A recent study conducted by researchers at West Virginia University has unveiled a series of diagnostic metabolic biomarkers, paving the way for the development of artificial intelligence (AI) tools aimed at detecting Alzheimer's disease at its initial stages, as well as identifying risk factors and potential treatment strategies.
The study sought to pinpoint the most relevant metabolic biomarkers associated with Alzheimer's disease, subsequently training AI models to predict the likelihood of disease occurrence or progression.
AI Predicts and Identifies Biomarkers
Led by Kesheng Wang, a professor at the WVU School of Nursing, the study capitalized on deep learning techniques within AI due to their adaptability in forecasting intricate biological phenomena, harnessing large datasets, and intricate algorithms for model training.
In the realm of medicine, biomarkers serve as quantifiable indicators of disease presence or severity, akin to the numbers reflecting cholesterol or glucose levels in routine blood tests.
Metabolic biomarkers, on the other hand, delve into the molecular intricacies of cells, tissues, and bodily fluids, shedding light on the interplay between genetics and lifestyle factors such as dietary habits and environmental influences, according to the research team.
Alzheimer's disease often manifests years or even decades before the onset of clinical symptoms, underscoring the importance of identifying predictive biomarkers in the preclinical phase to facilitate early intervention strategies and disease prevention efforts.
Moreover, early detection holds paramount significance in the realm of drug development, diagnostic accuracy, and therapeutic interventions aimed at mitigating functional decline and enhancing longevity.
The study drew upon data from the Alzheimer's Disease Neuroimaging Initiative, encompassing 78 individuals diagnosed with Alzheimer's disease and 99 cognitively normal counterparts aged between 75 to 82 years.
Deep Learning for Alzheimer's Detection
Leveraging LASSO software, researchers scrutinized 150 metabolic biomarkers, eventually isolating 21 markers deemed most pertinent to Alzheimer's disease, spanning glucose, amino acid, and lipid metabolism pathways.
Several of these biomarkers exhibited correlations with established clinical metrics such as plaque accumulation, cognitive performance measures, and hippocampal volume, a region frequently impacted in Alzheimer's pathology.
Subsequent analyses entailed testing various deep learning models until researchers identified the one yielding the highest accuracy for diagnostic assessment.
While the application of deep learning in Alzheimer's detection remains in its nascent stages, further investigations are imperative. Wang and his team are currently exploring an integrative approach that combines data from proteins and metabolism through deep learning methodologies.
"The metabolic basis of Alzheimer's disease is still poorly understood and the relationships between systemic abnormalities in metabolism and Alzheimer's disease pathogenesis are unclear," Wang said in a statement.
"This study shows there is potential to identify metabolic biomarkers that are predictive of Alzheimer's disease diagnosis and progression."
The study's findings were published in the Journal of the Neurological Sciences.
Related Article: Gum Diseases and Tooth Loss Linked to Brain Shrinkage, Higher Risk of Dementia, New Study Shows
