One of the most important ways in which artificial intelligence algorithms are proving beneficial is in diagnosing disease much faster than mere human beings can – and a new system for detecting potential indicators of Alzheimer's has reached close to 100 percent accuracy.

Through an analysis of fMRI brain scan images taken from 138 people, the new algorithm achieved an accuracy rate of over 99 percent. It performed better in terms of accuracy, sensitivity, and specificity than existing methods, the researchers report.

In particular, the method is able to pick out signs of mild cognitive impairment or MCI – which is the step between cognitive decline (as normally associated with aging) and Alzheimer's. Often, MCI won't come with any physical symptoms that can be spotted.

However, it's also important to note that MCI doesn't always necessarily mean Alzheimer's – but it's an important potential indicator of the disease in the future.

While manual analysis of MRI scans for signs of MCI is possible, humans are nowhere near as fast or reliable as deep learning techniques, which learn from vast databases of training data, then apply that knowledge to new data in intelligent ways.

"Modern signal processing allows delegating the image processing to the machine, which can complete it faster and accurately enough," says Rytis Maskeliūnas, an informatics professor from the Kaunas University of Technology (KTU) in Lithuania.

"Of course, we don't dare to suggest that a medical professional should ever rely on any algorithm 100 percent. Think of a machine as a robot capable of doing the most tedious task of sorting the data and searching for features."

Once the computer software has highlighted potential cases, specialists can then review and confirm them. An earlier diagnosis means earlier treatment, even if we're yet to discover a way of stopping Alzheimer's completely.

The AI model outlined in this new study is based on the existing ResNet18 neural network. The modified system was able to split brain scans into six categories, from healthy to full manifestations of Alzheimer's disease.

"Although this was not the first attempt to diagnose the early onset of Alzheimer's from similar data, our main breakthrough is the accuracy of the algorithm," says Maskeliūnas.

"Obviously, such high numbers are not indicators of true real-life performance, but we're working with medical institutions to get more data."

Various methods are used to detect Alzheimer's right now, including eye tracking, voice analysis, and even the installation of sensors in people's homes – but AI methods like the one outlined in this new study promise to be faster and simpler.

More than 78,000 fMRI scans were used to train and validate the model and hit the high accuracy rates, and the researchers say that their model could eventually be used to develop software that incorporates other data, including age and blood pressure.

Alzheimer's disease is the world's most frequent cause of dementia, contributing to some 70 percent of cases worldwide. Around 24 million people are currently thought to be affected globally, and as societies age, that figure is expected to rise sharply.

"Medical professionals all over the world attempt to raise awareness of an early Alzheimer's diagnosis, which provides the affected with a better chance of benefiting from treatment," says Maskeliūnas.

The research has been published in Diagnostics.