Machine learning IDs biomarkers for Alzheimer’s disease

Researchers at the University of Southern California Viterbi School of Engineering have utilized machine learning to detect potential blood-based markers of Alzheimer’s disease. This will allow for earlier diagnosis and could potentially lead to non-invasive methods of tracking the progression of the disease in impacted patients. Findings were published in the journal Frontiers in Aging Neuroscience.

There is wide interest in how machine learning can provide more information into cognitive diseases including Alzheimer’s. Recently, the National Institute on Aging awarded a total of $5 million for two projects that will use AI and big data to better understand Alzheimer’s disease and other dementias.

Led by professor Greg Ver Steeg, PhD, of the USC Information Sciences Institute, the researchers sought to determine whether there could be hidden indicators of Alzheimer’s—namely, factors that are discovered in routine blood tests. Steeg and colleagues used his machine learning method CorEx, which can find patterns in areas that are “overwhelmed by large amounts of data.”

It is possible there is not a single predictor that could determine cognitive decline, if it’s underway, or the severity of it. Still, there may be clusters of indicators that may better detect cognitive decline.

“The question we were looking at was whether we could use the algorithm to find groups of features that could be a better predictor of Alzheimer’s than any factors measured individually,” Steeg said in a prepared statement. 

Steeg and colleagues assessed the medical data from more than 800 adults who are part of the Alzheimer’s Disease Neuroimaging Initiative to identify predictors of cognitive decline. The cohort fell into three categories—cognitively normal, mild cognitive impairment and those with Alzheimer’s disease. 

The researchers found more than 400 biomarkers that were collected from brain imaging, genetics, plasma and demographics. They found amyloid and tau to be “important.” Additionally, the machine learning showed a “strong relationship” with cardiovascular health, hormones, metabolism and immune system.

“Maybe fixing one of these things alone doesn’t make a huge difference but fixing a cluster of things could be helpful in reducing the risk of developing the disease," Steeg et al. wrote.

Professor and co-researcher Paul Thompson, of the USC Mark and Mary Stevens Neuroimaging and Informatics Institute and the Keck School of Medicine, noted some of the relationships between the measures had been previously established as being associated with Alzheimer’s. Their results show clusters of features could be a stronger predictor of Alzheimer’s, compared to individual features. 

The researchers will look to confirm their initial results in a larger cohort and will use CorEx to find hidden factors in mental health diseases including schizophrenia and depression.

""

As a senior news writer for TriMed, Subrata covers cardiology, clinical innovation and healthcare business. She has a master’s degree in communication management and 12 years of experience in journalism and public relations.

Around the web

Cardiovascular devices are more likely to be in a Class I recall than any other device type. The FDA's approval process appears to be at least partially responsible, though the agency is working to make some serious changes. We spoke to a researcher who has been tracking these data for years to learn more. 

Updated compensation data includes good news for multiple subspecialties. The new report also examines private equity's impact on employment models and how much male cardiologists earn compared to females.

When drugs are on the FDA’s shortage list, outsourcing facilities can produce their own compounded versions. When the FDA removed tirzepatide from that list with no warning, it created a considerable amount of chaos both behind the scenes and in pharmacies all over the country. 

Trimed Popup
Trimed Popup