Connectome-Based Prediction of Cocaine Abstinence

Yip, Sarah W.; Scheinost, Dustin; Potenza, Marc N.; Carroll, Kathleen M.

doi:10.1176/appi.ajp.2018.17101147

Cited by 154 publications

(151 citation statements)

References 42 publications

Supporting

Mentioning

127

Contrasting

Unclassified

Order By: Relevance

“…Other imaging studies have used topological approaches to characterize how PTSD impacts neural networks 20,21 , and recent work suggests the utility of using machine learning approaches to predict and potentially identify those at risk for PTSD 22 . This area of inquiry has already significantly advanced the field in psychosis research and substance use, where machine learning can now identify patients using brain-based pathology 23 and individualized treatment response 24,25 .…”

Section: Introductionmentioning

confidence: 99%

Mapping PTSD symptoms to brain networks: a machine learning study

et al. 2020

View full text Add to dashboard Cite

Posttraumatic Stress Disorder (PTSD) is a prevalent and debilitating condition with complex and variable presentation. While PTSD symptom domains (intrusion, avoidance, cognition/mood, and arousal/reactivity) correlate highly, the relative importance of these symptom subsets often differs across patients. In this study, we used machine learning to derive how PTSD symptom subsets differ based upon brain functional connectivity. We acquired resting-state magnetic resonance imaging in a sample (N = 50) of PTSD patients and characterized clinical features using the PTSD Checklist for DSM-5 (PCL-5). We compared connectivity among 100 cortical and subcortical regions within the default mode, salience, executive, and affective networks. We then used principal component analysis and least-angle regression (LARS) to identify relationships between symptom domain severity and brain networks. We found connectivity predicted PTSD symptom profiles. The goodness of fit (R 2) for total PCL-5 score was 0.29 and the R 2 for intrusion, avoidance, cognition/mood, and arousal/reactivity symptoms was 0.33, 0.23, −0.01, and 0.06, respectively. The model performed significantly better than chance in predicting total PCL-5 score (p = 0.030) as well as intrusion and avoidance scores (p = 0.002 and p = 0.034). It was not able to predict cognition and arousal scores (p = 0.412 and p = 0.164). While this work requires replication, these findings demonstrate that this computational approach can directly link PTSD symptom domains with neural network connectivity patterns. This line of research provides an important step toward data-driven diagnostic assessments in PTSD, and the use of computational methods to identify individual patterns of network pathology that can be leveraged toward individualized treatment.

show abstract

Section: Introductionmentioning

confidence: 99%

Mapping PTSD symptoms to brain networks: a machine learning study

et al. 2020

View full text Add to dashboard Cite

show abstract

“…To model the RSFC‐phenotype association, four regression algorithms including connectome‐based predictive modeling (CPM), support vector regression (SVR), least absolute shrinkage and selection operator (LASSO), and Ridge regression have been frequently adopted for their good performances and interpretabilities (Coloigner, Phlypo, Bush, Lepore, & Wood, ; Cui & Gong, ; Dadi et al, ; de Vos et al, ; Gao, Greene, Constable, & Scheinost, ; Jiang et al, ; Meng et al, ; Ryali, Chen, Supekar, & Menon, ; Shen et al, ; Toiviainen, Alluri, Brattico, Wallentin, & Vuust, ). CPM is in fact a linear regression model and has been successfully applied in predicting intelligence (Beaty et al, ; Finn et al, ; Jiang et al, ), attention (Rosenberg, Finn, Scheinost, Constable, & Chun, ), cocaine abstinence (Yip, Scheinost, Potenza, & Carroll, ), and reading comprehension (Jangraw et al, ). SVR implements space transformation by some kernel function in order to achieve better prediction (Basak, Pal, & Patranabis, ), and has been utilized in prediction of mental disease (Rizk‐Jackson et al, ), brain maturity (Dosenbach et al, ; Nielsen et al, ), and painful sensation (Tu et al, ).…”

Section: Introductionmentioning

confidence: 99%

Bootstrapping promotes the RSFC‐behavior associations: An application of individual cognitive traits prediction

Wei

Jing

2020

Human Brain Mapping

View full text Add to dashboard Cite

Resting-state functional connectivity (RSFC) records enormous functional interaction information between any pair of brain nodes, which enriches the individualphenotypic prediction. To reduce high-dimensional features, correlation analysis is a common way for feature selection. However, resting state fMRI signal exhibits typically low signal-to-noise ratio and the correlation analysis is sensitive to outliers and data distribution, which may bring unstable features to prediction. To alleviate this problem, a bootstrapping-based feature selection framework was proposed and applied to connectome-based predictive modeling, support vector regression, least absolute shrinkage and selection operator, and Ridge regression to predict a series of cognitive traits based on Human Connectome Project data. To systematically investigate the influences of different parameter settings on the bootstrapping-based framework, 216 parameter combinations were evaluated and the best performance among them was identified as the final prediction result for each cognitive trait. By using the bootstrapping methods, the best prediction performances outperformed the baseline method in all four prediction models. Furthermore, the proposed framework could effectively reduce the feature dimension by retaining the more stable features. The results demonstrate that the proposed framework is an easy-to-use and effective method to improve RSFC prediction of cognitive traits and is highly recommended in future RSFC-prediction studies. K E Y W O R D S bootstrapping, cognitive trait, feature selection, functional connectivity, individualized prediction

show abstract

“…Hence, imposing a movement threshold in this sample, a relatively common procedure to limit the impact of motion on functional connectivity analyses (e.g. [47, 48, 50]), revealed that the formal mock scan group had more low-motion data relative to the informal mock scan group.…”

Section: Resultsmentioning

confidence: 99%

“…If a participant’s scan had a mean FFD above a threshold, it was designated as a “high motion” scan; if it was below the threshold, it was considered a “low motion” scan. Mean FFD thresholds of 0.10, 0.15, and 0.20 mm were used, as thresholds of this magnitude have been shown to limit the impact of motion [25, 47] while allowing for sample sizes of adequate size in children/adolescents [47, 48] and in those with a disorder [10, 49, 50].…”

Section: Methodsmentioning

confidence: 99%

Implementation of a pre- and in-scan system to minimize head motion in pediatric participants undergoing fMRI scans

Horien

Fontenelle

Joseph

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

BackgroundPerforming fMRI scans of children can be a difficult task, as participants tend to move while being scanned. Head motion represents a significant confound in functional magnetic resonance imaging (fMRI) connectivity analyses, and methods to limit the impact of movement on data quality are needed. One approach has been to use shorter MRI protocols, though this potentially reduces the reliability of the results. ObjectiveHere we describe steps we have taken to limit head motion in an ongoing fMRI study of children undergoing a 60 minute MRI scan protocol. Specifically, we have used a mock scan protocol that trains participants to lie still while being scanned. We provide a detailed protocol and describe other in-scanner measures we have implemented, including an incentive system and the use of a weighted blanket. Materials and methodsParticipants who received a formal mock scan (n = 12) were compared to participants who had an informal mock scan (n = 7). A replication group of participants (n = 16), including five with autism spectrum disorder, who received a formal mock scan were also compared to the informal mock scan group. The primary measure of interest was the mean frame-to-frame displacement across eight functional runs during the fMRI protocol. ResultsParticipants in the formal mock scan and replication group tended to exhibit more low-motion functional scans than the informal mock scan group (P < 0.05). Across different functional scan conditions (i.e. while watching movie clips, performing an attention task, and during resting-state scans), effect sizes tended to be large (Hedge's g > 0.8). ConclusionResults indicate that with appropriate measures, it is possible to achieve low-motion fMRI data in younger participants undergoing a long scan protocol.

show abstract

Connectome-Based Prediction of Cocaine Abstinence

Cited by 154 publications

References 42 publications

Mapping PTSD symptoms to brain networks: a machine learning study

Mapping PTSD symptoms to brain networks: a machine learning study

Bootstrapping promotes the RSFC‐behavior associations: An application of individual cognitive traits prediction

Implementation of a pre- and in-scan system to minimize head motion in pediatric participants undergoing fMRI scans

Contact Info

Product

Resources

About