Machine Learning-Guided Discovery of NEU1-Targeting Natural Compounds with Relevance to Mitochondrial Dysfunction-Linked, Fatigue-Associated Neurodegeneration in Alzheimer’s Disease

Alzheimer’s disease (AD), the most common form of dementia, is characterized not only by amyloid-beta (Aβ) accumulation and tau pathology but also by mitochondrial dysfunction, reduced energy metabolism, oxidative stress, and fatigue-associated neuronal damage. Neuraminidase-1 (NEU1), a lysosomal sialidase involved in autophagy, neuroinflammation, and mitochondrial regulation, has recently emerged as a promising therapeutic target in AD. This study aimed to identify novel NEU1 inhibitors targeting mitochondrial dysfunction and fatigue-related pathways using an integrated computational drug discovery approach involving machine learning (ML), quantitative structure–activity relationship modeling, molecular docking, and molecular dynamics simulations. A curated dataset of 12,239 bioactive natural compounds was pre-processed and screened based on drug-likeness properties. Molecular descriptors were generated using PaDEL software (RRID:SCR_014272), and predictive models were developed using Support Vector Machine, Random Forest (RF), and Extreme Gradient Boosting algorithms. Among these, RF demonstrated the best predictive performance with a test accuracy of 82.9%, R 2 score of 0.6312, and root mean square error of 0.3221. Virtual screening against the NEU1 active site (PDB ID: 8DU5) identified several promising compounds, with 3-[1-(2,3-dihydro-1,4-benzodioxin-6-yl)-5-oxopyrrolidin-3-yl]-1-(3,4-dimethylphenyl) urea showing the highest binding affinity (−10.2 kcal/mol). Mitochondrial dysfunction and fatigue-associated neurodegeneration are discussed as literature-supported downstream consequences of NEU1-driven neuroinflammatory and desialylation pathways and direct mitochondrial or fatigue-related assays. Molecular dynamics simulations confirmed stable protein–ligand interactions with favorable RMSD, radius of gyration, and solvent accessible surface area profiles. Overall, this study highlights the potential of ML-guided screening in discovering NEU1 inhibitors targeting mitochondrial dysfunction and fatigue-associated neurodegenerative mechanisms in AD.