Systems-Level Modeling of Neurodegenerative Diseases

Abstract

Neurodegenerative diseases -- Alzheimer's disease (AD), Parkinson's disease (PD), amyotrophic lateral sclerosis (ALS), Huntington's disease (HD), and frontotemporal dementia (FTD) -- affect over 50 million people globally, represent the leading cause of dementia and disability in older populations, and share common pathological mechanisms (protein misfolding, mitochondrial dysfunction, neuroinflammation, synaptic dysfunction) despite distinct clinical presentations. Decades of reductionist single-gene, single-protein research have identified risk factors and therapeutic targets but have produced disappointingly few disease-modifying therapies -- the complexity of neurodegeneration requires systems-level models that integrate genetics, molecular pathology, cellular interactions, neural circuits, and clinical phenotypes. We present the Neurodegenerative Systems Modelling Framework (NSMF), evaluating five computational approaches - genome-wide risk network analysis, protein aggregation and spreading simulations, multi-scale neural circuit models, patient-derived iPSC digital twins, and longitudinal biomarker trajectory AI -- across four disease modelling tasks (disease progression prediction, therapeutic target identification, patient stratification, and clinical trial design optimisation). Our Neurodegenerative Modelling Score (NMS) measures biological mechanism fidelity, prediction accuracy, therapeutic insight generation, patient stratification validity, and translational relevance. Longitudinal biomarker trajectory AI achieves the highest NMS (0.926) through integration of multi-modal longitudinal data to predict disease progression and treatment response, while multi-scale neural circuit models achieve the highest mechanism fidelity (0.960) through biophysical simulation of neuronal dynamics and network-level pathology.