Cross-cohort replicable resting-state functional connectivity in predicting symptoms and cognition of schizophrenia.

Schizophrenia (SZ) is a debilitating mental illness characterized by adolescence or early adulthood onset of psychosis, positive and negative symptoms, as well as cognitive impairments. Despite a plethora of studies leveraging functional connectivity (FC) from functional magnetic resonance imaging (fMRI) to predict symptoms and cognitive impairments of SZ, the findings have exhibited great heterogeneity. We aimed to identify congruous and replicable connectivity patterns capable of predicting positive and negative symptoms as well as cognitive impairments in SZ. Predictable functional connections (FCs) were identified by employing an individualized prediction model, whose replicability was further evaluated across three independent cohorts (BSNIP, SZ = 174; COBRE, SZ = 100; FBIRN, SZ = 161). Across cohorts, we observed that altered FCs in frontal-temporal-cingulate-thalamic network were replicable in prediction of positive symptoms, while sensorimotor network was predictive of negative symptoms. Temporal-parahippocampal network was consistently identified to be associated with reduced cognitive function. These replicable 23 FCs effectively distinguished SZ from healthy controls (HC) across three cohorts (82.7%, 90.2%, and 86.1%). Furthermore, models built using these replicable FCs showed comparable accuracies to those built using the whole-brain features in predicting symptoms/cognition of SZ across the three cohorts (r = .17-.33, p < .05). Overall, our findings provide new insights into the neural underpinnings of SZ symptoms/cognition and offer potential targets for further research and possible clinical interventions.

Determining four confounding factors in individual cognitive traits prediction with functional connectivity: an exploratory study.

Resting-state functional connectivity (RSFC) has been widely adopted for individualized trait prediction. However, multiple confounding factors may impact the predicted brain-behavior relationships. In this study, we investigated the impact of 4 confounding factors including time series length, functional connectivity (FC) type, brain parcellation choice, and variance of the predicted target. The data from Human Connectome Project including 1,206 healthy subjects were employed, with 3 cognitive traits including fluid intelligence, working memory, and picture vocabulary ability as the prediction targets. We compared the prediction performance under different settings of these 4 factors using partial least square regression. Results demonstrated appropriate time series length (300 time points) and brain parcellation (independent component analysis, ICA100/200) can achieve better prediction performance without too much time consumption. FC calculated by Pearson, Spearman, and Partial correlation achieves higher accuracy and lower time cost than mutual information and coherence. Cognitive traits with larger variance among subjects can be better predicted due to the well elaboration of individual variability. In addition, the beneficial effects of increasing scan duration to prediction partially arise from the improved test-retest reliability of RSFC. Taken together, the study highlights the importance of determining these factors in RSFC-based prediction, which can facilitate standardization of RSFC-based prediction pipelines going forward.

Aberrant brain functional and structural developments in MECP2 duplication rats.

Transgenic animal models with homologous etiology provide a promising way to pursue the neurobiological substrates of the behavioral deficits in autism spectrum disorder (ASD). Gain-of-function mutations of MECP2 cause MECP2 duplication syndrome, a severe neurological disorder with core symptoms of ASD. However, abnormal brain developments underlying the autistic-like behavioral deficits of MECP2 duplication syndrome are rarely investigated. To this end, a human MECP2 duplication (MECP2-DP) rat model was created by the bacterial artificial chromosome transgenic method. Functional and structural magnetic resonance imaging (MRI) with high-field were performed on 16 male MECP2-DP rats and 15 male wildtype rats at postnatal 28 days, 42 days, and 56 days old. Multimodal fusion analyses guided by locomotor-relevant metrics and social novelty time separately were applied to identify abnormal brain networks associated with diverse behavioral deficits induced by MECP2 duplication. Aberrant functional developments of a core network primarily composed of the dorsal medial prefrontal cortex (dmPFC) and retrosplenial cortex (RSP) were detected to associate with diverse behavioral phenotypes in MECP2-DP rats. Altered developments of gray matter volume were detected in the hippocampus and thalamus. We conclude that gain-of-function mutations of MECP2 induce aberrant functional activities in the default-mode-like network and aberrant volumetric changes in the brain, resulting in autistic-like behavioral deficits. Our results gain critical insights into the biomarker of MECP2 duplication syndrome and the neurobiological underpinnings of the behavioral deficits in ASD.

Multimodal brain deficits shared in early-onset and adult-onset schizophrenia predict positive symptoms regardless of illness stage.

Incidence of schizophrenia (SZ) has two predominant peaks, in adolescent and young adult. Early-onset schizophrenia provides an opportunity to explore the neuropathology of SZ early in the disorder and without the confound of antipsychotic mediation. However, it remains unexplored what deficits are shared or differ between adolescent early-onset (EOS) and adult-onset schizophrenia (AOS) patients. Here, based on 529 participants recruited from three independent cohorts, we explored AOS and EOS common and unique co-varying patterns by jointly analyzing three MRI features: fractional amplitude of low-frequency fluctuations (fALFF), gray matter (GM), and functional network connectivity (FNC). Furthermore, a prediction model was built to evaluate whether the common deficits in drug-naive SZ could be replicated in chronic patients. Results demonstrated that (1) both EOS and AOS patients showed decreased fALFF and GM in default mode network, increased fALFF and GM in the sub-cortical network, and aberrant FNC primarily related to middle temporal gyrus; (2) the commonly identified regions in drug-naive SZ correlate with PANSS positive significantly, which can also predict PANSS positive in chronic SZ with longer duration of illness. Collectively, results suggest that multimodal imaging signatures shared by two types of drug-naive SZ are also associated with positive symptom severity in chronic SZ and may be vital for understanding the progressive schizophrenic brain structural and functional deficits.

Three-way parallel group independent component analysis: Fusion of spatial and spatiotemporal magnetic resonance imaging data.

Advances in imaging acquisition techniques allow multiple imaging modalities to be collected from the same subject. Each individual modality offers limited yet unique views of the functional, structural, or dynamic temporal features of the brain. Multimodal fusion provides effective ways to leverage these complementary perspectives from multiple modalities. However, the majority of current multimodal fusion approaches involving functional magnetic resonance imaging (fMRI) are limited to 3D feature summaries that do not incorporate its rich temporal information. Thus, we propose a novel three-way parallel group independent component analysis (pGICA) fusion method that incorporates the first-level 4D fMRI data (temporal information included) by parallelizing group ICA into parallel ICA via a unified optimization framework. A new variability matrix was defined to capture subject-wise functional variability and then link it to the mixing matrices of the other two modalities. Simulation results show that the three-way pGICA provides highly accurate cross-modality linkage estimation under both weakly and strongly correlated conditions, as well as comparable source estimation under different noise levels. Results using real brain imaging data identified one linked functional-structural-diffusion component associated to differences between schizophrenia and controls. This was replicated in an independent cohort, and the identified components were also correlated with major cognitive domains. Functional network connectivity revealed visual-subcortical and default mode-cerebellum pairs that discriminate between schizophrenia and controls. Overall, both simulation and real data results support the use of three-way pGICA to identify multimodal spatiotemporal links and to pursue the study of brain disorders under a single unifying multimodal framework.

Associations between grip strength, brain structure, and mental health in > 40,000 participants from the UK Biobank.

BACKGROUND: Grip strength is a widely used and well-validated measure of overall health that is increasingly understood to index risk for psychiatric illness and neurodegeneration in older adults. However, existing work has not examined how grip strength relates to a comprehensive set of mental health outcomes, which can detect early signs of cognitive decline. Furthermore, whether brain structure mediates associations between grip strength and cognition remains unknown. METHODS: Based on cross-sectional and longitudinal data from over 40,000 participants in the UK Biobank, this study investigated the behavioral and neural correlates of handgrip strength using a linear mixed effect model and mediation analysis. RESULTS: In cross-sectional analysis, we found that greater grip strength was associated with better cognitive functioning, higher life satisfaction, greater subjective well-being, and reduced depression and anxiety symptoms while controlling for numerous demographic, anthropometric, and socioeconomic confounders. Further, grip strength of females showed stronger associations with most behavioral outcomes than males. In longitudinal analysis, baseline grip strength was related to cognitive performance at ~9 years follow-up, while the reverse effect was much weaker. Further, baseline neuroticism, health, and financial satisfaction were longitudinally associated with subsequent grip strength. The results revealed widespread associations between stronger grip strength and increased grey matter volume, especially in subcortical regions and temporal cortices. Moreover, grey matter volume of these regions also correlated with better mental health and considerably mediated their relationship with grip strength. CONCLUSIONS: Overall, using the largest population-scale neuroimaging dataset currently available, our findings provide the most well-powered characterization of interplay between grip strength, mental health, and brain structure, which may facilitate the discovery of possible interventions to mitigate cognitive decline during aging.

Links between electroconvulsive therapy responsive and cognitive impairment multimodal brain networks in late-life major depressive disorder.

BACKGROUND: Although electroconvulsive therapy (ECT) is an effective treatment for depression, ECT cognitive impairment remains a major concern. The neurobiological underpinnings and mechanisms underlying ECT antidepressant and cognitive impairment effects remain unknown. This investigation aims to identify ECT antidepressant-response and cognitive-impairment multimodal brain networks and assesses whether they are associated with the ECT-induced electric field (E-field) with an optimal pulse amplitude estimation. METHODS: A single site clinical trial focused on amplitude (600, 700, and 800 mA) included longitudinal multimodal imaging and clinical and cognitive assessments completed before and immediately after the ECT series (n = 54) for late-life depression. Another two independent validation cohorts (n = 84, n = 260) were included. Symptom and cognition were used as references to supervise fMRI and sMRI fusion to identify ECT antidepressant-response and cognitive-impairment multimodal brain networks. Correlations between ECT-induced E-field within these two networks and clinical and cognitive outcomes were calculated. An optimal pulse amplitude was estimated based on E-field within antidepressant-response and cognitive-impairment networks. RESULTS: Decreased function in the superior orbitofrontal cortex and caudate accompanied with increased volume in medial temporal cortex showed covarying functional and structural alterations in both antidepressant-response and cognitive-impairment networks. Volume increases in the hippocampal complex and thalamus were antidepressant-response specific, and functional decreases in the amygdala and hippocampal complex were cognitive-impairment specific, which were validated in two independent datasets. The E-field within these two networks showed an inverse relationship with HDRS reduction and cognitive impairment. The optimal E-filed range as [92.7-113.9] V/m was estimated to maximize antidepressant outcomes without compromising cognitive safety. CONCLUSIONS: The large degree of overlap between antidepressant-response and cognitive-impairment networks challenges parameter development focused on precise E-field dosing with new electrode placements. The determination of the optimal individualized ECT amplitude within the antidepressant and cognitive networks may improve the treatment benefit-risk ratio. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT02999269.

Functional Connectivity Predicts Individual Development of Inhibitory Control during Adolescence.

Derailment of inhibitory control (IC) underlies numerous psychiatric and behavioral disorders, many of which emerge during adolescence. Identifying reliable predictive biomarkers that place the adolescents at elevated risk for future IC deficits can help guide early interventions, yet the scarcity of longitudinal research has hindered the progress. Here, using a large-scale longitudinal dataset in which the same subjects performed a stop signal task during functional magnetic resonance imaging at ages 14 and 19, we tracked their IC development individually and tried to find the brain features predicting their development by constructing prediction models using 14-year-olds' functional connections within a network or between a pair of networks. The participants had distinct between-subject trajectories in their IC development. Of the candidate connections used for prediction, ventral attention-subcortical network interconnections could predict the individual development of IC and formed a prediction model that generalized to previously unseen individuals. Furthermore, we found that connectivity between these two networks was related to substance abuse problems, an IC-deficit related problematic behavior, within 5 years. Our study reveals individual differences in IC development from mid- to late-adolescence and highlights the importance of ventral attention-subcortical network interconnections in predicting future IC development and substance abuse in adolescents.

Interpreting Brain Biomarkers: Challenges and solutions in interpreting machine learning-based predictive neuroimaging.

Predictive modeling of neuroimaging data (predictive neuroimaging) for evaluating individual differences in various behavioral phenotypes and clinical outcomes is of growing interest. However, the field is experiencing challenges regarding the interpretability of the results. Approaches to defining the specific contribution of functional connections, regions, or networks in prediction models are urgently needed, which may help explore the underlying mechanisms. In this article, we systematically review the methods and applications for interpreting brain signatures derived from predictive neuroimaging based on a survey of 326 research articles. Strengths, limitations, and the suitable conditions for major interpretation strategies are also deliberated. In-depth discussion of common issues in existing literature and the corresponding recommendations to address these pitfalls are provided. We highly recommend exhaustive validation on the reliability and interpretability of the biomarkers across multiple datasets and contexts, which thereby could translate technical advances in neuroimaging into concrete improvements in precision medicine.

Gender Differences in Connectome-based Predictions of Individualized Intelligence Quotient and Sub-domain Scores.

Scores on intelligence tests are strongly predictive of various important life outcomes. However, the gender discrepancy on intelligence quotient (IQ) prediction using brain imaging variables has not been studied. To this aim, we predicted individual IQ scores for males and females separately using whole-brain functional connectivity (FC). Robust predictions of intellectual capabilities were achieved across three independent data sets (680 subjects) and two intelligence measurements (IQ and fluid intelligence) using the same model within each gender. Interestingly, we found that intelligence of males and females were underpinned by different neurobiological correlates, which are consistent with their respective superiority in cognitive domains (visuospatial vs verbal ability). In addition, the identified FC patterns are uniquely predictive on IQ and its sub-domain scores only within the same gender but neither for the opposite gender nor on the IQ-irrelevant measures such as temperament traits. Moreover, females exhibit significantly higher IQ predictability than males in the discovery cohort. This findings facilitate our understanding of the biological basis of intelligence by demonstrating that intelligence is underpinned by a variety of complex neural mechanisms that engage an interacting network of regions-particularly prefrontal-parietal and basal ganglia-whereas the network pattern differs between genders.

Task-induced brain connectivity promotes the detection of individual differences in brain-behavior relationships.

Although both resting and task-induced functional connectivity (FC) have been used to characterize the human brain and cognitive abilities, the potential of task-induced FCs in individualized prediction for out-of-scanner cognitive traits remains largely unexplored. A recent study Greene et al. (2018) predicted the fluid intelligence scores using FCs derived from rest and multiple task conditions, suggesting that task-induced brain state manipulation improved prediction of individual traits. Here, using a large dataset incorporating fMRI data from rest and 7 distinct task conditions, we replicated the original study by employing a different machine learning approach, and applying the method to predict two reading comprehension-related cognitive measures. Consistent with their findings, we found that task-based machine learning models often outperformed rest-based models. We also observed that combining multi-task fMRI improved prediction performance, yet, integrating the more fMRI conditions can not necessarily ensure better predictions. Compared with rest, the predictive FCs derived from language and working memory tasks were highlighted with more predictive power in predominantly default mode and frontoparietal networks. Moreover, prediction models demonstrated high stability to be generalizable across distinct cognitive states. Together, this replication study highlights the benefit of using task-based FCs to reveal brain-behavior relationships, which may confer more predictive power and promote the detection of individual differences of connectivity patterns underlying relevant cognitive traits, providing strong evidence for the validity and robustness of the original findings.

Electroconvulsive therapy treatment responsive multimodal brain networks.

Electroconvulsive therapy is regarded as the most effective antidepressant treatment for severe and treatment-resistant depressive episodes. Despite the efficacy of electroconvulsive therapy, the neurobiological underpinnings and mechanisms underlying electroconvulsive therapy induced antidepressant effects remain unclear. The objective of this investigation was to identify electroconvulsive therapy treatment responsive multimodal biomarkers with the 17-item Hamilton Depression Rating Scale guided brain structure-function fusion in 118 patients with depressive episodes and 60 healthy controls. Results show that reduced fractional amplitude of low frequency fluctuations in the prefrontal cortex, insula and hippocampus, linked with increased gray matter volume in anterior cingulate, medial temporal cortex, insula, thalamus, caudate and hippocampus represent electroconvulsive therapy responsive covarying functional and structural brain networks. In addition, relative to nonresponders, responder-specific electroconvulsive therapy related brain networks occur in frontal-limbic network and are associated with successful therapeutic outcomes. Finally, electroconvulsive therapy responsive brain networks were unrelated to verbal declarative memory. Using a data-driven, supervised-learning method, we demonstrated that electroconvulsive therapy produces a remodeling of brain functional and structural covariance that was unique to antidepressant symptom response, but not linked to memory impairment.

Parallel group ICA+ICA: Joint estimation of linked functional network variability and structural covariation with application to schizophrenia.

There is growing evidence that rather than using a single brain imaging modality to study its association with physiological or symptomatic features, the field is paying more attention to fusion of multimodal information. However, most current multimodal fusion approaches that incorporate functional magnetic resonance imaging (fMRI) are restricted to second-level 3D features, rather than the original 4D fMRI data. This trade-off is that the valuable temporal information is not utilized during the fusion step. Here we are motivated to propose a novel approach called "parallel group ICA+ICA" that incorporates temporal fMRI information from group independent component analysis (GICA) into a parallel independent component analysis (ICA) framework, aiming to enable direct fusion of first-level fMRI features with other modalities (e.g., structural MRI), which thus can detect linked functional network variability and structural covariations. Simulation results show that the proposed method yields accurate intermodality linkage detection regardless of whether it is strong or weak. When applied to real data, we identified one pair of significantly associated fMRI-sMRI components that show group difference between schizophrenia and controls in both modalities, and this linkage can be replicated in an independent cohort. Finally, multiple cognitive domain scores can be predicted by the features identified in the linked component pair by our proposed method. We also show these multimodal brain features can predict multiple cognitive scores in an independent cohort. Overall, results demonstrate the ability of parallel GICA+ICA to estimate joint information from 4D and 3D data without discarding much of the available information up front, and the potential for using this approach to identify imaging biomarkers to study brain disorders.

MicroRNA132 associated multimodal neuroimaging patterns in unmedicated major depressive disorder.

There is compelling evidence that epigenetic factors contribute to the manifestation of depression, in which microRNA132 (miR-132) is suggested to play a pivotal role in the pathogenesis and neuronal mechanisms underlying the symptoms of depression. Additionally, several depression-associated genes [MECP2, ARHGAP32 (p250GAP), CREB, and period genes] were experimentally validated as miR-132 targets. However, most studies regarding miR-132 in major depressive disorder are based on post-mortem, animal models or genetic comparisons. This work will be the first attempt to investigate how miR-132 dysregulation may impact covariation of multimodal brain imaging data in 81 unmedicated major depressive patients and 123 demographically-matched healthy controls, as well as in a medication-naïve subset of major depressive patients. MiR-132 values in blood (patients > controls) was used as a prior reference to guide fusion of three MRI features: fractional amplitude of low frequency fluctuations, grey matter volume, and fractional anisotropy. The multimodal components correlated with miR-132 also show significant group difference in loadings. Results indicate that (i) higher miR-132 levels in major depressive disorder are associated with both lower fractional amplitude of low frequency fluctuations and lower grey matter volume in fronto-limbic network; and (ii) the identified brain regions linked with increased miR-132 levels were also associated with poorer cognitive performance in attention and executive function. Using a data-driven, supervised-learning method, we determined that miR-132 dysregulation in major depressive disorder is associated with multi-facets of brain function and structure in fronto-limbic network (the key network for emotional regulation and memory), which deepens our understanding of how miR-132 dysregulation in major depressive disorders contribute to the loss of specific brain areas and is linked to relevant cognitive impairments.

Connectome-based individualized prediction of temperament trait scores.

Temperament consists of multi-dimensional traits that affect various domains of human life. Evidence has shown functional connectome-based predictive models are powerful predictors of cognitive abilities. Putatively, individuals' innate temperament traits may be predictable by unique patterns of brain functional connectivity (FC) as well. However, quantitative prediction for multiple temperament traits at the individual level has not yet been studied. Therefore, we were motivated to realize the individualized prediction of four temperament traits (novelty seeking [NS], harm avoidance [HA], reward dependence [RD] and persistence [PS]) using whole-brain FC. Specifically, a multivariate prediction framework integrating feature selection and sparse regression was applied to resting-state fMRI data from 360 college students, resulting in 4 connectome-based predictive models that enabled prediction of temperament scores for unseen subjects in cross-validation. More importantly, predictive models for HA and NS could be successfully generalized to two relevant personality traits for unseen individuals, i.e., neuroticism and extraversion, in an independent dataset. In four temperament trait predictions, brain connectivities that show top contributing power commonly concentrated on the hippocampus, prefrontal cortex, basal ganglia, amygdala, and cingulate gyrus. Finally, across independent datasets and multiple traits, we show person's temperament traits can be reliably predicted using functional connectivity strength within frontal-subcortical circuits, indicating that human social and behavioral performance can be characterized by specific brain connectivity profile.

Predicting individualized clinical measures by a generalized prediction framework and multimodal fusion of MRI data.

Neuroimaging techniques have greatly enhanced the understanding of neurodiversity (human brain variation across individuals) in both health and disease. The ultimate goal of using brain imaging biomarkers is to perform individualized predictions. Here we proposed a generalized framework that can predict explicit values of the targeted measures by taking advantage of joint information from multiple modalities. This framework also enables whole brain voxel-wise searching by combining multivariate techniques such as ReliefF, clustering, correlation-based feature selection and multiple regression models, which is more flexible and can achieve better prediction performance than alternative atlas-based methods. For 50 healthy controls and 47 schizophrenia patients, three kinds of features derived from resting-state fMRI (fALFF), sMRI (gray matter) and DTI (fractional anisotropy) were extracted and fed into a regression model, achieving high prediction for both cognitive scores (MCCB composite r=0.7033, MCCB social cognition r=0.7084) and symptomatic scores (positive and negative syndrome scale [PANSS] positive r=0.7785, PANSS negative r=0.7804). Moreover, the brain areas likely responsible for cognitive deficits of schizophrenia, including middle temporal gyrus, dorsolateral prefrontal cortex, striatum, cuneus and cerebellum, were located with different weights, as well as regions predicting PANSS symptoms, including thalamus, striatum and inferior parietal lobule, pinpointing the potential neuromarkers. Finally, compared to a single modality, multimodal combination achieves higher prediction accuracy and enables individualized prediction on multiple clinical measures. There is more work to be done, but the current results highlight the potential utility of multimodal brain imaging biomarkers to eventually inform clinical decision-making.

Maximum Classifier Discrepancy Generative Adversarial Network for Jointly Harmonizing Scanner Effects and Improving Reproducibility of Downstream Tasks.

OBJECTIVE: Multi-site collaboration is essential for overcoming small-sample problems when exploring reproducible biomarkers in MRI studies. However, various scanner-specific factors dramatically reduce the cross-scanner replicability. Moreover, existing harmony methods mostly could not guarantee the improved performance of downstream tasks. METHODS: we proposed a new multi-scanner harmony framework, called 'maximum classifier discrepancy generative adversarial network', or MCD-GAN, for removing scanner effects in the original feature space while preserving substantial biological information for downstream tasks. Specifically, the adversarial generative network was utilized for persisting the structural layout of each sample, and the maximum classifier discrepancy module was introduced for regulating GAN generators by incorporating the downstream tasks. RESULTS: We compared the MCD-GAN with other state-of-the-art data harmony approaches (e.g., ComBat, CycleGAN) on simulated data and the Adolescent Brain Cognitive Development (ABCD) dataset. Results demonstrate that MCD-GAN outperformed other approaches in improving cross-scanner classification performance while preserving the anatomical layout of the original images. SIGNIFICANCE: To the best of our knowledge, the proposed MCD-GAN is the first generative model which incorporates downstream tasks while harmonizing, and is a promising solution for facilitating cross-site reproducibility in various tasks such as classification and regression.

Data leakage inflates prediction performance in connectome-based machine learning models.

Predictive modeling is a central technique in neuroimaging to identify brain-behavior relationships and test their generalizability to unseen data. However, data leakage undermines the validity of predictive models by breaching the separation between training and test data. Leakage is always an incorrect practice but still pervasive in machine learning. Understanding its effects on neuroimaging predictive models can inform how leakage affects existing literature. Here, we investigate the effects of five forms of leakage-involving feature selection, covariate correction, and dependence between subjects-on functional and structural connectome-based machine learning models across four datasets and three phenotypes. Leakage via feature selection and repeated subjects drastically inflates prediction performance, whereas other forms of leakage have minor effects. Furthermore, small datasets exacerbate the effects of leakage. Overall, our results illustrate the variable effects of leakage and underscore the importance of avoiding data leakage to improve the validity and reproducibility of predictive modeling.

Brain-phenotype predictions can survive across diverse real-world data.

Recent work suggests that machine learning models predicting psychiatric treatment outcomes based on clinical data may fail when applied to unharmonized samples. Neuroimaging predictive models offer the opportunity to incorporate neurobiological information, which may be more robust to dataset shifts. Yet, among the minority of neuroimaging studies that undertake any form of external validation, there is a notable lack of attention to generalization across dataset-specific idiosyncrasies. Research settings, by design, remove the between-site variations that real-world and, eventually, clinical applications demand. Here, we rigorously test the ability of a range of predictive models to generalize across three diverse, unharmonized samples: the Philadelphia Neurodevelopmental Cohort (n=1291), the Healthy Brain Network (n=1110), and the Human Connectome Project in Development (n=428). These datasets have high inter-dataset heterogeneity, encompassing substantial variations in age distribution, sex, racial and ethnic minority representation, recruitment geography, clinical symptom burdens, fMRI tasks, sequences, and behavioral measures. We demonstrate that reproducible and generalizable brain-behavior associations can be realized across diverse dataset features with sample sizes in the hundreds. Results indicate the potential of functional connectivity-based predictive models to be robust despite substantial inter-dataset variability. Notably, for the HCPD and HBN datasets, the best predictions were not from training and testing in the same dataset (i.e., cross-validation) but across datasets. This result suggests that training on diverse data may improve prediction in specific cases. Overall, this work provides a critical foundation for future work evaluating the generalizability of neuroimaging predictive models in real-world scenarios and clinical settings.