Computational approaches to neurodegenerative diseases – Volume II

Neurodegenerative diseases remain among the hardest problems in neuroscience. They are multifactorial, they progress over decades, and the pathology that drives them is underway before symptoms appear. Computational neuroscience offers powerful ways to address these challenges. Mathematical modelling, machine learning, and large-scale simulations can bring together data from molecules, cells, brain networks, and clinical studies to track disease progression and test potential treatments.Volume I of this Research Topic brought together work that did exactly this. It spanned physics-informed neural networks for fractional-order modelling of Alzheimer’s disease, metaheuristic-optimised classification of disease stages, diagnostic modelling from sparse EEG channels, and gait-based prediction of cerebral small vessel disease. Taken together, the collection showed both the reach of data-driven modelling and where it reaches its limits: strong predictive performance on curated cohorts, but far weaker evidence that these models transfer to new populations, explain mechanisms, or inform clinical decisions.Volume II aims to help close this gap by reflecting the field’s progress since the first volume. For example, advances in blood-based biomarkers are enabling larger studies of people at earlier disease stages, particularly in Alzheimer’s disease, although their development varies considerably across neurodegenerative conditions. Similarly, the emergence of disease-modifying therapies for some disorders is making the prediction of treatment response, progression and risk an increasingly pressing clinical question. Foundation models and generative methods are being applied to imaging, EEG and electronic health records. Digital twins and whole-brain simulation are maturing into testable frameworks, and wearable and passive sensing are producing continuous, real-world measures of decline. At the same time, expectations of computational work have risen. Reproducibility, external validation, interpretability, fairness across populations, and privacy-preserving analysis across sites are no longer optional extras; they are what make a model credible.With this Research Topic, we aim to bring together computational work that models, analyses or predicts aspects of neurodegeneration, and that is explicit about mechanism, validation or clinical use. We particularly welcome studies that integrate experimental or clinical data with simulation, that test their models beyond the cohort on which they were trained, and that make their methods and data available to others.Themes of interest include, but are not limited to:Modelling disease mechanisms and progressionMultiscale and mechanistic models linking molecular pathology to network and behavioural declineNetwork spreading models of tau, amyloid and α-synuclein propagation across the connectomeData-driven disease progression modelling, subtyping and staging across long time horizonsWhole-brain simulation and digital twins as testbeds for interventionBiomarkers, diagnosis and predictionMultimodal integration of neuroimaging, EEG/MEG, genomics, fluid biomarkers and clinical recordsEarly and differential diagnosis, including prodromal and preclinical stagesDigital biomarkers from wearables, gait, speech, eye movements and passive monitoringPredicting progression, treatment response and adverse events, including in the context of emerging disease-modifying therapiesMethods, trust and translationFoundation models, generative and self-supervised approaches applied to neurodegenerative dataInterpretable and explainable modelling that yields mechanistic insight, not only accuracyFederated and privacy-preserving learning across sites and countriesExternal validation, benchmarking, reproducibility, and open data and codeFairness and generalisation across populations that remain underrepresented in existing cohortsComputational frameworks for therapeutic discovery, target identification and trial design Topic editors: Miodrag Zivkovic, HengJin Ke, Milos Antonijevic Source: https://www.frontiersin.org/research-topics/87507/computational-approaches-to-neurodegenerative-diseases---volume-ii