Using language in social media posts to study the network dynamics of depression longitudinally

Digital phenotyping of depression during pregnancy using self-report data

BACKGROUND

Depression is a common pregnancy complication yet is often under-detected and, subsequently, undertreated. Data collected through mobile health tools may be used to support the identification of depression symptoms in pregnancy.

METHODS

An observational cohort study of 2062 pregnancies collected self-reports of patient history, mood, pregnancy-specific symptoms, and written language using a prenatal support app. These app inputs were used to model depression risk in subsequent 30- and 60-day periods throughout pregnancy. A selective inference lasso modeling approach examined the individual and additive value of each type of patient-reported app input.

RESULTS

Depression models ranged in predictive power (AUC value of 0.64-0.83), depending on the type of inputs. The most predictive model included personal history, daily mood, and acute pregnancy-related symptoms (e.g., severe vomiting, cramping). Across models, daily mood was the strongest indicator of depression symptoms in the following month. Models that retained natural language inputs typically improved predictive accuracy and offered insight into the lived context associated with experiencing depression.

LIMITATIONS

Our findings are not generalizable beyond a digitally literate patient population that is self-motivated to report data during pregnancy.

CONCLUSIONS

Simple patient reported data, including sparse language, shared directly via digital tools may support earlier depression symptom identification and a more nuanced understanding of depression context.

"How" web searches change under stress

To adjust to stressful environments, people seek information. Here, we show that in response to stressful public and private events the high-level features of information people seek online alter, reflecting their motives for seeking knowledge. We first show that when people want information to guide action they selectively ask "How" questions. Next, we reveal that "How" searches submitted to Google increased dramatically during the pandemic (controlling for search volume). Strikingly, the proportion of these searches predicted weekly self-reported stress of ~ 17K individuals. To rule out third factors we manipulate stress and find that "How" searches increase in response to stressful, personal, events. The findings suggest that under stress people ask questions to guide action, and mental state is reflected in features that tap into why people seek information rather than the topics they search for. Tracking such features may provide clues regrading population stress levels.

The 'L-factor': Language as a transdiagnostic dimension in psychopathology

Thoughts and moods constituting our mental life incessantly change. When the steady flow of this dynamics diverges in clinical directions, the possible pathways involved are captured through discrete diagnostic labels. Yet a single vulnerable neurocognitive system may be causally involved in psychopathological deviations transdiagnostically. We argue that language viewed as integrating cortical functions is the best current candidate, whose forms of breakdown along its different dimensions are then manifest as symptoms - from prosodic abnormalities and rumination in depression to distortions of speech perception in verbal hallucinations, distortions of meaning and content in delusions, or disorganized speech in formal thought disorder. Spontaneous connected speech provides continuous objective readouts generating a highly accessible bio-behavioral marker with the potential of revolutionizing neuropsychological measurement. This argument turns language into a transdiagnostic 'L-factor' providing an analytical and mechanistic substrate for previously proposed latent general factors of psychopathology ('p-factor') and cognitive functioning ('c-factor'). Together with immense practical opportunities afforded by rapidly advancing natural language processing (NLP) technologies and abundantly available data, this suggests a new era of translational clinical psychiatry, in which both psychopathology and language may be rethought together.

Understanding Social Media Information Sharing in Individuals with Depression: Insights from the Elaboration Likelihood Model and Schema Activation Theory

Purpose

How individuals engage with social media can significantly impact their psychological well-being. This study examines the impact of social media interactions on mental health, grounded in the frameworks of the Elaboration Likelihood Model and Schema Activation Theory. It aims to uncover behavioral differences in information sharing between the general population and individuals with depression, while also elucidating the psychological mechanisms underlying these disparities.

Methods

A pre-experiment (N=30) and three experiments (Experiment 1a N=200, Experiment 1b N=180, Experiment 2 N=128) were executed online. These experiments investigated the joint effects of information quality, content valence, self-referential processing, and depression level on the intention to share information. The research design incorporated within-subject and between-subject methods, utilizing SPSS and SPSS Process to conduct independent sample t-tests, two-factor ANOVA analyses, mediation analyses, and moderated mediation analyses to test our hypotheses.

Results

Information quality and content valence significantly influence sharing intention. In scenarios involving low-quality information, individuals with depression are more inclined to share negative emotional content compared to the general population, and this tendency intensifies with the severity of depression. Moreover, self-referential processing acts as a mediator between emotional content and intention to share, yet this mediation effect weakens as the severity of depression rises.

Conclusion

Our study highlights the importance of promoting viewpoint diversity and breaking the echo chamber effect in social media to improve the mental health of individuals with depression. To achieve this goal, tailoring emotional content on social media could be a practical starting point for practice.

Linguistic markers for major depressive disorder: a cross-sectional study using an automated procedure

Introduction

The identification of language markers, referring to both form and content, for common mental health disorders such as major depressive disorder (MDD), can facilitate the development of innovative tools for early recognition and prevention. However, studies in this direction are only at the beginning and are difficult to implement due to linguistic variability and the influence of cultural contexts.

Aim

This study aims to identify language markers specific to MDD through an automated analysis process based on RO-2015 LIWC (Linguistic Inquiry and Word Count).

Materials and methods

A sample of 62 medicated patients with MDD and a sample of 43 controls were assessed. Each participant provided language samples that described something that was pleasant for them.

Assessment tools

(1) Screening tests for MDD (MADRS and DASS-21); (2) Ro-LIWC2015 - Linguistic Inquiry and Word Count - a computerized text analysis software, validated for Romanian Language, that analyzes morphology, syntax and semantics of word use.

Results

Depressive patients use different approaches in sentence structure, and communicate in short sentences. This requires multiple use of the punctuation mark period, which implicitly requires directive communication, limited in exchange of ideas. Also, participants from the sample with depression mostly use impersonal pronouns, first person pronoun in plural form - not singular, a limited number of prepositions and an increased number of conjunctions, auxiliary verbs, negations, verbs in the past tense, and much less in the present tense, increased use of words expressing negative affects, anxiety, with limited use of words indicating positive affects. The favorite topics of interest of patients with depression are leisure, time and money.

Conclusion

Depressive patients use a significantly different language pattern than people without mood or behavioral disorders, both in form and content. These differences are sometimes associated with years of education and sex, and might also be explained by cultural differences.

Language use on Twitter reflects social structure and social disparities

Large-scale mental health assessments increasingly rely upon user-contributed social media data. It is widely known that mental health and well-being are affected by minority group membership and social disparity. But do these factors manifest in the language use of social media users? We elucidate this question using spatial lag regressions. We examined the county-level (N = 1069) associations of lexical indicators linked to well-being and mental health, notably depression (e.g., first-person singular pronouns, negative emotions) with markers of social disparity (e.g., the Area Deprivation Index-3) and ethnicity, using a sample of approximately 30 million content-coded tweets (U.S. county-level aggregation). Results confirmed most expected associations: County-level lexical indicators of depression are positively linked with county-level area disparity (e.g., economic hardship and inequity) and percentage of ethnic minority groups. Predictive validity checks show that lexical indicators are related to future health and mental health outcomes. Lexical indicators of depression and adjustment coded from tweets aggregated at the county level could play a crucial role in prioritizing public health campaigns, particularly in socially deprived counties.

Violent behavior and the network properties of psychopathological symptoms and real-life functioning in patients with schizophrenia

Objective

To assess the interplay among psychopathological symptoms and real-life functioning, and to further detect their influence with violent behavior in patient with schizophrenia.

Methods

A sample of 1,664 patients with post-violence assessments and their propensity score-matched controls without violence from a disease registration report system of community mental health service in Guangdong, China, were studied by network analysis. Ising-Model was used to estimate networks of psychopathological symptoms and real-life functioning. Then, we tested whether network properties indicated the patterns of interaction were different between cases and controls, and calculated centrality indices of each node to identify the central nodes. Sensitivity analysis was conducted to examine the difference of interaction patterns between pre-violence and post-violence assessments in violence cases.

Results

Some nodes in the same domain were highly positive interrelations, while psychopathological symptoms were negatively related to real-life functioning in all networks. Many symptom-symptom connections and symptom-functioning connections were disconnected after the violence. The network density decreased from 23.53% to 12.42% without statistical significance (p = 0.338). The network structure, the global network strength, and the global clustering coefficient decreased significantly after the violence (p < 0.001, p = 0.019, and p = 0.045, respectively). Real-life functioning had a higher node strength. The strength of sleeping, lack of spontaneity and flow of conversation, and preoccupation were decreased in post-violence network of patients.

Conclusion

The decreasing connectivity may indicate an increased risk of violence and early warning for detecting violence. Interventions and improving health state based on nodes with high strength might prevent violence in schizophrenia patients.

Using natural language from a smartphone pregnancy app to identify maternal depression

Depression is highly prevalent in pregnancy, yet it often goes undiagnosed and untreated. Language can be an indicator of psychological well-being. This longitudinal, observational cohort study of 1,274 pregnancies examined written language shared in a prenatal smartphone app. Natural language feature of text entered in the app (e.g. in a journaling feature) throughout the course of participants' pregnancies were used to model subsequent depression symptoms. Language features were predictive of incident depression symptoms in a 30-day window (AUROC = 0.72) and offer insights into topics most salient in the writing of individuals experiencing those symptoms. When natural language inputs were combined with self-reported current mood, a stronger predictive model was produced (AUROC = 0.84). Pregnancy apps are a promising way to illuminate experiences contributing to depression symptoms. Even sparse language and simple patient-reports collected directly from these tools may support earlier, more nuanced depression symptom identification.

Functional neuroimaging in psychiatry and the case for failing better

Psychiatric disorders encompass complex aberrations of cognition and affect and are among the most debilitating and poorly understood of any medical condition. Current treatments rely primarily on interventions that target brain function (drugs) or learning processes (psychotherapy). A mechanistic understanding of how these interventions mediate their therapeutic effects remains elusive. From the early 1990s, non-invasive functional neuroimaging, coupled with parallel developments in the cognitive neurosciences, seemed to signal a new era of neurobiologically grounded diagnosis and treatment in psychiatry. Yet, despite three decades of intense neuroimaging research, we still lack a neurobiological account for any psychiatric condition. Likewise, functional neuroimaging plays no role in clinical decision making. Here, we offer a critical commentary on this impasse and suggest how the field might fare better and deliver impactful neurobiological insights.

Using Remotely Supervised At-Home TES for Enhancing Mental Resilience

We are in the midst of a mental health crisis with major depressive disorder being the most prevalent among mental health disorders and up to 30% of patients not responding to first-line treatments. Noninvasive Brain Stimulation (NIBS) techniques have proven to be effective in treating depression. However, there is a fundamental problem of scale. Currently, any type of NIBS treatment requires patients to repeatedly visit a clinic to receive brain stimulation by trained personnel. This is an often-insurmountable barrier to both patients and healthcare providers in terms of time and cost. In this perspective, we assess to what extent Transcranial Electrical Stimulation (TES) might be administered with remote supervision in order to address this scaling problem and enable neuroenhancement of mental resilience at home. Social, ethical, and technical challenges relating to hardware- and software-based solutions are discussed alongside the risks of stimulation under- or over-use. Solutions to provide users with a safe and transparent ongoing assessment of aptitude, tolerability, compliance, and/or misuse are proposed, including standardized training, eligibility screening, as well as compliance and side effects monitoring. Looking into the future, such neuroenhancement could be linked to prevention systems which combine home-use TES with digital sensor and mental monitoring technology to index decline in mental wellbeing and avoid relapse. Despite the described social, ethical legal, and technical challenges, the combination of remotely supervised, at-home TES setups with dedicated artificial intelligence systems could be a powerful weapon to combat the mental health crisis by bringing personalized medicine into people’s homes.

Machine learning of language use on Twitter reveals weak and non-specific predictions

Depressed individuals use language differently than healthy controls and it has been proposed that social media posts can be used to identify depression. Much of the evidence behind this claim relies on indirect measures of mental health and few studies have tested if these language features are specific to depression versus other aspects of mental health. We analysed the Tweets of 1006 participants who completed questionnaires assessing symptoms of depression and 8 other mental health conditions. Daily Tweets were subjected to textual analysis and the resulting linguistic features were used to train an Elastic Net model on depression severity, using nested cross-validation. We then tested performance in a held-out test set (30%), comparing predictions of depression versus 8 other aspects of mental health. The depression trained model had modest out-of-sample predictive performance, explaining 2.5% of variance in depression symptoms (R² = 0.025, r = 0.16). The performance of this model was as-good or superior when used to identify other aspects of mental health: schizotypy, social anxiety, eating disorders, generalised anxiety, above chance for obsessive-compulsive disorder, apathy, but not significant for alcohol abuse or impulsivity. Machine learning analysis of social media data, when trained on well-validated clinical instruments, could not make meaningful individualised predictions regarding users’ mental health. Furthermore, language use associated with depression was non-specific, having similar performance in predicting other mental health problems.

Possible Futures for Network Psychometrics

This commentary reflects on the articles included in the Psychometrika Special Issue on Network Psychometrics in Action. The contributions to the special issue are related to several possible future paths for research in this area. These include the development of models to analyze and represent interventions, improvement in exploratory and inferential techniques in network psychometrics, the articulation of psychometric theories in addition to psychometric models, and extensions of network modeling to novel data sources. Finally, network psychometrics is part of a larger movement in psychology that revolves around the analysis of human beings as complex systems, and it is timely that psychometricians start extending their rich modeling tradition to improve and extend the analysis of systems in psychology.

Association between depressive symptoms and diagnosis of diabetes and its complications: A network analysis in electronic health records

OBJECTIVES

Diabetes and its complications are commonly associated with depressive symptoms, and few studies have investigated the diagnosis effect of depressive symptoms in patients with diabetes. The present study used a network-based approach to explore the association between depressive symptoms, which are annotated from electronic health record (EHR) notes by a deep learning model, and the diagnosis of type 2 diabetes mellitus (T2DM) and its complications.

METHODS

In this study, we used anonymous admission notes of 52,139 inpatients diagnosed with T2DM at the first affiliated hospital of Nanjing Medical University from 2008 to 2016 as input for a symptom annotation model named T5-depression based on transformer architecture which helps to annotate depressive symptoms from present illness. We measured the performance of the model by using the F1 score and the area under the receiver operating characteristic curve (AUROC). We constructed networks of depressive symptoms to examine the connectivity of these networks in patients diagnosed with T2DM, including those with certain complications.

RESULTS

The T5-depression model achieved the best performance with an F1-score of 91.71 and an AUROC of 96.25 compared with the benchmark models. The connectivity of depressive symptoms in patients diagnosed with T2DM (p = 0.025) and hypertension (p = 0.013) showed a statistically significant increase 2 years after the diagnosis, which is consistent with the number of patients diagnosed with depression.

CONCLUSION

The T5-depression model proposed in this study can effectively annotate depressive symptoms in EHR notes. The connectivity of annotated depressive symptoms is associated with the diagnosis of T2DM and hypertension. The changes in the network of depressive symptoms generated by the T5-depression model could be used as an indicator for screening depression.