Brain simulation augments machine-learning-based classification of dementia

Abstract

IntroductionWhile the prevalence of neurodegenerative diseases and dementia increases, our knowledge of the underlying pathomechanisms and related diagnostic biomarkers, outcome predictors, or therapeutic targets remains limited. In this article, we show how computational multi-scale brain network modeling using The Virtual Brain (TVB) simulation platform supports revealing potential disease mechanisms and can lead to improved diagnostics.MethodsTVB allows standardized large-scale structural connectivity (SC)-based modeling and simulation of whole-brain dynamics. We combine TVB with a cause-and-effect model for amyloid-beta, and machine-learning classification with support vector machines and random forests. The amyloid-beta burden as quantified from individual AV-45 PET scans informs parameters of local excitation/inhibition balance. We use magnetic resonance imaging (MRI), positron emission tomography (PET, specifically Amyloid-beta (Abeta) binding tracer AV-45-PET, and Tau-protein (Tau) binding AV-1451-PET) from 33 participants of Alzheimer’s Disease Neuroimaging Initiative study 3 (ADNI3). The frequency compositions of simulated local field potentials (LFP) are under investigation for their potential to classify individuals between Alzheimer’s disease (AD), Mild Cognitive Impairment (MCI), and healthy controls (HC) using support vector machines and random forest classifiers.ResultsThe combination of empirical features (subcortical volumetry, AV-45- and AV-1451-PET standard uptake value ratio, SUVR per region) and simulated features (mean LFP frequency per brain region) significantly outperformed the classification accuracy of empirical data alone by about 10% in the accuracy index of weighted F1-score (empirical 64.34% vs. combined 74.28%). There was no significant difference between empirical and simulated features alone. The features with the highest feature importance showed high biological plausibility with respect to the AD-typical spatial distribution of the features. This was demonstrated for all feature types, e.g., increased importance indices for the left entorhinal cortex as the most important Tau-feature, the left dorsal temporopolar cortex for Abeta, the right thalamus for LFP frequency, and the right putamen for volume.DiscussionIn summary, here we suggest a strategy and provide proof of concept for TVB-inferred mechanistic biomarkers that are direct indicators of pathogenic processes in neurodegenerative disease. We show how the cause-and-effect implementation of local hyperexcitation caused by Abeta can improve the machine-learning-driven classification of AD. This proves TVBs ability to decode information in empirical data by means of SC-based brain simulation.

Personalized modeling of neurodegeneration determines dementia severity from EEG recordings

“…We used as benchmark for EEG performances in classification of MCI state algorithms based on spectral and connectivity features, which are the most common tool to assess MCI conditions from EEG [11,[45][46][47][48][49][50][51][52] and are even investigated as possible markers of progression to AD [37]. The final result was in line with previous studies performing MCI classification with these features [9,10,13,14,53] but was vastly outperformed by the modelbased procedure (Figure 6). However, the range of possible EEG features extend beyond the ones we tested.…”

“…Novel approaches can improve not only classification with EEG features, but also improve the design personalized models. Previous works utilized model personalized with structural scans as [19] to derive novel features to support classification of AD and MCI patients [53,58]. Beside the difference in the approach highlighted in the previous section, we also note that in our case network model features replace features based on recordings in the classification algorithm, rather than being combined with them, nonetheless enabling us to obtain a much higher accuracy in discriminating healthy and MCI subject.…”

Probabilistic Inference on Virtual Brain Models of Disorders

“…It emphasizes the incorporation of domain knowledge (Gelman et al, 2013), hypothesis testing (MacKay, 2003), interpretability (Hastie et al, 2009), generalizability (Vapnik, 1999), and causality (Pearl, 2009). It also often provides superior performance over purely data-driven methods, particularly in the context of brain disorders, such as epilepsy (Hashemi et al, 2020;Wang et al, 2023;Jirsa et al, 2023), and Alzheimer's disease (Triebkorn et al, 2022;Yalcinkaya et al, 2023). One class of computational network models commonly used to analyze functional neuroimaging modalities, such as fMRI, MEG, and EEG, is the class of connectome-based models (Ghosh et al, 2008;Sanz-Leon et al, 2015;Bassett and Sporns, 2017), also known as Virtual Brain Models (VBMs; Sanz Leon et al (2013); Jirsa et al (2023)).…”

Virtual brain simulations reveal network-specific parameters in neurodegenerative dementias

“…Importantly, all neurological conditions involve changes at multiple scales and can gain from the use of TVB for understanding the impact of cellular and microcircuit properties alterations on brain function. The promise for clinical use of TVB has been already suggested in epilepsy surgery [8], stroke [9], brain tumors [10], Multiple Sclerosis [11] and neurodegenerative conditions like dementia [12][13][14][15]. Interestingly, the central position of an E/I imbalance in the cascade of pathophysiological events in AD is increasingly recognized [16].…”