Researchers at Worcester Polytechnic Institute (WPI) employ machine learning to examine brain anatomy and forecast Alzheimer’s disease with 92.87% accuracy. The study highlights how brain volume loss varies by age and sex, offering new insights into early detection.
Challenges in Early Alzheimer’s Diagnosis
Alzheimer’s disease, a progressive neurodegenerative condition, disrupts cognitive functions and leads to neuronal death. Approximately 6.9 million Americans aged 65 and older live with the disease. Early symptoms often mimic normal aging, complicating diagnosis. Machine learning overcomes this by processing vast MRI data to spot subtle changes predictive of Alzheimer’s and related cognitive decline.
“Early diagnosis of Alzheimer’s disease can be difficult because symptoms can be mistaken for normal aging,” states Benjamin Nephew, assistant research professor in the Department of Biology and Biotechnology at WPI. “We found that machine-learning technologies can analyze large amounts of data from scans to identify subtle changes and accurately predict Alzheimer’s disease and related cognitive states.”
Study Methodology
Nephew, PhD student Senbao Lu, and recent MS graduate Bhaavin Jogeshwar analyzed 815 MRI scans from the Alzheimer’s Disease Neuroimaging Initiative. These scans, from individuals aged 69 to 84, represent normal cognition, mild cognitive impairment, and Alzheimer’s disease.
The team first measured volumes in 95 brain regions using machine learning. An algorithm then predicted disease states by comparing healthy brains to those with impairment, achieving high accuracy across groups.
Key Brain Regions Implicated
Volume reductions in the hippocampus, amygdala, and entorhinal cortex emerge as strongest predictors, regardless of age or sex. The hippocampus supports memory and learning, the amygdala manages emotions, and the entorhinal cortex handles memory, navigation, and perception—often hit first in Alzheimer’s.
Sex and Age Variations
In the youngest group studied (ages 69-76), both males and females show volume loss in the right hippocampus, underscoring its role in early detection.
Sex-specific patterns appear elsewhere: females exhibit loss in the left middle temporal cortex, linked to language, memory, and visual perception. Males show notable reductions in the right entorhinal cortex. “The degree of these differences was surprising,” Nephew notes, suggesting ties to age-related hormone shifts like declining estrogen and testosterone.
Toward Generalizable Models and Future Work
“The critical challenge in this research is to build a generalizable machine-learning model that captures the difference between healthy brains and brains from people with mild cognitive impairment or Alzheimer’s disease,” Nephew explains. “A generalizable model means that the biomarkers we found are not unique to this dataset but could be universal to all patients with mild cognitive impairment or Alzheimer’s.”
The team advances with deep learning models and factors like diabetes. WPI’s interdisciplinary approach draws students from biology, neuroscience, psychology, computer science, and bioinformatics.
“This research exemplifies the strength of neuroscience at WPI, which is interdisciplinary and computational,” Nephew adds. “The brain is an extremely complicated organ, and we need to think broadly about how to better understand, predict, and treat the diseases that afflict the brain.”
Findings appear in Neuroscience (DOI: 10.1016/j.neuroscience.2025.12.030).
