An AI-assisted multiplex fluorescence sensing platform for grading diagnosis of Alzheimer's disease

Abstract

Blood-based biomarkers have become increasingly important for Alzheimer's disease (AD) diagnosis. However, due to individual variations, diagnostic accuracy using a single blood biomarker remains low, making it challenging to implement in large-scale AD screening efforts. Herein, we developed a multiplex fluorescent sensing platform for simultaneously measuring Aβ40, Aβ42, and P-tau181 in the blood, and constructed an artificial intelligence (AI) model. These three biomarkers were analyzed in 60 clinical samples: 15 healthy control, 15 subjective cognitive decline, 15 mild cognitive impairment, and 15 AD samples. The AI model based on these three biomarkers exhibited high predictive accuracy (91%), high positive predictive value (PPV) and low false rate (8.8%). The diagnostic accuracy and PPV of the AI model exceeded 90% for AD grading diagnosis in clinical samples. This study introduces a promising strategy for disease diagnosis and grading based on multi-biomarker analysis.