Intraindividual time-varying dynamic network of affects: linear autoregressive mixed-effects models for ecological momentary assessment

An interesting recent development in emotion research and clinical psychology is the discovery that affective states can be modeled as a network of temporally interacting moods or emotions. Additionally, external factors like stressors or treatments can influence the mood network by amplifying or dampening the activation of specific moods. Researchers have turned to multilevel autoregressive models to fit these affective networks using intensive longitudinal data gathered through ecological momentary assessment. Nonetheless, a more comprehensive examination of the performance of such models is warranted. In our study, we focus on simple directed intraindividual networks consisting of two interconnected mood nodes that mutually enhance or dampen each other. We also introduce a node representing external factors that affect both mood nodes unidirectionally. Importantly, we disregard the potential effects of a current mood/emotion on the perception of external factors. We then formalize the mathematical representation of such networks by exogenous linear autoregressive mixed-effects models. In this representation, the autoregressive coefficients signify the interactions between moods, while external factors are incorporated as exogenous covariates. We let the autoregressive and exogenous coefficients in the model have fixed and random components. Depending on the analysis, this leads to networks with variable structures over reasonable time units, such as days or weeks, which are captured by the variability of random effects. Furthermore, the fixed-effects parameters encapsulate a subject-specific network structure. Leveraging the well-established theoretical and computational foundation of linear mixed-effects models, we transform the autoregressive formulation to a classical one and utilize the existing methods and tools. To validate our approach, we perform simulations assuming our model as the true data-generating process. By manipulating a predefined set of parameters, we investigate the reliability and feasibility of our approach across varying numbers of observations, levels of noise intensity, compliance rates, and scalability to higher dimensions. Our findings underscore the challenges associated with estimating individualized parameters in the context of common longitudinal designs, where the required number of observations may often be unattainable. Moreover, our study highlights the sensitivity of autoregressive mixed-effect models to noise levels and the difficulty of scaling due to the substantial number of parameters.

Psychological Resilience Factors and Their Association With Weekly Stressor Reactivity During the COVID-19 Outbreak in Europe: Prospective Longitudinal Study

BACKGROUND

Cross-sectional relationships between psychosocial resilience factors (RFs) and resilience, operationalized as the outcome of low mental health reactivity to stressor exposure (low "stressor reactivity" [SR]), were reported during the first wave of the COVID-19 pandemic in 2020.

OBJECTIVE

Extending these findings, we here examined prospective relationships and weekly dynamics between the same RFs and SR in a longitudinal sample during the aftermath of the first wave in several European countries.

METHODS

Over 5 weeks of app-based assessments, participants reported weekly stressor exposure, mental health problems, RFs, and demographic data in 1 of 6 different languages. As (partly) preregistered, hypotheses were tested cross-sectionally at baseline (N=558), and longitudinally (n=200), using mixed effects models and mediation analyses.

RESULTS

RFs at baseline, including positive appraisal style (PAS), optimism (OPT), general self-efficacy (GSE), perceived good stress recovery (REC), and perceived social support (PSS), were negatively associated with SR scores, not only cross-sectionally (baseline SR scores; all P<.001) but also prospectively (average SR scores across subsequent weeks; positive appraisal (PA), P=.008; OPT, P<.001; GSE, P=.01; REC, P<.001; and PSS, P=.002). In both associations, PAS mediated the effects of PSS on SR (cross-sectionally: 95% CI -0.064 to -0.013; prospectively: 95% CI -0.074 to -0.0008). In the analyses of weekly RF-SR dynamics, the RFs PA of stressors generally and specifically related to the COVID-19 pandemic, and GSE were negatively associated with SR in a contemporaneous fashion (PA, P<.001; PAC,P=.03; and GSE, P<.001), but not in a lagged fashion (PA, P=.36; PAC, P=.52; and GSE, P=.06).

CONCLUSIONS

We identified psychological RFs that prospectively predict resilience and cofluctuate with weekly SR within individuals. These prospective results endorse that the previously reported RF-SR associations do not exclusively reflect mood congruency or other temporal bias effects. We further confirm the important role of PA in resilience.

Investigating two mobile just-in-time adaptive interventions to foster psychological resilience: research protocol of the DynaM-INT study

BACKGROUND

Stress-related disorders such as anxiety and depression are highly prevalent and cause a tremendous burden for affected individuals and society. In order to improve prevention strategies, knowledge regarding resilience mechanisms and ways to boost them is highly needed. In the Dynamic Modelling of Resilience - interventional multicenter study (DynaM-INT), we will conduct a large-scale feasibility and preliminary efficacy test for two mobile- and wearable-based just-in-time adaptive interventions (JITAIs), designed to target putative resilience mechanisms. Deep participant phenotyping at baseline serves to identify individual predictors for intervention success in terms of target engagement and stress resilience.

METHODS

DynaM-INT aims to recruit N = 250 healthy but vulnerable young adults in the transition phase between adolescence and adulthood (18-27 years) across five research sites (Berlin, Mainz, Nijmegen, Tel Aviv, and Warsaw). Participants are included if they report at least three negative burdensome past life events and show increased levels of internalizing symptoms while not being affected by any major mental disorder. Participants are characterized in a multimodal baseline phase, which includes neuropsychological tests, neuroimaging, bio-samples, sociodemographic and psychological questionnaires, a video-recorded interview, as well as ecological momentary assessments (EMA) and ecological physiological assessments (EPA). Subsequently, participants are randomly assigned to one of two ecological momentary interventions (EMIs), targeting either positive cognitive reappraisal or reward sensitivity. During the following intervention phase, participants' stress responses are tracked using EMA and EPA, and JITAIs are triggered if an individually calibrated stress threshold is crossed. In a three-month-long follow-up phase, parts of the baseline characterization phase are repeated. Throughout the entire study, stressor exposure and mental health are regularly monitored to calculate stressor reactivity as a proxy for outcome resilience. The online monitoring questionnaires and the repetition of the baseline questionnaires also serve to assess target engagement.

DISCUSSION

The DynaM-INT study intends to advance the field of resilience research by feasibility-testing two new mechanistically targeted JITAIs that aim at increasing individual stress resilience and identifying predictors for successful intervention response. Determining these predictors is an important step toward future randomized controlled trials to establish the efficacy of these interventions.

Coping with COVID: risk and resilience factors for mental health in a German representative panel study

BACKGROUND

The coronavirus disease 2019 (COVID-19) pandemic might affect mental health. Data from population-representative panel surveys with multiple waves including pre-COVID data investigating risk and protective factors are still rare.

METHODS

In a stratified random sample of the German household population (n = 6684), we conducted survey-weighted multiple linear regressions to determine the association of various psychological risk and protective factors assessed between 2015 and 2020 with changes in psychological distress [(PD; measured via Patient Health Questionnaire for Depression and Anxiety (PHQ-4)] from pre-pandemic (average of 2016 and 2019) to peri-pandemic (both 2020 and 2021) time points. Control analyses on PD change between two pre-pandemic time points (2016 and 2019) were conducted. Regularized regressions were computed to inform on which factors were statistically most influential in the multicollinear setting.

RESULTS

PHQ-4 scores in 2020 (M = 2.45) and 2021 (M = 2.21) were elevated compared to 2019 (M = 1.79). Several risk factors (catastrophizing, neuroticism, and asking for instrumental support) and protective factors (perceived stress recovery, positive reappraisal, and optimism) were identified for the peri-pandemic outcomes. Control analyses revealed that in pre-pandemic times, neuroticism and optimism were predominantly related to PD changes. Regularized regression mostly confirmed the results and highlighted perceived stress recovery as most consistent influential protective factor across peri-pandemic outcomes.

CONCLUSIONS

We identified several psychological risk and protective factors related to PD outcomes during the COVID-19 pandemic. A comparison of pre-pandemic data stresses the relevance of longitudinal assessments to potentially reconcile contradictory findings. Implications and suggestions for targeted prevention and intervention programs during highly stressful times such as pandemics are discussed.

Deep learning and differential equations for modeling changes in individual-level latent dynamics between observation periods

When modeling longitudinal biomedical data, often dimensionality reduction as well as dynamic modeling in the resulting latent representation is needed. This can be achieved by artificial neural networks for dimension reduction and differential equations for dynamic modeling of individual-level trajectories. However, such approaches so far assume that parameters of individual-level dynamics are constant throughout the observation period. Motivated by an application from psychological resilience research, we propose an extension where different sets of differential equation parameters are allowed for observation subperiods. Still, estimation for intra-individual subperiods is coupled for being able to fit the model also with a relatively small dataset. We subsequently derive prediction targets from individual dynamic models of resilience in the application. These serve as outcomes for predicting resilience from characteristics of individuals, measured at baseline and a follow-up time point, and selecting a small set of important predictors. Our approach is seen to successfully identify individual-level parameters of dynamic models that allow to stably select predictors, that is, resilience factors. Furthermore, we can identify those characteristics of individuals that are the most promising for updates at follow-up, which might inform future study design. This underlines the usefulness of our proposed deep dynamic modeling approach with changes in parameters between observation subperiods.

Effects of general and corona-specific stressors on mental burden during the SARS-CoV-2 pandemic in Germany

The SARS-CoV-2 pandemic turned out to be a serious threat to mental and physical health. However, the relative contribution of corona-specific (DH_s) and general stressors (DH_g) on mental burden, and specific protective and risk factors for mental health are still not well understood. In a representative sample (N = 3,055) of the German adult population, mental health, potential risk, and protective factors as well as DH_s and DH_g exposure were assessed online during the SARS-CoV-2 pandemic (June and July 2020). The impact of these factors on mental health was analyzed using descriptive statistics, data visualizations, multiple regressions, and moderation analyses. The most burdensome DH_g were financial and sleeping problems, respectively, and DH_s corona-media reports and exclusion from recreational activities/important social events. 31 and 24% of total mental health was explained by DH_g and DH_s, respectively. Both predictors combined explained 36%, resulting in an increase in variance due to DH_s of only 5% (R² adjusted). Being female, older and a lower educational level were identified as general risk factors, somatic diseases as a corona-specific risk factor, and self-efficacy and locus of control (LOC) proved to be corona-specific protective factors. Further analyses showed that older age and being diagnosed with a somatic illness attenuated the positive influence of LOC, self-efficacy, and social support on resilience. Although the data showed that after the first easing restrictions, the stressor load was comparable to pre-pandemic data (with DH_s not making a significant contribution), different risk and protective factors could be identified for general and corona-specific stressors. In line with observations from network analysis from other groups, the positive impact of resilience factors was especially diminished in the most vulnerable groups (elderly and somatically ill). This highlights the need to especially target these vulnerable groups to foster their resilience in upcoming waves of the corona pandemic.

Individualizing deep dynamic models for psychological resilience data

Deep learning approaches can uncover complex patterns in data. In particular, variational autoencoders achieve this by a non-linear mapping of data into a low-dimensional latent space. Motivated by an application to psychological resilience in the Mainz Resilience Project, which features intermittent longitudinal measurements of stressors and mental health, we propose an approach for individualized, dynamic modeling in this latent space. Specifically, we utilize ordinary differential equations (ODEs) and develop a novel technique for obtaining person-specific ODE parameters even in settings with a rather small number of individuals and observations, incomplete data, and a differing number of observations per individual. This technique allows us to subsequently investigate individual reactions to stimuli, such as the mental health impact of stressors. A potentially large number of baseline characteristics can then be linked to this individual response by regularized regression, e.g., for identifying resilience factors. Thus, our new method provides a way of connecting different kinds of complex longitudinal and baseline measures via individualized, dynamic models. The promising results obtained in the exemplary resilience application indicate that our proposal for dynamic deep learning might also be more generally useful for other application domains.

Carotid Geometry and Wall Shear Stress Independently Predict Increased Wall Thickness-A Longitudinal 3D MRI Study in High-Risk Patients

Introduction: Carotid geometry and wall shear stress (WSS) have been proposed as independent risk factors for the progression of carotid atherosclerosis, but this has not yet been demonstrated in larger longitudinal studies. Therefore, we investigated the impact of these biomarkers on carotid wall thickness in patients with high cardiovascular risk.

Methods: Ninety-seven consecutive patients with hypertension, at least one additional cardiovascular risk factor and internal carotid artery (ICA) plaques (wall thickness ≥ 1.5 mm and degree of stenosis ≤ 50%) were prospectively included. They underwent high-resolution 3D multi-contrast and 4D flow MRI at 3 Tesla both at baseline and follow-up. Geometry (ICA/common carotid artery (CCA)-diameter ratio, bifurcation angle, tortuosity and wall thickness) and hemodynamics [WSS, oscillatory shear index (OSI)] of both carotid bifurcations were measured at baseline. Their predictive value for changes of wall thickness 12 months later was calculated using linear regression analysis for the entire study cohort (group 1, 97 patients) and after excluding patients with ICA stenosis ≥10% to rule out relevant inward remodeling (group 2, 61 patients).

Results: In group 1, only tortuosity at baseline was independently associated with carotid wall thickness at follow-up (regression coefficient = −0.52, p < 0.001). However, after excluding patients with ICA stenosis ≥10% in group 2, both ICA/CCA-ratio (0.49, p < 0.001), bifurcation angle (0.04, p = 0.001), tortuosity (−0.30, p = 0.040), and WSS (−0.03, p = 0.010) at baseline were independently associated with changes of carotid wall thickness at follow-up.

Conclusions: A large ICA bulb and bifurcation angle and low WSS seem to be independent risk factors for the progression of carotid atherosclerosis in the absence of ICA stenosis. By contrast, a high carotid tortuosity seems to be protective both in patients without and with ICA stenosis. These biomarkers may be helpful for the identification of patients who are at particular risk of wall thickness progression and who may benefit from intensified monitoring and treatment.

The Frequent Stressor and Mental Health Monitoring-Paradigm: A Proposal for the Operationalization and Measurement of Resilience and the Identification of Resilience Processes in Longitudinal Observational Studies

Resilience has been defined as the maintenance or quick recovery of mental health during and after times of adversity. How to operationalize resilience and to determine the factors and processes that lead to good long-term mental health outcomes in stressor-exposed individuals is a matter of ongoing debate and of critical importance for the advancement of the field. One of the biggest challenges for implementing an outcome-based definition of resilience in longitudinal observational study designs lies in the fact that real-life adversity is usually unpredictable and that its substantial qualitative as well as temporal variability between subjects often precludes defining circumscribed time windows of inter-individually comparable stressor exposure relative to which the maintenance or recovery of mental health can be determined. To address this pertinent issue, we propose to frequently and regularly monitor stressor exposure (E) and mental health problems (P) throughout a study's observation period [Frequent Stressor and Mental Health Monitoring (FRESHMO)-paradigm]. On this basis, a subject's deviation at any single monitoring time point from the study sample's normative E-P relationship (the regression residual) can be used to calculate that subject's current mental health reactivity to stressor exposure ("stressor reactivity," SR). The SR score takes into account the individual extent of experienced adversity and is comparable between and within subjects. Individual SR time courses across monitoring time points reflect intra-individual temporal variability in SR, where periods of under-reactivity (negative SR score) are associated with accumulation of fewer mental health problems than is normal for the sample. If FRESHMO is accompanied by regular measurement of potential resilience factors, temporal changes in resilience factors can be used to predict SR time courses. An increase in a resilience factor measurement explaining a lagged decrease in SR can then be considered to index a process of adaptation to stressor exposure that promotes a resilient outcome (an allostatic resilience process). This design principle allows resilience research to move beyond merely determining baseline predictors of resilience outcomes, which cannot inform about how individuals successfully adjust and adapt when confronted with adversity. Hence, FRESHMO plus regular resilience factor monitoring incorporates a dynamic-systems perspective into resilience research.

Psycho-social factors associated with mental resilience in the Corona lockdown

The SARS-CoV-2 pandemic is not only a threat to physical health but is also having severe impacts on mental health. Although increases in stress-related symptomatology and other adverse psycho-social outcomes, as well as their most important risk factors have been described, hardly anything is known about potential protective factors. Resilience refers to the maintenance of mental health despite adversity. To gain mechanistic insights about the relationship between described psycho-social resilience factors and resilience specifically in the current crisis, we assessed resilience factors, exposure to Corona crisis-specific and general stressors, as well as internalizing symptoms in a cross-sectional online survey conducted in 24 languages during the most intense phase of the lockdown in Europe (22 March to 19 April) in a convenience sample of N = 15,970 adults. Resilience, as an outcome, was conceptualized as good mental health despite stressor exposure and measured as the inverse residual between actual and predicted symptom total score. Preregistered hypotheses (osf.io/r6btn) were tested with multiple regression models and mediation analyses. Results confirmed our primary hypothesis that positive appraisal style (PAS) is positively associated with resilience (p < 0.0001). The resilience factor PAS also partly mediated the positive association between perceived social support and resilience, and its association with resilience was in turn partly mediated by the ability to easily recover from stress (both p < 0.0001). In comparison with other resilience factors, good stress response recovery and positive appraisal specifically of the consequences of the Corona crisis were the strongest factors. Preregistered exploratory subgroup analyses (osf.io/thka9) showed that all tested resilience factors generalize across major socio-demographic categories. This research identifies modifiable protective factors that can be targeted by public mental health efforts in this and in future pandemics.

[123I]FP-CIT SPECT in Clinically Uncertain Parkinsonism Predicts Survival: A Data-Driven Analysis

BACKGROUND

Dopamine transporter SPECT is an established method to investigate nigrostriatal integrity in case of clinically uncertain parkinsonism.

OBJECTIVE

The present study explores whether a data-driven analysis of [123I]FP-CIT SPECT is able to stratify patients according to mortality after SPECT.

METHODS

Patients from our clinical registry were included if they had received [123I]FP-CIT SPECT between 10/2008 and 06/2016 for diagnosis of parkinsonism and if their vital status could be determined in 07/2017. Specific binding ratios (SBR) of the whole striatum, its asymmetry (asymmetry index, AI; absolute value), and the rostrocaudal gradient of striatal binding (C/pP: caudate SBR divided by posterior putamen SBR) were used as input for hierarchical clustering of patients. We tested differences in survival between these groups (adjusted for age) with a Cox proportional hazards model.

RESULTS

Data from 518 patients were analyzed. Median follow-up duration was 3.3 years [95% C.I. 3.1 to 3.7]. Three subgroups identified by hierarchical clustering were characterized by relatively low striatal SBR, high AI, and low C/pP (group 1), low striatal SBR, high AI, and high C/pP (group 2), and high striatal SBR, low AI, and low C/pP (group 3). Mortality was significantly higher in group 1 compared to each of the other two groups (p = 0.029 and p = 0.003, respectively).

CONCLUSION

Data-driven analysis of [123I]FP-CIT SPECT identified a subgroup of patients with significantly increased mortality during follow-up. This suggests that [123I]-FP-CIT SPECT might not only serve as a diagnostic tool to verify nigrostriatal degeneration but also provide valuable prognostic information.