We introduce an infrared metasurface sensor based on plasmonic surface-enhanced infrared absorption spectroscopy combined with immunoassay, which detects alpha-synuclein, an early structural biomarker protein for neurodegenerative diseases with clinical specificity and identifies its different structural species using their unique spectroscopic signals. Unprecedentedly, we augmented the sensor with a Deep Neural Network, enabling quantitative differentiation of alpha-synuclein aggregates. Capable of multiplexing and retrieving aggregate signatures from complex biomatrix, our sensor shows promise for neurodegenerative disease diagnosis, disease progression monitoring, and drug efficacy assessment.