Connectome-based Predictive Modeling Research Articles

Individualized prediction of consummatory anhedonia from functional connectome in major depressive disorder.

Anhedonia is a key symptom of major depressive disorder (MDD) and other psychiatric diseases. The neural basis of anhedonia has been widely examined, yet the interindividual variability in neuroimaging biomarkers underlying individual-specific symptom severity is not well understood. To establish an individualized prediction model of anhedonia, we applied connectome-based predictive modeling (CPM) to whole-brain resting-state functional connectivity profiles of MDD patients. The CPM can successfully and reliably predict individual consummatory but not anticipatory anhedonia. The predictive model mainly included salience network (SN), frontoparietal network (FPN), default mode network (DMN), and motor network. Importantly, subsequent computational lesion prediction and consummatory-specific model prediction revealed that connectivity of the SN with DMN and FPN is essential and specific for the prediction of consummatory anhedonia. This study shows that brain functional connectivity, especially the connectivity of SN-FPN and SN-DMN, can specifically predict individualized consummatory anhedonia in MDD. These findings suggest the potential of functional connectomes for the diagnosis and prognosis of anhedonia in MDD and other disorders.

Individualized morphometric similarity predicts body mass index and food approach behavior in school-age children.

Childhood obesity is associated with alterations in brain structure. Previous studies generally used a single structural index to characterize the relationship between body mass index(BMI) and brain structure, which could not describe the alterations of structural covariance between brain regions. To cover this research gap, this study utilized two independent datasets with brain structure profiles and BMI of 155 school-aged children. Connectome-based predictive modeling(CPM) was used to explore whether children's BMI is reliably predictable by the novel individualized morphometric similarity network(MSN). We revealed the MSN can predict the BMI in school-age children with good generalizability to unseen dataset. Moreover, these revealed significant brain structure covariant networks can further predict children's food approach behavior. The positive predictive networks mainly incorporated connections between the frontoparietal network(FPN) and the visual network(VN), between the FPN and the limbic network(LN), between the default mode network(DMN) and the LN. The negative predictive network primarily incorporated connections between the FPN and DMN. These results suggested that the incomplete integration of the high-order brain networks and the decreased dedifferentiation of the high-order networks to the primary reward networks can be considered as a core structural basis of the imbalance between inhibitory control and reward processing in childhood obesity.

Connectome-based prediction of eating disorder-associated symptomatology.

Despite increasing knowledge on the neuroimaging patterns of eating disorder (ED) symptoms in non-clinical populations, studies using whole-brain machine learning to identify connectome-based neuromarkers of ED symptomatology are absent. This study examined the association of connectivity within and between large-scale functional networks with specific symptomatic behaviors and cognitions using connectome-based predictive modeling (CPM). CPM with ten-fold cross-validation was carried out to probe functional networks that were predictive of ED-associated symptomatology, including body image concerns, binge eating, and compensatory behaviors, within the discovery sample of 660 participants. The predictive ability of the identified networks was validated using an independent sample of 821 participants. The connectivity predictive of body image concerns was identified within and between networks implicated in cognitive control (frontoparietal and medial frontal), reward sensitivity (subcortical), and visual perception (visual). Crucially, the set of connections in the positive network related to body image concerns identified in one sample was generalized to predict body image concerns in an independent sample, suggesting the replicability of this effect. These findings point to the feasibility of using the functional connectome to predict ED symptomatology in the general population and provide the first evidence that functional interplay among distributed networks predicts body shape/weight concerns.

Reduced hippocampal-cortical connectivity during memory suppression predicts the ability to forget unwanted memories.

The ability to suppress unwelcome memories is important for productivity and well-being. Successful memory suppression is associated with hippocampal deactivations and a concomitant disruption of this region's functionality. Much of the previous neuroimaging literature exploring such suppression-related hippocampal modulations has focused on the region's negative coupling with the prefrontal cortex. Task-based changes in functional connectivity between the hippocampus and other brain regions still need further exploration. In the present study, we utilize psychophysiological interactions and seed connectome-based predictive modeling to investigate the relationship between the hippocampus and the rest of the brain as 134 participants attempted to suppress unwanted memories during the Think/No-Think task. The results show that during retrieval suppression, the right hippocampus exhibited decreased functional connectivity with visual cortical areas (lingual and cuneus gyrus), left nucleus accumbens and the brain-stem that predicted superior forgetting of unwanted memories on later memory tests. Validation tests verified that prediction performance was not an artifact of head motion or prediction method and that the negative features remained consistent across different brain parcellations. These findings suggest that systemic memory suppression involves more than the modulation of hippocampal activity-it alters functional connectivity patterns between the hippocampus and visual cortex, leading to successful forgetting.

Functional connectivity of the central autonomic and default mode networks represent neural correlates and predictors of individual personality.

There is solid evidence for the prominent involvement of the central autonomic and default mode systems in shaping personality. However, whether functional connectivity of these systems can represent neural correlates and predictors of individual variation in personality traits is largely unknown. Resting-state functional magnetic resonance imaging data of 215 healthy young adults were used to construct the sympathetic (SN), parasympathetic (PN), and default mode (DMN) networks, with intra- and internetwork functional connectivity measured. Personality factors were assessed using the five-factor model. We examined the associations between personality factors and functional network connectivity, followed by performance of personality prediction based on functional connectivity using connectome-based predictive modeling (CPM), a recently developed machine learning approach. All personality factors (neuroticism, extraversion, conscientiousness, and agreeableness) other than openness were significantly correlated with intra- and internetwork functional connectivity of the SN, PN, and DMN. Moreover, the CPM models successfully predicted conscientiousness and agreeableness at the individual level using functional network connectivity. Our findings may expand existing knowledge regarding the neural substrates underlying personality.

Resting-state functional connectome predicts individual differences in depression during COVID-19 pandemic.

Stressful life events are significant risk factors for depression, and increases in depressive symptoms have been observed during the COVID-19 pandemic. The aim of this study is to explore the neural makers for individuals' depression during COVID-19, using connectome-based predictive modeling (CPM). Then we tested whether these neural markers could be used to identify groups at high/low risk for depression with a longitudinal dataset. The results suggested that the high-risk group demonstrated a higher level and increment of depression during the pandemic, as compared to the low-risk group. Furthermore, a support vector machine (SVM) algorithm was used to discriminate major depression disorder patients and healthy controls, using neural features defined by CPM. The results confirmed the CPM's ability for capturing the depression-related patterns with individuals' resting-state functional connectivity signature. The exploration for the anatomy of these functional connectivity features emphasized the role of an emotion-regulation circuit and an interoception circuit in the neuropathology of depression. In summary, the present study augments current understanding of potential pathological mechanisms underlying depression during an acute and unpredictable life-threatening event and suggests that resting-state functional connectivity may provide potential effective neural markers for identifying susceptible populations. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Connectome-based predictive models using resting-state fMRI for studying brain aging.

Changes in the brain with age can provide useful information regarding an individual's chronological age. studies have suggested that functional connectomes identified via resting-state functional magnetic resonance imaging (fMRI) could be a powerful feature for predicting an individual's age. We applied connectome-based predictive modeling (CPM) to investigate individual chronological age predictions via resting-state fMRI using open-source datasets. The significant feature for age prediction was confirmed in 168 subjects from the Southwest University Adult Lifespan Dataset. The higher contributing nodes for age production included a positive connection from the left inferior parietal sulcus and a negative connection from the right middle temporal sulcus. On the network scale, the subcortical-cerebellum network was the dominant network for age prediction. The generalizability of CPM, which was constructed using the identified features, was verified by applying this model to independent datasets that were randomly selected from the Autism Brain Imaging Data Exchange I and the Open Access Series of Imaging Studies 3. CPM via resting-state fMRI is a potential robust predictor for determining an individual's chronological age from changes in the brain.

Personality traits modulate the neural responses to handwriting processing.

Handwriting is a vital skill for everyday human activities. It has a wealth of information about writers' characteristics and can hint toward underlying neurological conditions, such as Parkinson's disease, autism, dyslexia, and attention-deficit/hyperactivity disorder (ADHD). Many previous studies have reported a link between personality and individual differences in handwriting, but the evidence for the relationship tends to be anecdotal in nature. Using functional magnetic resonance imaging (fMRI), we examined whether the association between personality traits and handwriting was instantiated at the neural level. Results showed that the personality trait of conscientiousness modulated brain activation in the left premotor cortex and right inferior/middle frontal gyrus, which may reflect the impact of personality on orthography-to-grapheme transformation and executive control involved in handwriting. Such correlations were not observed in symbol-drawing or word-reading tasks, suggesting the specificity of the link between conscientiousness and handwriting in these regions. Moreover, using a connectome-based predictive modeling approach, we found that individuals' conscientiousness scores could be predicted based on handwriting-related functional brain networks, suggesting that the influence of personality on handwriting may occur within a broader network. Our findings provide neural evidence for the link between personality and handwriting processing, extending our understanding of the nature of individual differences in handwriting.

Neural mechanisms underlying empathy during alcohol abstinence: evidence from connectome-based predictive modeling.

Empathy impairments have been linked to alcohol dependence even during abstinent periods. Nonetheless, the neural underpinnings of abstinence-induced empathy deficits remain unclear. In this study, we employed connectome-based predictive modeling (CPM) by using whole brain resting-state functional connectivity (rs-FC) to predict empathy capability of abstinent alcoholics (n = 47) versus healthy controls (n = 59). In addition, the generalizability of the predictive model (i.e., one group treated as a training dataset and another one treated as a test dataset) was performed to determine whether healthy controls and abstinent alcoholics share common neural fingerprints of empathy. Our results showed that abstinent alcoholics relative to healthy controls had decreased empathy capacity. Although no predictive models were observed in the abstinence group, we found that individual empathy scores in the healthy group can be reliably predicted by functional connectivity from the default mode network (DMN) to the sensorimotor network (SMN), occipital network, and cingulo-opercular network (CON). Moreover, the identified connectivity fingerprints of healthy controls could be generalized to predict empathy in the abstinence group. These findings indicate that neural circuits accounting for empathy may be disrupted by alcohol use and the impaired degree varies greatly among abstinent individuals. The large inter-individual variation may impede identification of the predictive model of empathy in alcohol abstainers.

Structural connectome-based prediction of trait anxiety.

Neurobiological research on anxiety has shown that trait-anxious individuals may be characterized by weaker structural connectivity of the amygdala-prefrontal circuitry, representing a reduced capacity for efficient communication between the two brain regions. However, comparison of available studies has been inconsistent, possibly related to factors such as aging that influences both trait anxiety and structural connectivity of the brain. To help clarify the nature of brain-anxiety relationship, we applied a connectome-based predictive modeling framework on 148 diffusion-weighted imaging data from the Leipzig Study for Mind-Body Emotion Interactions dataset and identified multivariate patterns of whole-brain structural connectivity that predicted trait anxiety. Results showed that networks predictive of trait anxiety differed across age groups. Specifically, an isolated negative network, which shared overlapping features with the amygdala-prefrontal circuitry, was found in younger adults (20-30 years of age), whereas a widespread positive network highlighted by frontotemporal and frontolimbic connectivity was identified when both younger and older adults (20-80 years of age) were examined. No predictive network was observed when only older adults (30-80 years of age) were considered. Our findings highlight an important age-dependent effect on the structural connectome-based prediction of trait anxiety, supporting ongoing efforts to develop potential neural biomarkers of anxiety.

Prediction of childhood maltreatment and subtypes with personalized functional connectome of large-scale brain networks.

Childhood maltreatment (CM) has a long impact on physical and mental health of children. However, the neural underpinnings of CM are still unclear. In this study, we aimed to establish the associations between functional connectome of large‐scale brain networks and influences of CM evaluated through Childhood Trauma Questionnaire (CTQ) at the individual level based on resting‐state functional magnetic resonance imaging data of 215 adults. A novel individual functional mapping approach was employed to identify subject‐specific functional networks and functional network connectivities (FNCs). A connectome‐based predictive modeling (CPM) was used to estimate CM total and subscale scores using individual FNCs. The CPM established with FNCs can well predict CM total scores and subscale scores including emotion abuse, emotion neglect, physical abuse, physical neglect, and sexual abuse. These FNCs primarily involve default mode network, fronto‐parietal network, visual network, limbic network, motor network, dorsal and ventral attention networks, and different networks have distinct contributions to predicting CM and subtypes. Moreover, we found that CM showed age and sex effects on individual functional connections. Taken together, the present findings revealed that different types of CM are associated with different atypical neural networks which provide new clues to understand the neurobiological consequences of childhood adversity.

Behavioral Studies Using Large-Scale Brain Networks - Methods and Validations.

Mapping human behaviors to brain activity has become a key focus in modern cognitive neuroscience. As methods such as functional MRI (fMRI) advance cognitive scientists show an increasing interest in investigating neural activity in terms of functional connectivity and brain networks, rather than activation in a single brain region. Due to the noisy nature of neural activity, determining how behaviors are associated with specific neural signals is not well-established. Previous research has suggested graph theory techniques as a solution. Graph theory provides an opportunity to interpret human behaviors in terms of the topological organization of brain network architecture. Graph theory-based approaches, however, only scratch the surface of what neural connections relate to human behavior. Recently, the development of data-driven methods, e.g., machine learning and deep learning approaches, provide a new perspective to study the relationship between brain networks and human behaviors across the whole brain, expanding upon past literatures. In this review, we sought to revisit these data-driven approaches to facilitate our understanding of neural mechanisms and build models of human behaviors. We start with the popular graph theory approach and then discuss other data-driven approaches such as connectome-based predictive modeling, multivariate pattern analysis, network dynamic modeling, and deep learning techniques that quantify meaningful networks and connectivity related to cognition and behaviors. Importantly, for each topic, we discuss the pros and cons of the methods in addition to providing examples using our own data for each technique to describe how these methods can be applied to real-world neuroimaging data.

The brain network underlying attentional blink predicts symptoms of attention deficit hyperactivity disorder in children.

Attention deficit hyperactivity disorder (ADHD) is a chronic neuropsychiatric disease that can markedly impair educational, social, and occupational function throughout life. Behavioral deficits may provide clues to the underlying neurological impairments. Children with ADHD exhibit a larger attentional blink (AB) deficit in rapid serial visual presentation (RSVP) tasks than typically developing children, so we examined whether brain connectivity in the neural network associated with AB can predict ADHD symptoms and thus serve as potential biomarkers of the underlying neuropathology. We first employed a connectome-based predictive model analysis of adult resting-state functional magnetic resonance imaging data to identify a distributed brain network for AB. The summed functional connectivity (FC) strength within the AB network reliably predicted individual differences in AB magnitude measured by a classical dual-target RSVP task. Furthermore, the summed FC strength within the AB network predicted individual differences in ADHD Rating Scale scores from an independent dataset of pediatric patients. Our findings suggest that the individual AB network could serve as an applicable neuroimaging-based biomarker of AB deficit and ADHD symptoms.

Multi-modality connectome-based predictive modeling of individualized compulsions in obsessive-compulsive disorder

BackgroundWhile previous neuroimaging studies are mainly focused on dichotomous classification of obsessive-compulsive disorder (OCD) from controls, predicting continuous severity of specific symptom is also pivotal to clinical diagnosis and treatment. MethodsWe applied a machine-learning approach, connectome-based predictive modeling, on functional and structural brain networks constructed from resting-state functional magnetic resonance imaging and diffusion tensor imaging data to decode compulsions and obsessions of fifty-four patients with OCD. ResultsWe successfully predicted individualized compulsions with a positive model of structural brain network and with a negative model of functional brain network. The structural predictive brain network comprises the motor cortex, cerebellum and limbic lobe, which are involved in basic motor control, motor execution and emotion processing, respectively. The functional predictive brain network is composed by the prefrontal and limbic systems which are related to cognitive and affective control. Computational lesion analysis shows that functional connectivity among the salience network (SN), the frontal parietal network and the default mode network, as well as structural connectivity within the SN are vital in the individualized prediction of compulsions in OCD. LimitationsThere was no external validation of large samples to test the robustness of our predictive model. ConclusionsThese findings provide the first evidence for the predictive role of the triple network model in individualized compulsions and have important implications in diagnosis, prognosis and treatment of patients with OCD.

Abnormal Brain Network Interaction Associated With Positive Symptoms in Drug-Naive Patients With First-Episode Schizophrenia.

Positive symptoms are marked features of schizophrenia, and emerging evidence has suggested that abnormalities of the brain network underlying these symptoms may play a crucial role in the pathophysiology of the disease. We constructed two brain functional networks based on the positive and negative correlations between positive symptom scores and brain connectivity in drug-naive patients with first-episode schizophrenia (FES, n = 45) by using a machine-learning approach (connectome-based predictive modeling, CPM). The accuracy of the model was r = 0.47 (p = 0.002). The positively and negatively associated network strengths were then compared among FES subjects, individuals at genetic high risk (GHR, n = 41) for schizophrenia, and healthy controls (HCs, n = 48). The results indicated that the positively associated network contained more cross-subnetwork connections (96.02% of 176 edges), with a focus on the default-mode network (DMN)-salience network (SN) and the DMN-frontoparietal task control (FPT) network. The negatively associated network had fewer cross-subnetwork connections (71.79% of 117 edges) and focused on the sensory/somatomotor hand (SMH)-Cingulo opercular task control (COTC) network, the DMN, and the visual network with significantly decreased connectivity in the COTC-SMH network in FES (FES < GHR, p = 0.01; FES < HC, p = 0.01). Additionally, the connectivity strengths of the right supplementary motor area (SMA) (p < 0.001) and the right precentral gyrus (p < 0.0001) were reduced in FES. To the best of our knowledge, this is the first study to generate two brain networks associated with positive symptoms by utilizing CPM in FES. Abnormal segregation, interactions of brain subnetworks, and impaired SMA might lead to salience attribution abnormalities and, thus, as a result, induce positive symptoms in schizophrenia.

Connectome-based prediction of marital quality in husbands' processing of spousal interactions.

Marital quality may decrease during the early years of marriage. Establishing models predicting individualized marital quality may help develop timely and effective interventions to maintain or improve marital quality. Given that marital interactions have an important impact on marital well-being cross-sectionally and prospectively, neural responses during marital interactions may provide insight into neural bases underlying marital well-being. The current study applies connectome-based predictive modeling, a recently developed machine-learning approach, to functional magnetic resonance imaging (fMRI) data from both partners of 25 early-stage Chinese couples to examine whether an individual's unique pattern of brain functional connectivity (FC) when responding to spousal interactive behaviors can reliably predict their own and their partners' marital quality after 13 months. Results revealed that husbands' FC involving multiple large networks, when responding to their spousal interactive behaviors, significantly predicted their own and their wives' marital quality, and this predictability showed gender specificity. Brain connectivity patterns responding to general emotional stimuli and during the resting state were not significantly predictive. This study demonstrates that husbands' differences in large-scale neural networks during marital interactions may contribute to their variability in marital quality and highlights gender-related differences. The findings lay a foundation for identifying reliable neuroimaging biomarkers for developing interventions for marital quality early in marriages.

Reporting details of neuroimaging studies on individual traits prediction: A literature survey

Using machine-learning tools to predict individual phenotypes from neuroimaging data is one of the most promising and hence dynamic fields in systems neuroscience. Here, we perform a literature survey of the rapidly work on phenotype prediction in healthy subjects or general population to sketch out the current state and ongoing developments in terms of data, analysis methods and reporting. Excluding papers on age-prediction and clinical applications, which form a distinct literature, we identified a total 108 papers published since 2007. In these, memory, fluid intelligence and attention were most common phenotypes to be predicted, which resonates with the observation that roughly a quarter of the papers used data from the Human Connectome Project, even though another half recruited their own cohort. Sample size (in terms of training and external test sets) and prediction accuracy (from internal and external validation respectively) did not show significant temporal trends. Prediction accuracy was negatively correlated with sample size of the training set, but not the external test set. While known to be optimistic, leave-one-out cross-validation (LOO CV) was the prevalent strategy for model validation (n = 48). Meanwhile, 27 studies used external validation with external test set. Both numbers showed no significant temporal trends. The most popular learning algorithm was connectome-based predictive modeling introduced by the Yale team. Other common learning algorithms were linear regression, relevance vector regression (RVR), support vector regression (SVR), least absolute shrinkage and selection operator (LASSO), and elastic net. Meanwhile, the amount of data from self-recruiting studies (but not studies using open, shared dataset) was positively correlated with internal validation prediction accuracy. At the same time, self-recruiting studies also reported a significantly higher internal validation prediction accuracy than those using open, shared datasets. Data type and participant age did not significantly influence prediction accuracy. Confound control also did not influence prediction accuracy after adjusted for other factors. To conclude, most of the current literature is probably quite optimistic with internal validation using LOO CV. More efforts should be made to encourage the use of external validation with external test sets to further improve generalizability of the models.

Neuromarkers of Risky Alcohol Use From Age 14 to 19 Years

This study tested the ability of a whole-brain, connectome-based predictive modeling (CPM) approach, to identify neuromarkers of current and future alcohol use severity in a large, multi-site, longitudinal sample of adolescents from the IMAGEN consortium (N∼2,000). CPM is a machine learning method of generating behavioral predictions from functional connectivity data (referred to as ‘neural fingerprinting’).

Predicting Transdiagnostic Social Impairments in Childhood Using Connectome-Based Predictive Modeling

Social impairments are core features of multiple neurodevelopmental disorders. Previous neuroimaging studies have focused on predicting associations between brain function within disorders but rarely use brain-based models for the transdiagnostic prediction of symptoms. This study used a machine learning approach to examine functional connectivity that predicts elevated social impairments in a transdiagnostic sample of youths. We hypothesized a widespread network of edges would predict social impairments, across diagnoses, particularly in brain regions involved in social cognition.

Resting-state functional connectivity of social brain regions predicts motivated dishonesty

Motivated dishonesty is a typical social behavior varying from person to person. Resting-state fMRI (rsfMRI) is capable of identifying unique patterns from functional connectivity (FC) between brain regions. Recent work has built a link between brain networks in resting state to dishonesty in Western participants. To determine and reproduce the relevant neural patterns and build an interpretable model to predict dishonesty, we analyzed two conceptually similar datasets containing rsfMRI data with different dishonesty tasks. Both tasks implemented the information-passing paradigm, in which monetary rewards were employed to induce dishonesty. We applied connectome-based predictive modeling (CPM) to build a model among FC within and between four social brain networks (reward, self-referential, moral, and cognitive control). The CPM analysis indicated that FCs of social brain networks are predictive of dishonesty rate, especially FCs within reward network, and between self-referential and cognitive control networks. Our study offers an conceptual replication with integrated model to predict dishonesty with rsfMRI, and the results suggest that frequent motivated dishonest decisions may require the higher engagement of social brain regions.