Big Data, Big Opportunity: Using Normative Models to Bridge Cross‐Sectional and Longitudinal Insights

Normative modelling is becoming more popular in neuroimaging due to its ability to make predictions of deviation from a normal trajectory at the level of individual participants. It allows the user to model the distribution of several neuroimaging modalities, giving an estimation for the mean and centiles of variation. With the increase in the availability of big data in neuroimaging, there is a need to scale normative modelling to big data sets. However, the scaling of normative models has come with several challenges.So far, most normative modelling approaches used Gaussian process regression, and although suitable for smaller datasets (up to a few thousand participants) it does not scale well to the large cohorts currently available and being acquired. Furthermore, most neuroimaging modelling methods that are available assume the predictive distribution to be Gaussian in shape. However, deviations from Gaussianity can be frequently found, which may lead to incorrect inferences, particularly in the outer centiles of the distribution. In normative modelling, we use the centiles to give an estimation of the deviation of a particular participant from the ‘normal’ trend. Therefore, especially in normative modelling, the correct estimation of the outer centiles is of utmost importance, which is also where data are sparsest.Here, we present a novel framework based on Bayesian linear regression with likelihood warping that allows us to address these problems, that is, to correctly model non-Gaussian predictive distributions and scale normative modelling elegantly to big data cohorts. In addition, this method provides likelihood-based statistics, which are useful for model selection.To evaluate this framework, we use a range of neuroimaging-derived measures from the UK Biobank study, including image-derived phenotypes (IDPs) and whole-brain voxel-wise measures derived from diffusion tensor imaging. We show good computational scaling and improved accuracy of the warped BLR for certain IDPs and voxels if there was a deviation from normality of these parameters in their residuals.The present results indicate the advantage of a warped BLR in terms of; computational scalability and the flexibility to incorporate non-linearity and non-Gaussianity of the data, giving a wider range of neuroimaging datasets that can be correctly modelled.

Accommodating site variation in neuroimaging data using normative and hierarchical Bayesian models

Longitudinal neuroimaging studies offer valuable insight into intricate dynamics of brain development, ageing, and disease progression over time. However, prevailing analytical approaches rooted in our understanding of population variation are primarily tailored for cross-sectional studies. To fully harness the potential of longitudinal neuroimaging data, we have to develop and refine methodologies that are adapted to longitudinal designs, considering the complex interplay between population variation and individual dynamics.We build on normative modelling framework, which enables the evaluation of an individual’s position compared to a population standard. We extend this framework to evaluate an individual’s longitudinal change compared to the longitudinal change reflected by the (population) standard dynamics. Thus, we exploit the existing normative models pre-trained on over 58,000 individuals and adapt the framework so that they can also be used in the evaluation of longitudinal studies. Specifically, we introduce a quantitative metric termed “z-diff” score, which serves as an indicator of a temporal change of an individual compared to a population standard. Notably, our framework offers advantages such as flexibility in dataset size and ease of implementation.To illustrate our approach, we applied it to a longitudinal dataset of 98 patients diagnosed with early-stage schizophrenia who underwent MRI examinations shortly after diagnosis and one year later.Compared to cross-sectional analyses, which showed global thinning of grey matter at the first visit, our method revealed a significant normalisation of grey matter thickness in the frontal lobe over time. Furthermore, this result was not observed when using more traditional methods of longitudinal analysis, making our approach more sensitive to temporal changes.Overall, our framework presents a flexible and effective methodology for analysing longitudinal neuroimaging data, providing insights into the progression of a disease that would otherwise be missed when using more traditional approaches.

Longitudinal neuroimaging studies offer valuable insight into intricate dynamics of brain development, ageing, and disease progression over time. However, prevailing analytical approaches rooted in our understanding of population variation are primarily tailored for cross-sectional studies. To fully harness the potential of longitudinal neuroimaging data, we have to develop and refine methodologies that are adapted to longitudinal designs, considering the complex interplay between population variation and individual dynamics.We build on normative modelling framework, which enables the evaluation of an individual’s position compared to a population standard. We extend this framework to evaluate an individual’s longitudinal change compared to the longitudinal change reflected by the (population) standard dynamics. Thus, we exploit the existing normative models pre-trained on over 58,000 individuals and adapt the framework so that they can also be used in the evaluation of longitudinal studies. Specifically, we introduce a quantitative metric termed “z-diff” score, which serves as an indicator of a temporal change of an individual compared to a population standard. Notably, our framework offers advantages such as flexibility in dataset size and ease of implementation.To illustrate our approach, we applied it to a longitudinal dataset of 98 patients diagnosed with early-stage schizophrenia who underwent MRI examinations shortly after diagnosis and one year later.Compared to cross-sectional analyses, which showed global thinning of grey matter at the first visit, our method revealed a significant normalisation of grey matter thickness in the frontal lobe over time. Furthermore, this result was not observed when using more traditional methods of longitudinal analysis, making our approach more sensitive to temporal changes.Overall, our framework presents a flexible and effective methodology for analysing longitudinal neuroimaging data, providing insights into the progression of a disease that would otherwise be missed when using more traditional approaches.

Coronavirus disease 19 (COVID‐19) has dramatically altered everyday life, including the field of Alzheimer's disease (AD) research. This perspective article explores some of the ways in which COVID‐19 has already impacted the field, anticipates some of the long‐lasting effects, and explores strategies for addressing current and future needs. Areas of impact include study integrity, regulatory and industry issues, and participant engagement. Proposed strategies for addressing these challenges include analytic methods to deal with large degrees of missing data and development of patient‐centered, user‐friendly, remote data collection tools and assessments. We also highlight the importance of maintaining participant well‐being as a first and constant priority.

The paper, "Ethnicity Moderates Dementia's Biomarkers", by Royall and Palmer in this issue of Journal of Alzheimer's Disease represents the cutting edge of Alzheimer's disease (AD) research. The authors capitalize on several powerful and emerging trends in AD research that will surely reap benefits for our discipline during the next decade: latent variable models, biomarkers, and ethnicity. In this study, the authors specifically find that self-reported ethnicity moderates the dementing process and hypothesize that this is more likely due to distinct biological mechanisms than environmental influences.

Longitudinal neuroimaging studies offer valuable insight into brain development, ageing, and disease progression over time. However, prevailing analytical approaches rooted in our understanding of population variation are primarily tailored for cross-sectional studies. To fully leverage the potential of longitudinal neuroimaging, we need methodologies that account for the complex interplay between population variation and individual dynamics. We extend the normative modelling framework, which evaluates an individual’s position relative to population standards, to assess an individual’s longitudinal change compared to the population’s standard dynamics. Using normative models pre-trained on over 58,000 individuals, we introduce a quantitative metric termed ‘ z-diff ’ score, which quantifies a temporal change in individuals compared to a population standard. This approach offers advantages in flexibility in dataset size and ease of implementation. We applied this framework to a longitudinal dataset of 98 patients with early-stage schizophrenia who underwent MRI examinations shortly after diagnosis and 1 year later. Compared to cross-sectional analyses, showing global thinning of grey matter at the first visit, our method revealed a significant normalisation of grey matter thickness in the frontal lobe over time—an effect undetected by traditional longitudinal methods. Overall, our framework presents a flexible and effective methodology for analysing longitudinal neuroimaging data, providing insights into the progression of a disease that would otherwise be missed when using more traditional approaches.

Background: Alzheimer's disease, the most common form of dementia, has tremendous social and economic impact worldwide. This study aimed to analyze global trends in Alzheimer's disease research and to investigate China's contribution to this research.Methods: The quantity and influence of publications related to Alzheimer's disease in China and elsewhere were compared. The Web of Science (WOS) and PubMed databases were searched from 1988 to 2017 using the terms “Alzheimer's disease” or “Alzheimers disease.” Global Alzheimer's disease publications were classified and analyzed. Keywords, countries, and institutions publishing articles on Alzheimer's disease were analyzed, and citations of these articles were examined.Results: A total of 181,116 articles regarding Alzheimer's disease research were identified and analyzed. Neuroscience and neurology were the main research categories both globally and in China. Basic research dominated Alzheimer's publications, accounting for 30.93% of global publications and 95.31% of publications in China. A total of 8,935 journals published articles related to Alzheimer's disease. The journal Neurobiology of Aging published the most Alzheimer's disease-related articles, numbering 5,206 over the time period examined. The National Institutes of Health, the National Institute on Aging, and the Department of Health and Human Services jointly sponsored 11,809 articles, ranking first in the world. The National Natural Science Foundation of China funded the largest number of studies on Alzheimer's disease in China and recognized the importance of traditional Chinese medicine in Alzheimer's disease research.Conclusions: The present study provides data for global researchers to understand research perspectives and develop future research directions. In recent years, Chinese researchers have contributed significantly to global Alzheimer's research. Still, strengthening international cooperation could improve the quality and number of publications regarding Alzheimer's disease.

BackgroundFamilial Alzheimer's disease research necessitates innovative methodologies to disentangle the intricate relationships between genetic factors and neuroimaging measures. Traditional frequentist approaches, often hampered by small sample sizes in this population and challenges in incorporating prior knowledge transparently, may limit the robustness of findings.MethodsWe analyzed neuroimaging data of preclinical PSNE1 single mutation carriers, utilizing the software JASP to test effects of carrier status on measures of basal forebrain functional connectivity using both frequentist and Bayesian approach. Bayesian Analysis of Covariance was first implemented to test the hypothesis of a basal forebrain functional connectivity reduction in mutation carriers compared to non‐mutation carriers. To get more precise information, Informative priors, derived from the existing literature on preclinical sporadic Alzheimer's disease, were integrated into post‐hoc Independent samples t‐tests. Student t‐tests were utilized for parameter estimation, using an Informed Cauchy specification of prior's location (‐0.7) and scale (0.707) allowing the derivation of posterior distributions.ResultsOur findings demonstrated that preclinical mutation carriers exhibited no significant alterations in basal forebrain functional connectivity compared to non‐carriers. However, the Bayesian analysis revealed distinct advantages over the traditional frequentist approach (anterior basal forebrain: t=‐1.126 (213), p= 0.131, Cohen's d = ‐0.165, 95 % CI for Cohen's d: [‐Inf., 0.077]; posterior basal forebrain: t=‐1.337 (213), p= 0.091, Cohen's d = ‐0.196, 95 % CI for Cohen's d: [‐Inf., 0.046]. Incorporating informative priors significantly improved the precision of parameter estimates, allowing us to reject our hypothesis with a relatively high degree of confidence, from anecdotal (anterior basal forebrain: BF10 = 1.167, δ = ‐ 0.319, 95 % CI: [‐0.599, ‐0.061]) to strong evidence (posterior basal forebrain: BF10 = 0.033, δ = ‐0.067, 95 % CI: [‐0.260, ‐0.003]).DiscussionOur results underscore the importance of adopting Bayesian frameworks in familial Alzheimer's disease research. The explicit integration of prior knowledge, as facilitated by Bayesian analysis, enhances the reliability of conclusions drawn from neuroimaging data. These methodological advantages are crucial in advancing the field and encouraging replication efforts.

Normative Modelling (‘brain growth charting’) is now a well-established method for computational psychiatry and involves charting centiles of variation across a population in terms of mappings between biology and behavior, providing statistical inferences at the level of the individual. These models have helped the field to move away from case-control analysis toward individual-level analysis. Correspondingly, normative modelling has now been applied to chart brain development and ageing in many populations and has been used to quantify individual deviations across various neurological and psychiatric conditions. This has been supported by large-scale models that are openly accessible for diverse brain imaging modalities. As normative modelling continues to grow, several recent methodological developments, such as non-Gaussian models, longitudinal models, and federated learning, have been implemented in different software tools, including the Predictive Clinical Neuroscience toolkit (PCNtoolkit). In this protocol update, we provide: (i) a revised overview of this methodological landscape; (ii) an update to our 2022 standardised analytical protocol for normative modelling of neuroimaging data, including options for federated and longitudinal normative models; (iii) practical guidance suited to both novice and experienced practitioners supported by open-source code examples implemented in the refactored version of PCNtoolkit; and (iv) updated models for cortical thickness, volumetric data, diffusion-weighted imaging and longitudinal data for use by the community.

BackgroundThe health-related quality of life (HRQoL) of patients with non–muscle-invasive bladder cancer (NMIBC) may be impaired due to the chronic and burdensome disease course, but longitudinal data are limited. ObjectiveTo evaluate HRQoL outcomes during the first 4 yr after NMIBC diagnosis, and to compare HRQoL across patient characteristics and with a normative population. Design, setting, and participantsPatients with NMIBC (n = 1019) were included from the multicentre prospective cohort UroLife. Data were collected at 6 wk (baseline), and 3, 15, and 51 mo after diagnosis. Longitudinal reference data were obtained from an age- and sex-matched normative population (n = 490). Outcome measurements and statistical analysisCancer- and NMIBC-specific HRQoL outcomes (range 0–100) were evaluated by the European Organisation for Research and Treatment of Cancer (EORTC) QLQ-C30 and EORTC QLQ-NMIBC24 questionnaires, respectively. Linear mixed modelling was used to analyse within-group changes and between-group differences. Results and limitationsThe majority of HRQoL outcomes remained stable over time. Observed changes were only of small clinical relevance. Improvements were noted in insomnia, social functioning, and three NMIBC-specific symptoms, while minor deteriorations in appetite and diarrhoea lasted until 51 mo. HRQoL in some domains was worse for high-grade NMIBC, high European Association of Urology (EAU) risk group, initial Bacillus Calmette-Guérin (BCG) treatment, being female, and being younger (<65 yr); yet differences were few, small, and temporary. No differences were observed across recurrence status. Compared with a normative population, clinically relevant worse scores were observed for six of 15 outcomes, which mostly recovered at 51 mo, except for minor symptoms of appetite loss and diarrhoea. ConclusionsNo remarkable changes in HRQoL were observed during the first 4 yr after NMIBC diagnosis. Grade, EAU risk group, initial treatment, recurrence, sex, and age did not importantly affect HRQoL. HRQoL was largely comparable with that of the general population, especially after 4 yr. Patient summaryQuality of life is not largely affected during the first 4 yr after non–muscle-invasive bladder cancer diagnosis.

Normative modelling is a method for understanding the underlying heterogeneity within brain disorders like Alzheimer Disease (AD), by quantifying how each patient deviates from the expected normative pattern that has been learned from a healthy control distribution. Existing deep learning based normative models have been applied on only single modality Magnetic Resonance Imaging (MRI) neuroimaging data. However, these do not take into account the complementary information offered by multimodal M RI, which is essential for understanding a multifactorial disease like AD. To address this limitation, we propose a multi-modal variational autoencoder (mmVAE) based normative modelling framework that can capture the joint distribution between different modalities to identify abnormal brain volume deviations due to AD. Our multi-modal framework takes as input Freesurfer processed brain region volumes from T1-weighted (cortical and subcortical) and T2-weighed (hippocampal) scans of cognitively normal participants to learn the morphological characteristics of the healthy brain. The estimated normative model is then applied on AD patients to quantify the deviation in brain volumes and identify abnormal brain pattern deviations due to the progressive stages of AD. We compared our proposed mmVAE with a baseline unimodal VAE having a single encoder and decoder and the two modalities concatenated as unimodal input. Our experimental results show that deviation maps generated by mmVAE are more sensitive to disease staging within AD, have a better correlation with patient cognition and result in higher number of brain regions with statistically significant deviations compared to the unimodal baseline model.

Normative modelling is an emerging method for quantifying how individuals deviate from the healthy populational pattern. Several machine learning models have been implemented to develop normative models to investigate brain disorders, including regression, support vector machines and Gaussian process models. With the advance of deep learning technology, the use of deep neural networks has also been proposed. In this study, we assessed normative models based on deep autoencoders using structural neuroimaging data from patients with Alzheimer’s disease (n = 206) and mild cognitive impairment (n = 354). We first trained the autoencoder on an independent dataset (UK Biobank dataset) with 11,034 healthy controls. Then, we estimated how each patient deviated from this norm and established which brain regions were associated to this deviation. Finally, we compared the performance of our normative model against traditional classifiers. As expected, we found that patients exhibited deviations according to the severity of their clinical condition. The model identified medial temporal regions, including the hippocampus, and the ventricular system as critical regions for the calculation of the deviation score. Overall, the normative model had comparable cross-cohort generalizability to traditional classifiers. To promote open science, we are making all scripts and the trained models available to the wider research community.

In the field of computer vision, in particular, deep learning, a cutting-edge method for machine learning, has surpassed conventional machine learning in its ability to detect intricate structures in complicated, high-dimensional data. With the advent of large-scale multi-modal neuroimaging data made possible by the fast development of neuroimaging techniques, there has been a surge of interest in using deep learning for the automated classification and early detection of Alzheimer's disease (AD). This chapter presents the results of a literature review on the topic of Alzheimer's disease diagnostic classification using deep learning methods and neuroimaging data. The authors also summarize the data sources, processing steps, training protocols, and evaluation methods to help future deep-learning studies on Alzheimer's disease researchers. There are limitations to deep learning, despite its promising performance across studies and tasks.

Participant retention is important to maintaining statistical power, minimizing bias, and preventing scientific error in Alzheimer disease and related dementias research. We surveyed representative investigators from NIH-funded Alzheimer's Disease Research Centers (ADRC), querying their use of retention tactics across 12 strategies. We compared survey results to data from the National Alzheimer's Coordinating Center for each center. We used a generalized estimating equation with independent working covariance model and empirical standard errors to assess relationships between survey results and rates of retention, controlling for participant characteristics. Twenty-five (83%) responding ADRCs employed an average 42 (SD=7) retention tactics. In a multivariable model that accounted for participant characteristics, the number of retention tactics used by a center was associated with participant retention (odds ratio=1.68, 95% confidence interval: 1.42, 1.98; P<0.001 for the middle compared with the lowest tertile survey scores; odds ratio=1.59, 95% confidence interval: 1.30, 1.94; P<0.001 for the highest compared with the lowest tertile survey scores) at the first follow-up visit. Participant characteristics such as normal cognition diagnosis, older age, higher education, and Caucasian race were also associated with higher retention. Retention in clinical research is more likely to be achieved by employing a variety of tactics.