A study published in Nature's journal NPJ Digital Medicine revealed that new Artificial Intelligence (AI) software is capable of identifying strokes with twice the accuracy compared to traditional methods. Developed by researchers at Imperial College London, Technical University of Munich and Edinburgh University, the model can also estimate the exact time of stroke and the biological age of brain lesions, crucial information for more effective treatment.
The software stands out for overcoming challenges in identifying the moment the stroke occurred, a critical step in determining the viability of treatments. Currently, doctors assess the degree of brain damage by analyzing CT scan images, but the process is prone to errors. AI, on the other hand, automates and refines this analysis, making it significantly more accurate.
According to Paul Bentley from the Department of Brain Sciences at Imperial College London, “Not only is the software twice as efficient at reading stroke onset time, it can also be integrated into analytical systems already used in hospitals. This will allow doctors to make faster and more accurate decisions in emergency situations.”
Time is of the essence in stroke treatment. For patients within four to five hours of spillage, medical and surgical treatments are viable options. Between four, five and six hours, care becomes more limited, and, beyond this period, many cases become irreversible.
Adam Marcus, lead author of the study, believes that implementing the software can significantly change this scenario. “We estimate that up to 50 percent more patients could receive appropriate treatments thanks to the accuracy of our model.”
With the ability to identify whether brain damage can be reversed and determine the onset of stroke, the software promises to save thousands of lives and improve the recovery of countless patients. Researchers are now working to integrate the technology into the NHS, the British healthcare system, and expand its use globally.
(By Rafael Langa)
