Researchers studying Alzheimer’s disease have found it difficult to determine which of the complex molecular and cellular changes in a patient’s brain are related to specific symptoms, making it challenging to pinpoint the best treatments for each individual.
A team led by Daifeng Wang, PhD, an associate professor of biostatistics and medical informatics in the University of Wisconsin School of Medicine and Public Health and an investigator at UW’s Waisman Center, used an emerging artificial intelligence model called graph neural networks to create two new tools that aim to help link changes in brain cells to Alzheimer’s symptoms, and potentially improve diagnosis and identify therapeutic targets. The tools, called PASCode and iBrainMap, identify specific genes and cells involved in disease progression and cognitive resilience to create personalized roadmaps of patients’ brains.
PASCode helps researchers identify which cellular changes and alterations in the way genes are expressed may be linked to specific Alzheimer’s symptoms, as well as changes that may contribute to resilience to cognitive decline. During the study, which analyzed brain tissue from nearly 1,500 autopsy donors, Wang said that the goal was to correlate the clinical information, such as symptoms, with cellular information. PASCode assigned each cell a score that reflects how strongly it is associated with specific Alzheimer’s symptoms. The stronger the score, the more likely the cell is involved in disease-related changes.
With their new iBrainMap, researchers built a personalized brain gene roadmap for each of the autopsy donor brains used in the study. The roadmap identifies cell-type specific genes that may influence disease progression and indicates whether those genes are associated with specific symptoms, including weight loss, insomnia and depression. The new personalized roadmap improved classification of healthy individuals versus individuals with Alzheimer’s disease by 21%.
The tools could be an important first step toward identifying personalized treatment targets.
