How old is your brain? Artificial intelligence accurately predicts human cognitive function and Alzheimer's disease

Brain aging is considered a reliable biomarker of neurodegenerative disease risk
.
This risk increases
when a person's brain exhibits characteristics that are "older" than expected by their peers.
By harnessing the deep learning capabilities of the team's novel AI model to analyze the scans, the researchers were able to detect subtle anatomical markers of the brain that would otherwise be difficult to detect and that have been linked to
cognitive decline.
Their findings, published in the Proceedings of the National Academy of Sciences on Tuesday, Jan.
2, provide an unprecedented perspective
on human cognition.

"Our study harnesses the power of deep learning to identify aging areas of the brain that reflect cognitive decline that may contribute to Alzheimer's," said Andrei Irimia, assistant professor of gerontology, biomedical engineering, quantitative and computational biology and neuroscience at the University of Southern California's Leonard Davis School of Geriatrics, who also corresponding author
of the study.

"People age at different rates and have different
types of body tissue.
We colloquially say, 'So-and-so is forty, but looks thirty
.
' The same idea applies to the brain
.
A 40-year-old's brain may look 'young' as a 30-year-old or 'old'
as a 60-year-old.
”

A more precise alternative than existing methods

Irimia and his team collated the brain MRIs of 4681 cognitively normal participants, some of whom developed cognitive decline or Alzheimer's disease
later in life.

Using this data, they created an artificial intelligence model called a neural network that predicted the age
of participants through MRI of the brain.
First, the researchers trained the network to generate detailed maps of brain anatomy to reveal aging patterns
on specific topics.
They then compared
the perceived (biological) brain age of the study participants with their actual (temporal) age.
The greater the difference between the two, the worse the participants' cognitive scores, which reflect the risk of Alzheimer's

The results showed that the team's model could predict the true (chronological) age of cognitively normal participants with an average absolute error of 2.
3 years, which is about 1 year
more accurate than existing award-winning brain age estimation models that use different neural network architectures.

"Explainable AI can be a powerful tool for assessing the risk of Alzheimer's disease and other neurocognitive diseases," said Irimia, who also holds faculty positions
at the USC Viterbi School of Engineering and the USC Dornsife College of Arts and Sciences.
"The sooner we identify people at high risk for Alzheimer's disease, the sooner clinicians can intervene in treatment options, surveillance, and disease management
.
What is particularly powerful about AI is its ability to capture the subtle and complex features of aging that other methods cannot, which is key to
identifying a person's risk years before they develop the disease.
”

Aging in the brain varies by gender

The new model also reveals sex differences
in aging in various regions of the brain.
Some parts of the brain in men age faster than women and vice versa
.
Men are at higher risk of sports injuries due to Parkinson's disease, and the brain's motor cortex (the area responsible for motor function) ages faster
.
The findings also suggest that in women, typical aging in the right hemisphere of the brain may be relatively slow
.

An emerging area of research shows the promise of personalized medicine

The scope of this work goes far beyond disease risk assessment
.
Irhimia envisions a world in which new deep learning methods developed as part of the research are being used to help people understand the overall rate at
which they age.

"One of the most important applications of our work is that it has the potential to pave the way for tailored interventions that address each person's unique pattern of aging," Irimia said
.
"A lot of people want to know their true aging rate
.
This information can give us some hints about what people can do to make different lifestyle changes or interventions to improve their overall health and well-being
.
Our approach can be used to design patient-centered treatment plans and personalized maps of brain aging, which may be of interest
to people with different health needs and goals.
”