Biology subjects

Alexander, D. C.

Publications and source records attributed to Alexander, D. C..

At least 19 recordsLinked to original sources

bioRxiv2025

Microbiome-behavior coupling shapes infant adaptation to early maternal unpredictability

We asked whether infant microbiome composition, together with infants behavioral responses to early maternal unpredictability, can offer novel mechanistic insights into behavioral phenotypic variation in human development. Maternal unpredictability was computed as the entropy rate of transitions among maternal sensory signals during naturalistic mother-infant interaction (n=256 dyads; 2-6 months). Infant visual orienting behavior (VOB) was indexed by infants simultaneous gaze shifts to and from the mother. Higher maternal unpredictability predicted more frequent VOB shifts, indicating a mature-for-age profile, and longitudinally forecast poorer inhibitory control at 19-28 months. A subgroup of infants with high maternal unpredictability and high VOB had inhibitory control outcomes closer to those of infants who experienced low maternal unpredictability, suggesting VOB strategy might buffer against the effects of high maternal unpredictability. In participants with metagenomic data (n=93), VOB was associated with taxonomic and functional gut microbial profiles along a Bifidobacterium breve to Bifidobacterium longum axis. Neuroactive gene-set enrichment analysis linked VOB to increase in tryptophan synthesis and glutamate synthesis genes. In contrast, maternal unpredictability showed depletions of GABA and tryptophan synthesis in the unique-effects model and, when VOB and the maternal unpredictability by VOB term were included, enrichment of acetate synthesis and quinolinic acid degradation. Notably, the maternal unpredictability by VOB interactions for tryptophan and glutamate were negative, indicating that the VOB-pathway coupling attenuates as unpredictability increases. Together, these findings support a framework in which maternal unpredictability is the environmental challenge, infant VOB expresses the behavioral strategy, and microbial metabolism modulates whether that strategy is biochemically feasible. TeaserInfant gut microbiome is linked to infant behavioral responses to maternal unpredictability, which in turn impacts how they learn to control their actions.

neuroscience↗

bioRxiv2024

Phenotypic diversity and shared genomic determinants among isolates causing a large incidence of disseminated gonococcal infections in Canada

The incidence of disseminated gonococcal infection (DGI) has remained low since the advent of antibiotics, however recent surge in DGI have inexplicably emerged within several regions during the past decade. In an effort to understand whether Neisseria gonorrhoeae that cause disseminated disease can be differentiated from non-invasive strains, we have performed a phenotypic and genotypic analysis on a selection of isolates obtained from invasive and uncomplicated infections in Canada. Phenotypic analysis of a matched subset of 19 isolates obtained since 2013 found that these varied in their capacity to aggregate in suspension and in their association with serum complement proteins, however these interactions did not discriminate between the invasive and mucosal isolates. Sequence typing of 360 Canadian isolates revealed that two porB alleles are significantly associated with the DGI strains, one of these being present throughout the past decade whereas the other became associated more recently. A PopNet-based population dynamics analysis, which instead establishes relationships based upon variance among discrete chromosomal segments, found that DGI isolates were restricted in their phylogenetic distribution. While this implies a genetically-linked potential to cause invasive disease, it cannot distinguish between an inherent difference in the phenotype of these populations or the horizontal exchange of some virulence factor among closely related strains. Regardless, a large number of genetic determinants are enriched in the DGI strains, making these enticing candidates for future work to understand how they might either promote the gonococcal capacity to cause systemic infection or reduce the presentation of clinical symptoms from localized infection so that it remains untreated. AUTHOR SUMMARYNeisseria gonorrhoeae is a sexually transmitted bacteria that causes over 82 million cases of gonorrhea each year. With its re-emergence, rising incidence rates and the prevalence of multidrug resistant strains increasing, the bacteria is considered to be a high burden threat to global public health. While disseminated gonococcal disease arising from untreated infections are uncommon, there have recently been regions with high incidence invasive disease. Here, we take advantage of the active ongoing collection of gonococcal isolates in Canada to perform a combined phenotypic and genotypic analysis that aims to understand whether certain strains are more often linked to invasive infections. We found that all disseminated isolates bound to the complement regulatory factors factor H and/or C4 binding protein to facilitate their resistance to the bactericidal activity of serum, but this was not sufficient to explain the heightened instance of invasive disease. While classical genome-based phylogenetic analysis displayed little association between invasive strains, PopNet-based analysis revealed that invasive isolates fell within defined sub-populations and indicated variant alleles enriched among the disseminating bacteria, and a broader pan-genome approach revealed genes more likely to be present in invasive strains. Our study thereby provides support for a genetic contribution to the invasive potential of gonococcal isolates and provides candidate drivers of this virulent outcome.

microbiology↗

bioRxiv2024

ACSS2 contributes to transcriptional regulation in Cajal-Retzius cells in a mouse model of Alzheimer's disease

Dysregulation of histone acetylation in the brain has emerged as a major contributor to human Alzheimers disease (AD). The mechanisms by which these protective or risk-conferring epigenetic marks are established and maintained are under intense investigation. ACSS2 (Acetyl-CoA Synthetase 2) is a key metabolic enzyme that is chromatin-associated in neurons. ACSS2 is recruited to specific promoters and generates a local pool of acetyl-CoA from acetate, thereby fueling histone acetylation and driving the expression of neuronal genes that regulate learning and memory. Here, we examine the contribution of ACSS2-mediated histone acetylation to AD-related molecular and behavioral outcomes. Using a mouse model of human pathological AD-Tau injection, we show that loss of ACSS2 exacerbates Tau-related memory impairments, while dietary supplementation of acetate rescues learning in an ACSS2-dependent manner. Combining state-of-the-art proteomic and genomic approaches, we demonstrate that this effect is accompanied by ACSS2-dependent incorporation of acetate into hippocampal histone acetylation, which facilitates gene expression programs related to learning. Further, we identify Cajal-Retzius neurons as a critical hippocampal neuronal population affected, exhibiting the largest epigenetic and transcriptional dysregulation. Overall, these results reveal ACSS2 as a key neuroprotective metabolic enzyme, dysregulation of which might play an important role in the etiology of human AD, and guide the development of future therapies for AD and related dementia.

neuroscience↗

bioRxiv2023

Decreased voluntary alcohol intake and ventral striatal epigenetic and transcriptional remodeling in male Acss2 KO mice

Metabolic-epigenetic interactions are emerging as key pathways in regulating alcohol-related transcriptional changes in the brain. We previously demonstrated that this is mediated by the metabolic enzyme Acetyl-CoA synthetase 2 (Acss2), which is nuclear and chromatin-bound in neurons. Mice lacking Acss2 fail to deposit alcohol-derived acetate onto histones in the brain and show no conditioned place preference for ethanol reward. Here, we explored the role of this pathway during voluntary alcohol intake. We found that Acss2 KO mice consumed significantly less alcohol during drinking-in-the-dark, and this effect was primarily driven by males. We performed genome-wide transcriptional profiling of 7 key brain regions implicated in alcohol and drug use, and found that, following drinking, Acss2 KO mice exhibited blunted gene expression in the ventral striatum, and similar to the behavioral differences, transcriptional dysregulation was more pronounced in male mice. Further, we found that the gene expression changes were associated with depletion of ventral striatal histone acetylation (H3K27ac) in Acss2 KO mice compared to WT. Taken together, our data suggest that Acss2 plays an important role in orchestrating ventral striatal epigenetic and transcriptional changes during voluntary alcohol drinking, especially in males. Consequently, targeting this pathway could be a promising new therapeutic avenue.

neuroscience↗

bioRxiv2022

The variegation of human brain vulnerability to rare genetic disorders and convergence with behaviorally defined disorders

Diverse gene dosage disorders (GDDs) increase risk for psychiatric impairment, but characterization of GDD effects on the human brain has so far been piecemeal and lacked simultaneous analysis of multiple brain features across different GDDs. Here, through multimodal neuroimaging of 3 aneuploidy syndromes (XXY, XYY, trisomy 21), we reveal considerable diversity in cortical changes across GDDs and imaging-derived phenotypes (IDPs). This variegation of IDP change underlines the limitations of studying GDD effects unimodally. Integration across all IDP maps reveals highly distinct architectures of cortical change in each GDD, along with partial coalescence onto a common spatial axis of cortical vulnerability. This common axis shows strong alignment with shared cortical changes in behaviorally defined psychiatric disorders, and is enriched for specific molecular and cellular signatures - offering a high-priority target for future translational research.

neuroscience↗

bioRxiv2022

Accurate Bayesian segmentation of thalamic nuclei using diffusion MRI and an improved histological atlas

The human thalamus is a highly connected brain structure, which is key for the control of numerous functions and is involved in several neurological disorders. Recently, neuroimaging studies have increasingly focused on the volume and connectivity of the specific nuclei comprising this structure, rather than looking at the thalamus as a whole. However, accurate identification of cytoarchitectonically designed histological nuclei on standard in vivo structural MRI is hampered by the lack of image contrast that can be used to distinguish nuclei from each other and from surrounding white matter tracts. While diffusion MRI may offer such contrast, it has lower resolution and lacks some boundaries visible in structural imaging. In this work, we present a Bayesian segmentation algorithm for the thalamus. This algorithm combines prior information from a probabilistic atlas with likelihood models for both structural and diffusion MRI, allowing label boundaries to be informed by both modalities. We present an improved probabilistic atlas, incorporating 26 thalamic nuclei identified from histology and 45 white matter tracts identified in ultra-high gradient strength diffusion imaging. We present a family of likelihood models for diffusion tensor imaging, ensuring compatibility with the vast majority of neuroimaging datasets that include diffusion MRI data. The use of these diffusion likelihood models greatly improves identification of nuclei versus segmentation based solely on structural MRI. Dice comparison of 5 manually identifiable groups of nuclei to ground truth segmentations show improvements of up to 10 percentage points. Additionally, our chosen model shows a high degree of reliability, with median test-retest Dice scores above 0.85 for four out of five nuclei groups, whilst also offering improved detection of differential thalamic involvement in Alzheimers disease (AUROC 83.36%). The probabilistic atlas and segmentation tool will be made publicly available as part of the neuroimaging package FreeSurfer.

neuroscience↗

bioRxiv2022

S-EBM: Generalising event-based modelling of disease progression for simultaneous events

This study introduces the parsimonious event-based model of disease progression (P-EBM). The P-EBM generalises the event-based model of disease progression (EBM) to allow inference of fewer disease progression stages than the number of input biomarkers. The original EBM is designed to estimate a single distinct biomarker abnormality, termed an event, at each model stage. By allowing multiple events within a common stage, the P-EBM prevents redundant complexity to permit discovery of parsimonious sequences of disease progression - those that contain purely serial events, as in the original EBM, as well as those containing one or more sets of simultaneous events. This study describes P-EBM theory, evaluates its sequence estimation and staging performance and demonstrates its application to derive a parsimonious sequence of disease progression in sporadic Alzheimers disease (AD). Results show that the P-EBM can accurately recover a wider range of sequences than EBM under a range of realistic experimental scenarios, including different numbers of simultaneous events, biomarker disease signals and dataset sizes. The P-EBM sequence successfully highlights redundant biomarkers and stages subjects using fewer biomarkers. In sporadic AD, the P-EBM estimates a shorter sequence than the EBM with substantially higher likelihood which plausibly suggests that some biomarker events appear simultaneously. The P-EBM has potential application for generating new insights into disease evolution and for suggesting efficient biomarker collection strategies for patient staging.

neuroscience↗

bioRxiv2022

Targeting acetyl-CoA metabolism attenuates the formation of fear memories through reduced activity-dependent histone acetylation

Histone acetylation is a key component in the consolidation of long-term fear memories. Epigenetic enzymes involved in histone acetylation, including histone acetyltransferases and deacetylases, have been put forward as potential pharmacological targets in the treatment of pathological fear memories, such as those that underlie post-traumatic stress disorder (PTSD). However, these enzymes typically play a ubiquitous role in gene regulation, which precludes the clinical use of systemic manipulations. Recently, we have found that a nuclear-localized metabolic enzyme, Acetyl-coA synthetase 2 (Acss2), modulates histone acetylation during learning and memory. Loss of Acss2 is well-tolerated in mice, with no impact on general health or baseline behavior. Here, we show that an Acss2 null mouse model shows reduced acquisition of long-term fear memories in assays of contextual and cued fear conditioning. We find that loss of Acss2 leads to consolidation-specific reductions in both histone acetylation and the expression of critical learning and memory-related genes in the dorsal hippocampus. Further, we show that systemic administration of blood-brain-barrier (BBB)-permeable Acss2 inhibitors during the consolidation window reduces fear memory formation in mice and rats, and also reduces anxiety in a predator-scent-stress (PSS) paradigm. Our findings suggest that Acss2 plays a critical role in the formation of fear memories, and represents a potential pharmacological target in the treatment of PTSD.

neuroscience↗

bioRxiv2021

pySuStaIn: a Python implementation of the Subtype and Stage Inference algorithm

Progressive disorders are highly heterogeneous. Symptom-based clinical classification of these disorders may not reflect the underlying pathobiology. Data-driven subtyping and staging of patients has the potential to disentangle the complex spatiotemporal patterns of disease progression. Tools that enable this are in high demand from clinical and treatment-development communities. Here we describe the pySuStaIn software package, a Python-based implementation of the Subtype and Stage Inference (SuStaIn) algorithm. SuStaIn unravels the complexity of heterogeneous diseases by inferring multiple disease progression patterns (subtypes) and individual severity (stages) from cross-sectional data. The primary aims of pySuStaIn are to enable widespread application and translation of SuStaIn via an accessible Python package that supports simple extension and generalization to novel modelling situations within a single, consistent architecture. Code metadataCurrent code version v1.0 Permanent link to code/repository used of this code version https://github.com/ucl-pond/pySuStaIn Legal Code License MIT Code versioning system used git Software code languages, tools, and services used Python Compilation requirements, operating environments & dependencies Linux, Mac, Windows Support email for questions leon.aksman@loni.usc.edu, p.wijeratne@ucl.ac.uk, alexandra.young@kcl.ac.uk

neuroscience↗

bioRxiv2021

Training Data Distribution Significantly Impacts the Estimation of Tissue Microstructure with Machine Learning

PurposeSupervised machine learning (ML) provides a compelling alternative to traditional model fitting for parameter mapping in quantitative MRI. The aim of this work is to demonstrate and quantify the effect of different training strategies on the accuracy and precision of parameter estimates when supervised ML is used for fitting. MethodsWe fit a two-compartment biophysical model to diffusion measurements from in-vivo human brain, as well as simulated diffusion data, using both traditional model fitting and supervised ML. For supervised ML, we train several artificial neural networks, as well as random forest regressors, on different distributions of ground truth parameters. We compare the accuracy and precision of parameter estimates obtained from the different estimation approaches using synthetic test data. ResultsWhen the distribution of parameter combinations in the training set matches those observed in similar data sets, we observe high precision, but inaccurate estimates for atypical parameter combinations. In contrast, when training data is sampled uniformly from the entire plausible parameter space, estimates tend to be more accurate for atypical parameter combinations but may have lower precision for typical parameter combinations. ConclusionThis work highlights the need to consider the choice of training data when deploying supervised ML for estimating microstructural metrics, as performance depends strongly on the training-set distribution. We show that high precision obtained using ML may mask strong bias, and visual assessment of the parameter maps is not sufficient for evaluating the quality of the estimates.

neuroscience↗

bioRxiv2021

On the generalizability of diffusion MRI signal representations across acquisition parameters, sequences and tissue types: chronicles of the MEMENTO challenge.

Diffusion MRI (dMRI) has become an invaluable tool to assess the microstructural organization of brain tissue. Depending on the specific acquisition settings, the dMRI signal encodes specific properties of the underlying diffusion process. In the last two decades, several signal representations have been proposed to fit the dMRI signal and decode such properties. Most methods, however, are tested and developed on a limited amount of data, and their applicability to other acquisition schemes remains unknown. With this work, we aimed to shed light on the generalizability of existing dMRI signal representations to different diffusion encoding parameters and brain tissue types. To this end, we organized a community challenge - named MEMENTO, making available the same datasets for fair comparisons across algorithms and techniques. We considered two state-of-the-art diffusion datasets, including single-diffusion-encoding (SDE) spin-echo data from a human brain with over 3820 unique diffusion weightings (the MASSIVE dataset), and double (oscillating) diffusion encoding data (DDE/DODE) of a mouse brain including over 2520 unique data points. A subset of the data sampled in 5 different voxels was openly distributed, and the challenge participants were asked to predict the remaining part of the data. After one year, eight participant teams submitted a total of 80 signal fits. For each submission, we evaluated the mean squared error, the variance of the prediction error and the Bayesian information criteria. Most predictions predicted either multi-shell SDE data (37%) or DODE data (22%), followed by cartesian SDE data (19%) and DDE (18%). Most submissions predicted the signals measured with SDE remarkably well, with the exception of low and very strong diffusion weightings. The prediction of DDE and DODE data seemed more challenging, likely because none of the submissions explicitly accounted for diffusion time and frequency. Next to the choice of the model, decisions on fit procedure and hyperparameters play a major role in the prediction performance, highlighting the importance of optimizing and reporting such choices. This work is a community effort to highlight strength and limitations of the field at representing dMRI acquired with trending encoding schemes, gaining insights into how different models generalize to different tissue types and fiber configurations over a large range of diffusion encodings.

neuroscience↗

bioRxiv2020

Tau-first subtype of Alzheimer's disease consistently identified across in vivo and post mortem studies

Alzheimers disease (AD) is marked by the spread of misfolded amyloid-{beta} and tau proteins throughout the brain. While it is commonly believed that amyloid-{beta} abnormality drives the cascade of AD pathogenesis, several in vivo and post mortem studies indicate that in some subjects localized tau-based neurofibrillary tangles precede amyloid-{beta} pathology. This suggests that there may be multiple distinct subtypes of protein aggregation pathways within AD, with potentially different demographic, cognitive and comorbidity profiles. We investigated this hypothesis, applying data-driven disease progression subtyping models to post mortem immunohistochemistry and in vivo positron emission tomography (PET) and cerebrospinal fluid (CSF) based measures of protein pathologies in two large observational cohorts. We consistently identified both amyloid-first and tau-first AD subtypes, where tau-first subjects had higher levels of soluble TREM2 compared to amyloid-first subjects. Our work provides insight into AD progression that may be valuable for interventional trials targeting amyloid-{beta} and tau.

neuroscience↗

bioRxiv2020

Deep learning model fitting for diffusion-relaxometry: a comparative study

Quantitative Magnetic Resonance Imaging (qMRI) signal model fitting is traditionally performed via non-linear least square (NLLS) estimation. NLLS is slow and its performance can be affected by the presence of different local minima in the fitting objective function. Recently, machine learning techniques, including deep neural networks (DNNs), have been proposed as robust alternatives to NLLS. Here we present a deep learning implementation of qMRI model fitting, which uses DNNs to perform the inversion of the forward signal model. We compare two DNN training strategies, based on two alternative definitions of the loss function, since at present it is not known which definition leads to the most accurate, precise and robust parameter estimation. In strategy 1 we define the loss as the l2-norm of tissue parameter prediction errors, while in strategy 2 as the l2-norm of MRI signal prediction errors. We compare the two approaches on synthetic and 3T in vivo saturation inversion recovery (SIR) diffusion-weighted (DW) MRI data, using a model for joint diffusion-T1 mapping. Strategy 1 leads to lower tissue parameter root mean squared errors (RMSEs) when realistic noise distributions are considered (e.g. Rician vs Gaussian). However, strategy 2 offers lower signal reconstruction RMSE, and allows training to be performed on both synthetic and actual in vivo MRI measurements. In conclusion, both strategies are valid choices for DNN-based fitting. Strategy 2 is more practical, as it does not require pre-computation of reference tissue parameters, but may lead to worse parameter estimation.

neuroscience↗

bioRxiv2020

"Select and retrieve via direct upsampling" network (SARDU-Net): a data-driven, model-free, deep learning approach for quantitative MRI protocol design

PurposeWe introduce "Select and retrieve via direct upsampling" network (SARDU-Net), a data-driven framework for model-free quantitative MRI (qMRI) protocol design, and demonstrate it on in vivo brain and prostate diffusion-relaxation imaging (DRI). MethodsSARDU-Net selects subsets of informative measurements within lengthy pilot scans, without the requirement to identify tissue parameters for which to optimise for. The algorithm consists of a selector, identifying measurement subsets, and a predictor, estimating fully-sampled signals from the subsets. We implement both using deep neural networks, which are trained jointly end-to-end. We demonstrate the algorithm on brain (32 diffusion-/T1-weightings) and prostate (16 diffusion-/T2-weightings) DRI scans acquired on 3 healthy volunteers on two separate 3T Philips systems each. We used SARDU-Net to identify sub-protocols of fixed size, assessing the reproducibility of the procedure and testing sub-protocols for their potential to inform multi-contrast analyses via T1-weighted spherical mean diffusion tensor (T1-SMDT, brain) and hybrid multi-dimensional MRI (HM-MRI, prostate) modelling. ResultsIn both brain and prostate, SARDU-Net identifies sub-protocols that maximise information content in a reproducible manner across training instantiations. The sub-protocols enable multi-contrast modelling for which they were not optimised explicitly, providing robust T1-SMDT and HM-MRI maps and goodness-of-fit in the top 5% against extensive sub-protocol comparisons. ConclusionsSARDU-Net gives new opportunities to identify economical but informative qMRI protocols from a subset of the pilot scans that can be used for acquisition-time-sensitive applications. The simple architecture makes the algorithm easy to train when exhaustive searches are intractable, and applicable to a variety of anatomical contexts.

bioengineering↗

bioRxiv2019

Multi-parametric quantitative spinal cord MRI with unified signal readout and image denoising

Multi-parametric quantitative MRI (qMRI) of the spinal cord is a promising non-invasive tool to probe early microstructural damage in neurological disorders. It is usually performed by combining acquisitions with multiple signal readouts, which exhibit different thermal noise levels, geometrical distortions and susceptibility to physiological noise. This ultimately hinders joint multi-contrast modelling and makes the geometric correspondence of parametric maps challenging. We propose an approach to overcome these limitations, by implementing state-of-the-art microstructural MRI of the spinal cord with a unified signal readout. We base our acquisition on single-shot echo planar imaging with reduced field-of-view, and obtain data from two different vendors (vendor 1: Philips Achieva; vendor 2: Siemens Prisma). Importantly, the unified acquisition allows us to compare signal and noise across contrasts, thus enabling overall quality enhancement via Marchenko-Pastur (MP) Principal Component Analysis (PCA) denoising. MP-PCA is a recent method relying on redundant acquisitions, i.e. such that the number of measurements is much larger than the number of informative principal components. Here we used in vivo and synthetic data to test whether a unified readout enables more efficient denoising of less redundant acquisitions, since these can be denoised jointly with more redundant ones. We demonstrate that a unified readout provides robust multi-parametric maps, including diffusion and kurtosis tensors from diffusion MRI, myelin metrics from two-pool magnetisation transfer, and T1 and T2 from relaxometry. Moreover, we show that MP-PCA improves the quality of our multi-contrast acquisitions, since it reduces the coefficient of variation (i.e. variability) by up to 15% for mean kurtosis, 8% for bound pool fraction (BPF, myelin-sensitive), and 13% for T1, while enabling more efficient denoising of modalities limited in redundancy (e.g. relaxometry). In conclusion, multi-parametric spinal cord qMRI with unified readout is feasible and provides robust microstructural metrics with matched resolution and distortions, whose quality benefits from MP-PCA denoising, a useful pre-processing tool for spinal cord MRI.

neuroscience↗

bioRxiv2019

Predicting Alzheimer’s disease progression using deep recurrent neural networks

Early identification of individuals at risk of developing Alzheimers disease (AD) dementia is important for developing disease-modifying therapies. In this study, given multimodal AD markers and clinical diagnosis of an individual from one or more timepoints, we seek to predict the clinical diagnosis, cognition and ventricular volume of the individual for every month (indefinitely) into the future. We proposed and applied a minimal recurrent neural network (minimalRNN) model to data from The Alzheimers Disease Prediction Of Longitudinal Evolution (TADPOLE) challenge, comprising longitudinal data of 1677 participants (Marinescu et al. 2018) from the Alzheimers Disease Neuroimaging Initiative (ADNI). We compared the performance of the minimalRNN model and four baseline algorithms up to 6 years into the future. Most previous work on predicting AD progression ignore the issue of missing data, which is a prevalent issue in longitudinal data. Here, we explored three different strategies to handle missing data. Two of the strategies treated the missing data as a "preprocessing" issue, by imputing the missing data using the previous timepoint ("forward filling") or linear interpolation ("linear filling). The third strategy utilized the minimalRNN model itself to fill in the missing data both during training and testing ("model filling"). Our analyses suggest that the minimalRNN with "model filling" compared favorably with baseline algorithms, including support vector machine/regression, linear state space (LSS) model, and long short-term memory (LSTM) model. Importantly, although the training procedure utilized longitudinal data, we found that the trained minimalRNN model exhibited similar performance, when using only 1 input timepoint or 4 input timepoints, suggesting that our approach might work well with just cross-sectional data. An earlier version of our approach was ranked 5th (out of 53 entries) in the TADPOLE challenge in 2019. The current approach is ranked 2nd out of 63 entries as of June 3rd, 2020.

neuroscience↗

bioRxiv2019

Modeling longitudinal imaging biomarkers with parametric Bayesian multi-task learning

Longitudinal imaging biomarkers are invaluable for understanding the course of neurodegeneration, promising the ability to track disease progression and to detect disease earlier than cross-sectional biomarkers. To properly realize their potential, biomarker trajectory models must be robust to both under-sampling and measurement errors and should be able to integrate multi-modal information to improve trajectory inference and prediction. Here we present a parametric Bayesian multi-task learning based approach to modeling univariate trajectories across subjects that addresses these criteria.\n\nOur approach learns multiple subjects trajectories within a single model that allows for different types of information sharing, i.e. coupling, across subjects. It optimizes a combination of uncoupled, fully coupled and kernel coupled models. Kernel-based coupling allows linking subjects trajectories based on one or more biomarker measures. We demonstrate this using Alzheimers Disease Neuroimaging Initiative (ADNI) data, where we model longitudinal trajectories of MRI-derived cortical volumes in neurodegeneration, with coupling based on APOE genotype, cerebrospinal fluid (CSF) and amyloid PET-based biomarkers. In addition to detecting established disease effects, we detect disease related changes within the insula that have not received much attention within the literature.\n\nDue to its sensitivity in detecting disease effects, its competitive predictive performance and its ability to learn the optimal parameter covariance from data rather than choosing a specific set of random and fixed effects a priori, we propose that our model can be used in place of or in addition to linear mixed effects models when modeling biomarker trajectories. A software implementation of the method is publicly available.

neuroscience↗

bioRxiv2018

Application of mechanistic methods to clinical trials in multiple sclerosis: the simvastatin case

The analysis of clinical trials is limited to pre-specified outcomes, thereby precluding a mechanistic understanding of the treatment response. Multivariate mechanistic models can elucidate the causal chain of events by simultaneous analysis of multimodal data that link intermediate variables to outcomes of interest. A double-blind, randomised, controlled, phase 2 clinical trial in secondary progressive multiple sclerosis (MS-STAT, NCT00647348) demonstrated that simvastatin (80mg/day) over two years reduced the brain atrophy rate and was associated with beneficial effects on cognitive and disability outcomes. Therefore, this trial offers an opportunity to apply mechanistic models to investigate the hypothesised pathways that link simvastatin to clinical outcome measures, either directly or indirectly via changes in serum total cholesterol levels and to determine which is the more likely.

neuroscience↗