Monitoring Depression Trends on Twitter During the COVID-19 Pandemic: Observational Study

JMIR Publications Inc. - Tập 1 Số 1 - Trang e26769
Yipeng Zhang1, Hanjia Lyu1, Yubao Liu1, Xiyang Zhang2, Yu Wang1, Jiebo Luo1
1University of Rochester, Rochester, NY, United States
2University of Akron, Akron, OH, United States

Tóm tắt

Background The COVID-19 pandemic has affected people’s daily lives and has caused economic loss worldwide. Anecdotal evidence suggests that the pandemic has increased depression levels among the population. However, systematic studies of depression detection and monitoring during the pandemic are lacking. Objective This study aims to develop a method to create a large-scale depression user data set in an automatic fashion so that the method is scalable and can be adapted to future events; verify the effectiveness of transformer-based deep learning language models in identifying depression users from their everyday language; examine psychological text features’ importance when used in depression classification; and, finally, use the model for monitoring the fluctuation of depression levels of different groups as the disease propagates. Methods To study this subject, we designed an effective regular expression-based search method and created the largest English Twitter depression data set containing 2575 distinct identified users with depression and their past tweets. To examine the effect of depression on people’s Twitter language, we trained three transformer-based depression classification models on the data set, evaluated their performance with progressively increased training sizes, and compared the model’s tweet chunk-level and user-level performances. Furthermore, inspired by psychological studies, we created a fusion classifier that combines deep learning model scores with psychological text features and users’ demographic information, and investigated these features’ relations to depression signals. Finally, we demonstrated our model’s capability of monitoring both group-level and population-level depression trends by presenting two of its applications during the COVID-19 pandemic. Results Our fusion model demonstrated an accuracy of 78.9% on a test set containing 446 people, half of which were identified as having depression. Conscientiousness, neuroticism, appearance of first person pronouns, talking about biological processes such as eat and sleep, talking about power, and exhibiting sadness were shown to be important features in depression classification. Further, when used for monitoring the depression trend, our model showed that depressive users, in general, responded to the pandemic later than the control group based on their tweets (n=500). It was also shown that three US states—New York, California, and Florida—shared a similar depression trend as the whole US population (n=9050). When compared to New York and California, people in Florida demonstrated a substantially lower level of depression. Conclusions This study proposes an efficient method that can be used to analyze the depression level of different groups of people on Twitter. We hope this study can raise awareness among researchers and the public of COVID-19’s impact on people’s mental health. The noninvasive monitoring system can also be readily adapted to other big events besides COVID-19 and can be useful during future outbreaks.

Từ khóa


Tài liệu tham khảo

Coronavirus disease (COVID-19) pandemicWorld Health Organization2020-12-21https://www.who.int/emergencies/diseases/novel-coronavirus-2019

Education: from disruption to recoveryUNESCO202003042020-12-21https://en.unesco.org/covid19/educationresponse

Mental disordersWorld Health Organization2020-12-21https://www.who.int/news-room/fact-sheets/detail/mental-disorders

10.1192/bjp.172.1.35

10.1016/j.jad.2019.08.054

10.1186/s13104-017-3010-1

10.3386/w26983

10.1016/j.ijsu.2020.04.018

10.1016/j.jad.2020.08.001

10.3389/fpsyt.2020.00824

10.3389/fpsyt.2020.569981

10.1016/j.bbi.2020.04.069

10.3115/v1/w15-1202

10.3115/v1/w15-1203

10.1016/j.copsyc.2016.01.004

ErnalaSKLabetoulleTBaneFBirnbaumMLRizviAFKaneJMDe ChoudhuryMCharacterizing audience engagement and assessing its impact on social media disclosures of mental illnesses2018Twelfth International AAAI Conference on Web and Social MediaJune 25-28, 2018Palo Alto, CA

10.2196/17188

10.1038/nature07634

10.1109/bigdata50022.2020.9378114

Wu, W, arXiv.

Lyu, H, arXiv.

10.1016/j.ipm.2020.102486

10.1016/j.bbi.2020.04.028

Brandwatch2020-12-21https://www.brandwatch.com/

De ChoudhuryMGamonMCountsSHorvitzEPredicting depression via social media2013Seventh International AAAI Conference on Weblogs and Social MediaJuly 8-11, 2013Boston, MA

10.3115/v1/w14-3207

10.3115/v1/w15-1204

10.18653/v1/w17-3104

10.18653/v1/w18-0609

NadeemMIdentifying depression on TwitterarXiv201607252020-12-21http://arxiv.org/abs/1607.07384

10.18653/v1/w19-3003

10.18653/v1/w19-3005

10.3115/1614108

10.2196/17758

10.3390/ijerph15092028

10.3390/ijerph15050928

Ramirez-EsparzaNChungCKacewiczEPennebakerJWThe psychology of word use in depression forums in English and in SpanishProceedings of the 2008 International Conference on Weblogs and Social Media2008ICWSM '08March 30-April 2, 2008Seattle, WA

10.18653/v1/w18-5903

10.1109/access.2019.2909180

LiuYOttMGoyalNDuJJoshiMChenDLevyOLewisMZettlemoyerLStoyanovVRoBERTa: a robustly optimized BERT pretraining approachOpenReview201909252020-12-21https://openreview.net/forum?id=SyxS0T4tvS

Yang, Z, 2019, Adv Neural Inf Process Syst, 32, 5753

10.1145/2702123.2702280

Blei, DM, 2003, J Machine Learning Res, 3, 993

ZhouDLuoJSilenzioVZhouYHuJCurrierGKautzHTackling mental health by integrating unobtrusive multimodal sensingProceedings of the AAAI Conference on Artificial Intelligence2015AAAI-15January 25-30, 2015Austin, TX

10.2196/14199

10.2196/24012

10.24963/ijcai.2017/536

10.1145/2464464.2464480

10.1192/bjpo.bp.116.004960

10.1002/da.22376

Watson Personality InsightsIBM2020-12-21https://www.ibm.com/cloud/watson-personality-insights

10.2196/16023

HuttoCGilbertEVADER: a parsimonious rule-based model for sentiment analysis of social media textProceedings of the International AAAI Conference on Web and Social Media2014ICWSM-14June 1-4, 2014Ann Arbor, MI

10.2307/2137349

10.1016/s0165-0327(01)00420-7

10.3390/ijerph17051729

10.1503/jpn.150205

10.1145/3308558.3313684

10.1177/0261927X09351676

10.31234/osf.io/znj7h

10.1080/02699930441000030

10.1097/00006842-200107000-00001

Chen, L, arXiv.

10.1177/1077801207307801

10.1037//0021-843x.88.2.182

10.18653/v1/s17-2126

10.3115/v1/d14-1162

scikit-learn2020-12-21https://scikit-learn.org/stable/

10.1093/bioinformatics/btq134

10.1016/0165-0327(88)90043-2

10.21767/2471-805x.100008

10.1037/bul0000022

COVID data trackerCenters for Disease Control and Prevention202003282020-12-21https://covid.cdc.gov/covid-data-tracker

Ho, CS, 2020, Ann Acad Med Singap, 49, 155, 10.47102/annals-acadmedsg.202043

10.3233/THC-161261

10.1016/j.sleep.2020.08.020

10.2196/mental.7847

Thông tin
Thông tin xuất bản

JMIR Publications Inc.

Tập 1 Số 1

e26769

Thông tin tác giả