Longitudinal Network Research Articles

Depression and PTSD in the aftermath of strict COVID-19 lockdowns: a cross-sectional and longitudinal network analysis

ABSTRACT Background: Post-traumatic stress disorder (PTSD) and major depressive disorder (MDD) are two highly comorbid psychological outcomes commonly studied in the context of stress and potential trauma. In Hubei, China, of which Wuhan is the capital, residents experienced unprecedented stringent lockdowns in the early months of 2020 when COVID-19 was first reported. The comorbidity between PTSD and MDD has been previously studied using network models, but often limited to cross-sectional data and analysis. Objectives: This study aims to examine the cross-sectional and longitudinal network structures of MDD and PTSD symptoms using both undirected and directed methods. Methods: Using three types of network analysis – cross-sectional undirected network, longitudinal undirected network, and directed acyclic graph (DAG) – we examined the interrelationships between MDD and PTSD symptoms in a sample of Hubei residents assessed in April, June, August, and October 2020. We identified the most central symptoms, the most influential bridge symptoms, and causal links among symptoms. Results: In both cross-sessional and longitudinal networks, the most central depressive symptoms included sadness and depressed mood, whereas the most central PTSD symptoms changed from irritability and hypervigilance at the first wave to difficulty concentrating and avoidance of potential reminders at later waves. Bridge symptoms showed similarities and differences between cross-sessional and longitudinal networks with irritability/anger as the most influential bridge longitudinally. The DAG found feeling blue and intrusive thoughts the gateways to the emergence of other symptoms. Conclusions: Combining cross-sectional and longitudinal analysis, this study elucidated central and bridge symptoms and potential causal pathways among PTSD and depression symptoms. Clinical implications and limitations are discussed.

Examining Social Influences on Why We Drink: Perceived Drinking Motives in the Social Network Impact Individuals’ Own Drinking Motives and Alcohol Use

Introduction: A significant body of research has investigated the impacts of social influence and social selection on binge drinking and risk factors for binge drinking in emerging adults; however, one risk factor for binge drinking that has yet to be thoroughly investigated in this regard is drinking motives. Preliminary research suggests the motives of others may impact emerging adults’ own alcohol use indirectly through their own motives (i.e., social influence). While these are important findings, research to date has been only conducted with adolescents or dyads and has not examined selection (i.e., selecting social network members with similar motives). We filled these gaps with a longitudinal egocentric social network design. Methods: Emerging adults (N = 177) completed measures on their alcohol use, drinking motives, and social networks at baseline (T1) and four-month follow-up (T2). Results: A cross-lagged panel model indicated T1 perceived network drinking motives predicted T2 participant drinking motives (for all motives but social), but T1 participant drinking motives did not predict T2 perceived network drinking motives. Path analysis indicated T1 perceived network drinking motives predicted T2 participant binge drinking frequency indirectly through T2 participant drinking motives for enhancement, coping-with-anxiety, and conformity, but not social or coping-with-depression, motives. Discussion: Results suggests drinking motives of those around emerging adults impact their own drinking motives, and indirectly, their own alcohol use. We found evidence of social influence, but not social selection. Conclusion: It appears that those around emerging adults have the capacity to influence their drinking behaviors and drinking motives.

New trends in network modeling of psychopathology.

New trends in network modeling of psychopathology.

Network Structure of Autism Spectrum Disorder Behaviors and Its Evolution in Preschool Children: Insights from a New Longitudinal Network Analysis Method.

Network modeling of the social, communication and restrictive/repetitive behaviors (RRBs) included in the definition of Autism Spectrum Disorder was performed. The Autism Diagnostic Interview-Revised (ADI-R) assessed behaviors in 139 pre-school cases at two cross-sections that averaged 34.8months apart. Cross-sectional networks were based on the correlation matrix of the ADI-R behavioral items and the "bootCross" method was developed and enabled the estimation of a longitudinal network. At both stages, RRB items/nodes formed a consistent peripheral cluster, while social and communication nodes formed a core cluster that diverged with time. These differences in the nature and evolution of the RRB and socio-communicative dimensions indicate that their inter-behavior dynamics are very different. The most central behaviors across stages are proposed as prime targets for efficient therapeutic intervention.

Relationships in context and contexts of relationship: a normative transition at the end of Secondary Education

High school is an institution where teens spend a large part of their time. Consequently, in this context, a dense network of relationships tends to be formed during the period of study at the institute. However, at the end of secondary school, this set of contextualized relationships begins to disintegrate. In this study, we analyze the transition from high school to university of a cohort of high school students. First, we interviewed all the students who were in their last year of high school at an institute in the province of Seville (n = 69). A year and a half later, we interviewed them again when many of them were already studying at university (n = 57). Using a hybrid research design, we collected personal network data from each interviewee in the context of a longitudinal whole network design of the entire cohort analyzed. The results show a decrease in the structural cohesion of the networks, both for personal networks and for the complete network. Furthermore, less cohesive personal networks seem more likely to experience short-term changes in ecological transition processes. In the discussion we reflect on the impact of educational transitions on friendship relationships.

Hopelessness during and after the COVID-19 pandemic lockdown among Chinese college students:A longitudinal network analysis.

IntroductionIn China, recurrent pandemics require frequent city‐wide lockdowns and quarantine actions to contain the impact of COVID‐19, exposing college students to psychological problems, including hopelessness. Hence, the purpose of helping problematic college students alleviate hopelessness symptoms motivates us to carry out the present study to explore their interrelationship.MethodsHopelessness (i.e., a complex phenomenon with important clinical consequences, such as depression and suicidality) was investigated in a large longitudinal sample of college students (N = 2787; 58.59% female; age mean ± SD = 18.34 ± 0.92) who were recruited during and after the COVID‐19 lockdown using the Beck Hopelessness Scale (BHS).ResultsApplying the novel approach (i.e., symptom network analysis), the results indicated that the edge of #BHS1 (i.e., [NOT] hope‐enthusiasm)–#BHS15 (i.e., [NOT] faith‐in‐the‐future) showed the strongest association both in Wave 1 and Wave 2. Similarly, #BHS20 (i.e., not‐trying) had the highest node expectedinfluence (centrality) in the hopelessness symptoms network both among Wave 1 and Wave 2. The Network Comparison Test indicated that the global network strength significantly differed between the two time points. As expected, college students' hopelessness will gradually dissipate with the end of segregation control. The stability and accuracy indicated that the network analysis results were trustworthy.ConclusionsThe study findings provide evidence that central nodes and edges connecting symptoms should be addressed. Further interventions and treatments that may target these symptoms are essential to effectively alleviate the overall hopelessness level among college students. Theoretical and clinical potential consequences were discussed in detail.

Genetic network analysis of human immunodeficiency virus sexual transmission in rural Southwest China after the expansion of antiretroviral therapy: A population-based study.

This study is used to analyze the genetic network of HIV sexual transmission in rural areas of Southwest China after expanding antiretroviral therapy (ART) and to investigate the factors associated with HIV sexual transmission through the genetic network. This was a longitudinal genetic network study in Guangxi, China. The baseline survey and follow-up study were conducted among patients with HIV in 2015, and among those newly diagnosed from 2016 to 2018, respectively. A generalized estimating equation model was employed to explore the factors associated with HIV transmission through the genetic linkage between newly diagnosed patients with HIV (2016-2018) and those at baseline (2015-2017), respectively. Of 3,259 identified HIV patient sequences, 2,714 patients were at baseline, and 545 were newly diagnosed patients with HIV at follow-up. A total of 8,691 baseline objectives were observed by repeated measurement analysis. The prevention efficacy in HIV transmission for treated HIV patients was 33% [adjusted odds ratio (AOR): 0.67, 95% confidence interval (CI): 0.48-0.93]. Stratified analyses indicated the prevention efficacy in HIV transmission for treated HIV patients with a viral load (VL) of <50 copies/ml and those treated for 4 years with a VL of <50 copies/ml to be 41 [AOR: 0.59, 95% CI: 0.43-0.82] and 65% [AOR: 0.35, 95% CI: 0.24-0.50], respectively. No significant reduction in HIV transmission occurred among treated HIV patients with VL missing or treated HIV patients on dropout. Some factors were associated with HIV transmission, including over 50 years old, men, Zhuang and other nationalities, with less than secondary schooling, working as a farmer, and heterosexual transmission. This study reveals the role of ART in reducing HIV transmission, and those older male farmers with less than secondary schooling are at high risk of HIV infection at a population level. Improvements to ART efficacy for patients with HIV and precision intervention on high-risk individuals during the expansion of ART are urgently required.

Longitudinal Assessment of Posttreatment Diffuse Glioma Tissue Volumes with Three-dimensional Convolutional Neural Networks.

Neural networks were trained for segmentation and longitudinal assessment of posttreatment diffuse glioma. A retrospective cohort (from January 2018 to December 2019) of 298 patients with diffuse glioma (mean age, 52 years ± 14 [SD]; 177 men; 152 patients with glioblastoma, 72 patients with astrocytoma, and 74 patients with oligodendroglioma) who underwent two consecutive multimodal MRI examinations were randomly selected into training (n = 198) and testing (n = 100) samples. A posttreatment tumor segmentation three-dimensional nnU-Net convolutional neural network with multichannel inputs (T1, T2, and T1 postcontrast and fluid-attenuated inversion recovery [FLAIR]) was trained to segment three multiclass tissue types (peritumoral edematous, infiltrated, or treatment-changed tissue [ED]; active tumor or enhancing tissue [AT]; and necrotic core). Separate longitudinal change nnU-Nets were trained on registered and subtracted FLAIR and T1 postlongitudinal images to localize and better quantify and classify changes in ED and AT. Segmentation Dice scores, volume similarities, and 95th percentile Hausdorff distances ranged from 0.72 to 0.89, 0.90 to 0.96, and 2.5 to 3.6 mm, respectively. Accuracy rates of the posttreatment tumor segmentation and longitudinal change networks being able to classify longitudinal changes in ED and AT as increased, decreased, or unchanged were 76%-79% and 90%-91%, respectively. The accuracy levels of the longitudinal change networks were not significantly different from those of three neuroradiologists (accuracy, 90%-92%; κ, 0.58-0.63; P > .05). The results of this study support the potential clinical value of artificial intelligence-based automated longitudinal assessment of posttreatment diffuse glioma. Keywords: MR Imaging, Neuro-Oncology, Neural Networks, CNS, Brain/Brain Stem, Segmentation, Quantification, Convolutional Neural Network (CNN) Supplemental material is available for this article. © RSNA, 2022.

Measuring zero water level in stream reaches: A comparison of an image‐based versus a conventional method

AbstractA limited number of gauging stations, especially for nested catchments, hampers a process understanding of the interaction between streamflow, groundwater and water usage during drought. Non‐commercial measurement devices can help overcome this lack of monitoring, but they need to be thoroughly tested. The Dreisam River in the South‐West of Germany was affected by several hydrological drought events from 2015 to 2020 during which parts of the main stream and tributaries fell dry. Therefore it provided a useful case study area for a flexible longitudinal water quality and quantity monitoring network. Among other measurements the setup employs an image‐based method with QR codes as fiducial marker. In order to assess under which conditions the QR‐code based water level loggers (WLL) deliver data according to scientific standards, we compared its performance to conventional capacitive based WLL. The results from 20 monitoring stations reveal that the riverbed was dry for &gt;50% at several locations and even for &gt;70% at most severely affected locations during July and August 2020, with the north western parts of the catchment being especially concerned. Highly variable longitudinal drying patterns of the stream reaches emerged from the monitoring. The image‐based method was found valuable for identification and validation of zero level occurrences. Nevertheless, a simple image processing approach (based on an automatic thresholding algorithm) did not compensate for errors due to natural conditions and technical setup. Our findings highlight that the complexity of measurement environments is a major challenge when working with image‐based methods.

Social comparison effects on academic self-concepts-Which peers matter most?

Social comparisons with peers are important sources of self-development during adolescence. Many previous studies showed that students’ academic self-concepts (ASC) form by contrasting one’s own achievement with the average of one’s class or school (the Big-Fish-Little-Pond Effect [BFLPE]). Based on social comparison theory, however, we would expect some peers to be more likely social comparison targets than other peers, for example, because they are more visible or students perceive them as similar to themselves. In this study, we used sociometric data to analyze which peers play the most important role for social comparison effects on ASC. We examined how the average achievement of friends, study partners, peers perceived as popular by the student, as well as same-gender and same-ethnic peers affect the general ASC and how these effects compare to the effect of the classroom’s average achievement. The study was based on a German longitudinal sample of 2,438 students (44% no recent immigrant background, 19% Turkish immigrant background, 10% Eastern European immigrant background, 27% other immigrant background) from 117 school classes that were followed from grade 9 to 10. Results from longitudinal social network analysis do not confirm substantial incremental effects of specific types of peers, while class average achievement showed a stable negative effect (confirming the BFLPE). In addition, we could provide evidence for social selection effects based on ASC. We conclude that classrooms provide a specific setting that imposes social comparisons with the “generalized peer” rather than with specific subgroups of peers.

Peer Entrepreneur Networks in a Startup Accelerator Before and During the Covid-19 Pandemic

In response to the global spread of COVID-19, startup accelerators have shifted all activities online. To understand how this shift affects peer entrepreneur networks – a key constituent of accelerators value proposition to participating entrepreneurs – we draw on longitudinal network data from two editions of the same startup accelerator program before and during the COVID-19 pandemic. Based on interaction ritual theory and the physical proximity principle, we hypothesize and test how online acceleration limits social connectivity. We find that in the online, compared to the offline cohort, networks became less dense, entrepreneurs reached fewer peers via indirect connections; and clustering increased. We highlight the impact of physical distancing on entrepreneurship and discuss contributions to theory on peer networks and startup acceleration.

Internal Dynamics of Business Ecosystems: A Network Approach

Scholarly interest in business ecosystems has seen a steady increase over the past decades. Surprisingly, while many studies have dealt with theoretical approaches, the architecture or outcomes of ecosystems, literature on their internal dynamics remains scarce. Without understanding these dynamics, however, our understanding of business ecosystems remains incomplete. In this paper, we investigate the dynamics of business ecosystems in the pharmaceutical sector, arguing that firms exhibit different partner selection mechanisms depending on which attribute they choose their partners on. Specifically, we propose that firms will show tendencies for aspiration when considering the size of potential partners, homophily when considering their inventiveness and conformity when considering their profitability. Applying a longitudinal network approach, we empirically test our hypotheses on the business ecosystems of four major US-based pharmaceutical firms, which we observed over a 4-year period from 2016 to 2019. Our results give a nuanced depiction of the mechanisms underlying the creation and dissolution of partnerships in business ecosystems. With the exception of conformity as the dominant mechanism regarding profitability, our results support our predictions.

Socio-Cognitive Processes and Peer-Network Influences in Defending and Bystanding

Peers are critical to defending and bystanding during episodes of bullying. This study investigates the extent to which friends can shape defending and bystanding as well as social cognitions associated with these two behaviors (i.e., perceptions of self-efficacy and moral distress). The study sample consisted of n = 1354 early and middle adolescents (7th‒10th grade; 81.4% Italian; 51.3% boys) in northern Italy. Employing a longitudinal social network analytic approach, using stochastic actor-oriented modeling, this study found that adolescents become more similar or stay similar to their friends in both behaviors and perceptions, with no clear indication that students select friends based on similar levels of behaviors or perceptions. The findings illustrate how defending and bystanding behaviors and related social cognitions are developed within friend (peer) networks.

Extent and implications of omitted ties on network measures in a longitudinal social network survey of people who use drugs

BackgroundPrevious research has revealed under-reporting of personal network members (i.e., alters) in studies involving people who use drugs (PWUD). This analysis (1) characterizes relationships that were more likely to be omitted but later recalled with prompting and (2) identifies network structural characteristics most impacted by these omissions among a sample of PWUD in rural Appalachian Kentucky, an epicenter of the opioid epidemic. MethodsData were collected through longitudinal assessments as part of the Social Networks Among Appalachian People (SNAP) study (2008–2017). Study participants completed interviewer-administered questionnaires that collected social network data via free-listing at baseline and six-month intervals. At visit 5, after free-listing, interviewers prompted participants with the names of previously reported alters. We used modified Poisson regression with generalized estimating equations to identify individual- and relationship-level characteristics associated with an alter being reported only after prompting. We examined the impact of including vs. excluding relationships reported after prompting on local and global sociometric network measures (i.e., betweenness centrality, bridging, density, mean degree, transitivity, cliques, and 2-cores). ResultsRelationships reported only after prompting were more likely to be immediate family (Adjusted Prevalence Ratio [APR]:1.29; 95% Confidence Interval [CI]: 1.03–1.63) and less likely to involve sex (APR:0.54; 95% CI: 0.43–0.67). Considerable differences were observed for participant positional rankings of betweenness centrality and bridging, and differences in network density and average degree pre- and post-prompting were statistically significant. ConclusionLongitudinal network studies that aim to assess transmission dynamics, information diffusion, or peer influence should consider the effects of omitted relationships.

Abstract 5145: Project Survival®: Discovery of a molecular-clinical phenome biomarker panel to detect pancreatic ductal adenocarcinoma among at risk populations using high-fidelity longitudinal phenotypic and multi-omic analysis

Abstract Delayed diagnosis and rapid progression are major drivers of poor survival outcomes for pancreatic ductal adenocarcinoma (PDAC). PDAC is expected to be the second leading cause of cancer death and has a dismal 5 year survival rate of 10%. There is an urgent unmet need to detect the disease at an early stage and stratify patients into more effective treatment regimens within clinically meaningful timeframes. To accomplish this, robust quality controlled OMIC molecular profiling platforms and analytic solutions need to be deployed into precision medicine protocols to discover actionable biomarkers. Project Survival® is a multicenter (n=6), prospective biomarker study (NCT 02781012) of PDAC and relevant controls combining high-fidelity longitudinal phenotypic characterization, multi-omic profiling (proteomics, signaling lipidomics, structural lipidomics, and metabolomics), and agnostic Bayesian artificial intelligence network inference (bAIcis®) to discover biomarkers with diagnostic and therapeutic utility. This study utilizes a systems medicine approach for translational biomarker discovery by performing analysis of matched subject sera, plasma, buffy coat, saliva, urine, and tumor/adjacent normal tissues and integrating them with the respective full clinical annotation using the BERG Interrogative Biology® platform. Multiple longitudinal time points were taken over the course of the six-year timeline enabling dynamic modeling. Utilizing the Project Survival® molecular and clinical data, we have analyzed and integrated baseline samples from 121 at risk patients and 279 patients with PDAC. Samples were randomized and analyzed over the course of recruitment allowing for agnostic discovery and integration to determine diagnostic utility. Discovery analysis identified 123 potential molecular markers, of which, four demonstrated a combined AUC of 0.85, PPV 0.83, NPV 0.72, and OR 13.1. In parallel, 4 non-canonical clinical measurements were assessed for diagnostic utility providing an AUC 0.79, PPV 0.84, NPV 0.72 and OR 13.2. Combining molecular and clinical features demonstrated an AUC of 0.9, PPV 0.9, NPV 0.77, OR 29.2, and p-value 1.4 E-40. Molecular markers revealed no treatment associated expression effects. Taken together, these marker panels demonstrate diagnostic utility to detect PDAC and will be further validated using robust bioanalysis methods as well as in an independent cohort of samples to provide enhanced insight into their positioning in the diagnostic landscape for PDAC. Citation Format: Eric M. Grund, A. James Moser, Corinne L. DeCicco, Nischal M. Chand, Genesis L. Perez-Melara, Gregory M. Miller, Punit Shah, Valarie Bussberg, Vladimir Tolstikov, Rangaprasad Sarangarajan, Elder Granger, Niven Narian, Michael A. Kiebish. Project Survival®: Discovery of a molecular-clinical phenome biomarker panel to detect pancreatic ductal adenocarcinoma among at risk populations using high-fidelity longitudinal phenotypic and multi-omic analysis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5145.

The emergence of multiplex dynamics between information provision ties and rescue collaboration ties: a longitudinal network analytic approach to flooding cases in Myanmar

The emergence of multiplex dynamics between information provision ties and rescue collaboration ties: a longitudinal network analytic approach to flooding cases in Myanmar

Using Machine Learning to Examine Suicidal Ideation After Traumatic Brain Injury: A Traumatic Brain Injury Model Systems National Database Study.

The aim of the study was to predict suicidal ideation 1 yr after moderate to severe traumatic brain injury. This study used a cross-sectional design with data collected through the prospective, longitudinal Traumatic Brain Injury Model Systems network at hospitalization and 1 yr after injury. Participants who completed the Patient Health Questionnaire-9 suicide item at year 1 follow-up ( N = 4328) were included. A gradient boosting machine algorithm demonstrated the best performance in predicting suicidal ideation 1 yr after traumatic brain injury. Predictors were Patient Health Questionnaire-9 items (except suicidality), Generalized Anxiety Disorder-7 items, and a measure of heavy drinking. Results of the 10-fold cross-validation gradient boosting machine analysis indicated excellent classification performance with an area under the curve of 0.882. Sensitivity was 0.85 and specificity was 0.77. Accuracy was 0.78 (95% confidence interval, 0.77-0.79). Feature importance analyses revealed that depressed mood and guilt were the most important predictors of suicidal ideation, followed by anhedonia, concentration difficulties, and psychomotor disturbance. Overall, depression symptoms were most predictive of suicidal ideation. Despite the limited clinical impact of the present findings, machine learning has potential to improve prediction of suicidal behavior, leveraging electronic health record data, to identify individuals at greatest risk, thereby facilitating intervention and optimization of long-term outcomes after traumatic brain injury.

Czech version of the Emotion Regulation Skills Questionnaire (ERSQ): Psychometric evaluation and network model in an adult clinical sample.

The Emotion Regulation Skills Questionnaire (ERSQ, 27 items) is an instrument designed to measure nine emotion regulation skills. This study examined the psychometric properties and longitudinal network structure of the Czech translation of the ERSQ in the clinical sample (primarily mood or neurotic disorders). Czech patients N = 427 completed the ERSQ weekly during treatment. The data were analyzed using confirmatory factor analysis (CFA) and network modeling. The CFA supported an eight-dimensional factor solution (with merged Acceptance and Tolerance subscales). The subscales' internal consistency ranged from ωh = .669 (Bodily sensations) to ωh = .859 (Acceptance/Tolerance). The factor structure was invariant across genders, two age groups, and seven measurement waves (invariance tested as an assumption for network analysis). The longitudinal network model indicated the existence of perceptive and modulatory clusters of emotional regulation skills and revealed the central role of Modification, Bodily sensations, and Readiness for confrontation in emotional regulation. The Czech translation of the ERSQ is a psychometrically sound instrument comparable to the original version. Despite the high internal consistency of the total score (ωtot = .910), the ERSQ is not recommended for use as a unidimensional instrument. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

What kinds of social networks protect older adults’ health during a pandemic? The tradeoff between preventing infection and promoting mental health

When the coronavirus emerged in early 2020, older adults were at heightened risk of contracting the virus, and of suffering mental health consequences from the pandemic and from the precautions designed to mitigate it. In this paper, we examine how social networks prior to the pandemic helped to shape health beliefs, behaviors, and outcomes among older adults during its onset, focusing on (1) perceived risk of COVID-19, (2) preventative health behaviors, and (3) mental health, including loneliness, perceived stress, depression, and anxiety. Drawing on the longitudinal Social Networks in Alzheimer Disease study, we find that networks high in bridging social capital predict greater perceived risk and more precautions taken, but worse mental health. In contrast, networks high in bonding social capital predict less perceived risk and fewer precautions taken, but better mental health. We discuss this apparent tradeoff between physical and mental health.

Altered synaptic plasticity of the longitudinal dentate gyrus network in noise-induced anxiety

SummaryAnxiety is characteristic comorbidity of noise-induced hearing loss (NIHL), which causes physiological changes within the dentate gyrus (DG), a subfield of the hippocampus that modulates anxiety. However, which DG circuit underlies hearing loss-induced anxiety remains unknown. We utilize an NIHL mouse model to investigate short- and long-term synaptic plasticity in DG networks. The recently discovered longitudinal DG-DG network is a collateral of DG neurons synaptically connected with neighboring DG neurons and displays robust synaptic efficacy and plasticity. Furthermore, animals with NIHL demonstrate increased anxiety-like behaviors similar to a response to chronic restraint stress. These behaviors are concurrent with enhanced synaptic responsiveness and suppressed short- and long-term synaptic plasticity in the longitudinal DG-DG network but not in the transverse DG-CA3 connection. These findings suggest that DG-related anxiety is typified by synaptic alteration in the longitudinal DG-DG network.