible.
READ MORE:
Learning neuroimaging models from health system-scale data
Yiwei Lyu, Samir Harake, Asadur Chowdury, Soumyanil Banerjee, Rachel Gologorsky, Shixuan Liu, Anna-Katharina Meissner, Akshay Rao,
Chenhui Zhao, Akhil Kondepudi, Cheng Jiang, Xinhai Hou, Rushikesh S. Joshi,
Volker Neuschmelting, Ashok Srinivasan, Dawn Kleindorfer, Brian Athey, Vikas Gulani, Aditya Pandey, Honglak Lee, and Todd Hollon
Nature Biomedical Engineering (2026), 118
Abstract:
Neuroimaging is a ubiquitous tool for evaluating patients with neurological diseases. The global demand for magnetic resonance imaging (MRI) studies has risen steadily, placing substantial strain on health systems, prolonging turnaround times and intensifying physician burnout. These challenges disproportionately impact patients in low-resource and rural settings. Here we utilize data from a large academic health system to develop Prima, an AI foundation model for neuroimaging that supports real-world, clinical MRI studies as input. Trained on over 220,000 MRI studies, Prima uses a hierarchical vision architecture that provides general and transferable MRI features. Prima was tested in a 1-year health system-wide study that included 29,431 MRI studies. Across 52 radiologic diagnoses from major neurologic disorders, Prima achieved a mean diagnostic area under the curve (AUC) of 92.0%, outperforming other state-of-the-art general and medical AI models. Prima offers explainable differential diagnoses, worklist priority for radiologists and clinical referral recommendations. Prima demonstrates algorithmic fairness across sensitive groups. These findings highlight the transformative potential of health system-scale AI training and Prima’s role in advancing AI-driven healthcare.
Comments