Ultrasensitive SERS-LFA for the detection of neurofilament light chain and machine learning-assisted Alzheimer's disease classification
Elangovan Sarathkumar, Ramshekhar N. Menon, Ramapurath S. Jayasree
Abstract
= 0.97) and high specificity. To demonstrate clinical utility, spectral data from LFA test zones were acquired for plasma samples of AD patients, patients with mild cognitive impairment, and control cohorts. A machine learning pipeline integrating principal component analysis (PCA) and multilayer perceptron (MLP) classification enabled group-wise discrimination with 77.8% accuracy and macro-averaged precision, recall, and F1-score values of 0.83, 0.78, and 0.77, respectively. To our knowledge, this is the first report of a GNR@MQD-based nanohybrid applied in a competitive SERS-LFA for blood-based NfL detection and ML-assisted clinical differentiation. The proposed platform represents a cost-effective, portable, and intelligent diagnostic solution for early-stage, point-of-care screening of AD and other broader neurodegenerative conditions.