A team of scientists from the University of Bari creates a machine learning system capable of detecting Alzheimer’s symptoms in the brain before it is clinically evident.
The research of a team of scientists from the University of Bari, in Italy, could represent a leap in Alzheimer’s prevention. Like all works that come from the experimental field, it must be taken with caution. The conclusions at the moment are only preliminary and, in any case, it will take time for such a system to be available in the medical field.
But this does not detract from the merit of the scientists at the University of Bari. These have developed an artificial intelligence system capable of detect Alzheimer’s ten years in advance. It does this by studying the patient’s brain, detecting mild cognitive impairments.
The team of scientists has also done so using a non-intrusive technique. It is simply a question of analyze brain scans in detail obtained by magnetic resonance.
The artificial intelligence system developed by the researchers belongs to the category of machine learning. It is therefore framed within algorithms capable of learning from the information to which they have access. But that takes training. Project managers they nurtured their software with 67 brain magnetic resonances****. Of these, 38 belonged to Alzheimer’s patients, while another 29 belonged to people without the disease.
To raise the accuracy of the study, scientists they divided the resonance images into small parts. In this way the system analyzes the brain region by region, to evaluate the neuronal connectivity between them.
From this training, the algorithm was shown a set of brain scans. Specifically, there were 148, of which 52 belonged to brains without the disease, 48 with Alzheimer’s and 48 with mild cognitive impairment. The latter is a symptom of Alzheimer’s. And it usually leads to the disease within three to nine years.
The algorithm was able to distinguish between diseased and diseased brains with 86% accuracy. While differentiating between healthy brains and those who had mild cognitive impairment it was successful by 84%.
Given the importance of mild cognitive impairment in Alzheimer’s, it is thought that the algorithm could diagnose the disease up to 10 years in advance. The software would be able to identify small brain changes that pre-empt clinical symptoms. From that moment on it could work to delay the disease.
Images: Geralt, GR_Image