Using artificial intelligence technologies, they created an algorithm that analyzes electroretinogram (ERG) signals. It turns out that these signals can be used not only to identify eye problems but also to detect signs of various nervous system disorders. Moreover, the algorithm doesn't simply say "is there a disease?"—it shows which parts of the signal are important, helping doctors better understand the results and decide whether the patient needs further examination.
This method has already been tested on real data from people with and without various diagnoses. To train the model, the scientists used specialized approaches that allow the algorithm to explain why it reached a certain conclusion—this helps doctors trust the results.
Interestingly, unlike complex neural networks typically used to analyze such data and which require powerful computers, the developed algorithms are simpler, faster, and less demanding on technology. This means doctors can use them without expensive equipment, saving time and money, while still obtaining accurate preliminary results.
In the future, scientists plan to improve the algorithms so that they can help detect other eye and nerve diseases, such as night blindness or glaucoma.