Table 1.

Summary of articles estimating mental health burden from social media during a disaster.

Disaster category and reference			Disaster type		Disaster year		Disaster location		Mental health issue		Social media platform		Number of posts (number of users)		Analysis
Natural disaster
	Gruebner et al [16]	Meteorological		2012		United States		Affective response		Twitter		344,957 (NRa)		GISb analysis
	Gruebner et al [17]	Meteorological		2012		United States		Affective response		Twitter		1,018,140 (NR)		GIS analysis
	Karmegam and Mappillairaju [18]	Hydrological		2015		India		Affective response		Twitter		5696 (NR)		Mixed effect model, spatial regression model, thematic analysis
	Li et al [19]	Geophysical, biological		2009-2011		Japan, Haiti		Affective response		Twitter		50,000 (NR)		t test
	Shekhar and Setty [20]	Geophysical, climatological, and hydrological		2015		Global		Affective response		Twitter		60,519 (NR)		Text mining; k-means clustering
	Vo and Collier [21]	Geophysical		2011		Japan		Affective response		Twitter		70,725 (NR)		Naive Bayes, support vector machine, MaxEnt, J48, multinomial naive Bayes; Pearson correlation
Human-made disaster
	Doré et al [22]	Active shooter		2012-2013		United States		Affective response		Twitter		43,548 (NR)		Negative binomial regression
	Glasgow et al [23]	Active shooter		2012-2013		United States		Grief		Twitter		460,000 (NR)		Multinomial naive Bayes; support vector machine
	Gruebner et al [5]	Terrorist attack		2015		France		Affective response		Twitter		22,534 (NR)		GIS analysis
	Jones et al [24]	Active shooter		2014-2015		United States		Affective response		Twitter		325,736 (6314)		Piecewise regression
	Jones et al [25]	Terrorist attack		2015		United States		Affective response		Twitter		1,160,000 (25,894)		Time series, topic analysis
	Khalid et al [26]	Terrorist attack		NR		NR		Trauma		Unspecified blogs and discussion boards		17 (NR)		Semantic mapping and knowledge pathways
	Lin et al [27]	Terrorist attack		2015-2016		France, Belgium		Affective response		Twitter		18 Million (NR)		Multivariate regression analysis, survival analysis
	Sadasivuni and Zhang [28]	Terrorist attack		2019		Sri Lanka		Depression		Twitter		51,462 (NR)		Gradient-based trend analysis methods, correlation, learning quotient, text mining
	Saha and De Choudhury [29]	Active shooter		2012-2016		United States		Stress		Reddit		113,337 (NR)		Support vector machine classifier of stress, time-series analysis
	Woo et al [30]	Accident		2011-2014		Korea		Suicide		Twitter		NR		Time-series analysis
Epidemic or pandemic
	Da and Yang [31]	Epidemic or pandemic		2020		China		Affective response		Sina Weibo		340,456 (NR)		Linear regression
	Gupta and Agrawal [32]	Epidemic or pandemic		2020		India		Anxiety, depression, panic attacks, stress, suicide attempts		Twitter, Facebook, WhatsApp, and blogs		NR		Thematic analysis
	Hung et al [33]	Epidemic or pandemic		2020		United States		Psychological stress		Twitter		1,001,380 (334,438)		Latent Dirichlet allocation
	Koh and Liew [34]	Epidemic or pandemic		2020		Global		Loneliness		Twitter		NR (4492)		Hierarchical clustering
	Kumar and Chinnalagu [35]	Epidemic or pandemic		2020		NR		Affective response		Twitter, Facebook, YouTube, and blogs		80,689 (NR)		Sentiment analysis bidirectional long short-term memory
	Lee et al [36]	Epidemic or pandemic		2020		Japan, Korea		Affective response		Twitter		4,951,289 (NR)		Trend analysis
	Li et al [37]	Epidemic or pandemic		2020		China		Anxiety, depression, indignation, and Oxford happiness		Sina Weibo		NR (17,865)		t test
	Low et al [38]	Epidemic or pandemic		2018-2020		Global		Eating disorder, addiction, alcoholism, ADHDc, anxiety, autism, bipolar disorder, BPDd, depression, health anxiety, loneliness, PTSDe, schizophrenia, social anxiety, suicide, broad mental health, COVID-19 support		Reddit		NR (826,961)		Support vector machine, tree ensemble, stochastic gradient descent, linear regression, spectral clustering, latent Dirichlet allocation
	Mathur et al [39]	Epidemic or pandemic		2019-2020		Global		Affective response		Twitter		30,000 (NR)		Sentiment analysis
	Oyebode et al [40]	Epidemic or pandemic		2020		Global		General mental health concerns		Twitter, YouTube, Facebook, Archinect, LiveScience, and PushSquare		8,021,341 (NR)		Thematic analysis
	Pellert et al [41]	Epidemic or pandemic		2020		Austria		Affective response		Twitter and unspecified chat platform for students		2,159,422 (594,500)		Trend analysis
	Pran et al [42]	Epidemic or pandemic		2020		Bangladesh		Affective response		Facebook		1120 (NR)		Convolutional neural network and long short-term memory
	Sadasivuni and Zhang [43]	Epidemic or pandemic		2020		Global		Depression		Twitter		318,847 (NR)		Autoregressive integrated moving average model
	Song et al [44]	Epidemic or pandemic		2015		South Korea		Anxiety		Twitter, Unspecified blogs and discussion boards		8,671,695 (NR)		Multilevel analysis, association analysis
	Xu et al [45]	Epidemic or pandemic		2019-2020		China		Affective response		Sina Weibo		10,159 (8703)		Content analysis, regression
	Xue et al [46]	Epidemic or pandemic		2020		Global		Affective response		Twitter		4,196,020 (NR)		Latent Dirichlet allocation, sentiment analysis
	Zhang et al [47]	Epidemic or pandemic		2020		United States		Depression, anxiety		YouTube		294,294 (49)		Regression, correlation, feature vector

^aNR: not reported.

^bGIS: geographic information system.

^cADHD: attention-deficit/hyperactivity disorder.

^dBPD: borderline personality disorder.

^ePTSD: posttraumatic stress disorder.