Identifying schizophrenia stigma on Twitter: a proof of principle model using service user supervised machine learning

Jilka, Sagar; Odoi, Clarissa; Bilsen, Janet van; Morris, Daniel; Erturk, Sinan; Cummins, Nicholas; Cella, Matteo; Wykes, Til

doi:10.1038/s41537-021-00197-6

Cited by 17 publications

(8 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This serves as further testing for our ML model on a set of tweets independent from those used for training, which is commonly used in validating ML models [45,46]. Our past research has shown that this additional step can help clarify a small difference in accuracies and demonstrate a clearer difference in performance between top-performing models, confirming that this is an important step [31]. We randomly selected 150 tweets from our second sample of 96,356 tweets and split them into 3 batches of 50.…”

Section: Blind Validationmentioning

confidence: 88%

See 1 more Smart Citation

Codeveloping and Evaluating a Campaign to Reduce Dementia Misconceptions on Twitter: Machine Learning Study

Erturk¹,

Hudson²,

Jansli³

et al. 2022

JMIR Infodemiology

Self Cite

View full text Add to dashboard Cite

Background Dementia misconceptions on Twitter can have detrimental or harmful effects. Machine learning (ML) models codeveloped with carers provide a method to identify these and help in evaluating awareness campaigns. Objective This study aimed to develop an ML model to distinguish between misconceptions and neutral tweets and to develop, deploy, and evaluate an awareness campaign to tackle dementia misconceptions. Methods Taking 1414 tweets rated by carers from our previous work, we built 4 ML models. Using a 5-fold cross-validation, we evaluated them and performed a further blind validation with carers for the best 2 ML models; from this blind validation, we selected the best model overall. We codeveloped an awareness campaign and collected pre-post campaign tweets (N=4880), classifying them with our model as misconceptions or not. We analyzed dementia tweets from the United Kingdom across the campaign period (N=7124) to investigate how current events influenced misconception prevalence during this time. Results A random forest model best identified misconceptions with an accuracy of 82% from blind validation and found that 37% of the UK tweets (N=7124) about dementia across the campaign period were misconceptions. From this, we could track how the prevalence of misconceptions changed in response to top news stories in the United Kingdom. Misconceptions significantly rose around political topics and were highest (22/28, 79% of the dementia tweets) when there was controversy over the UK government allowing to continue hunting during the COVID-19 pandemic. After our campaign, there was no significant change in the prevalence of misconceptions. Conclusions Through codevelopment with carers, we developed an accurate ML model to predict misconceptions in dementia tweets. Our awareness campaign was ineffective, but similar campaigns could be enhanced through ML to respond to current events that affect misconceptions in real time.

show abstract

Section: Blind Validationmentioning

confidence: 88%

“…In accordance with the literature, tweets were preprocessed, which involved them being lemmatized first [ 31 ]. This ensured the words in the tweets were in their stem form (eg, “depression,” “depressed,” and “depressing,” would all be converted into “depress”); this removed typos and focused on the meaning of words.…”

Section: Methodsmentioning

confidence: 99%

Codeveloping and Evaluating a Campaign to Reduce Dementia Misconceptions on Twitter: Machine Learning Study

Erturk¹,

Hudson²,

Jansli³

et al. 2022

JMIR Infodemiology

Self Cite

View full text Add to dashboard Cite

show abstract

“…To extract these tweets, the team used keywords relating to seven of the most stigmatised conditions: Schizophrenia, Depression, Anxiety, Autism, Eating Disorders, obsessive compulsive disorder (OCD) (Robinson et al., 2019 ) and Addiction (Matthews et al., 2017 ). Search terms were based on those used previously in similar work (Robinson et al., 2019 ; Jilka et al., 2022 ; see supplementary material for keywords). We collected tweets during UK office hours (9 am–5 pm) spanning various pandemic stages: pre-UK lockdown (1st January–22nd March 2020), the first period of lockdown (23rd March–30th April 2020), and changing of lockdown rules (1st May–31st December 2020).…”

Section: Methodsmentioning

confidence: 99%

Investigating mental health service user views of stigma on Twitter during COVID-19: a mixed-methods study

Jansli

Hudson

Negbenose

et al. 2022

Journal of Mental Health

View full text Add to dashboard Cite

Background: Mental health stigma on social media is well studied, but not from the perspective of mental health service users. Coronavirus disease-19 (COVID-19) increased mental health discussions and may have impacted stigma. Objectives: (1) to understand how service users perceive and define mental health stigma on social media; (2) how COVID-19 shaped mental health conversations and social media use. Methods: We collected 2,700 tweets related to seven mental health conditions: schizophrenia, depression, anxiety, autism, eating disorders, OCD, and addiction. Twenty-seven service users rated them as stigmatising or neutral, followed by focus group discussions. Focus group transcripts were thematically analysed. Results: Participants rated 1,101 tweets (40.8%) as stigmatising. Tweets related to schizophrenia were most frequently classed as stigmatising (411/534, 77%). Tweets related to depression or anxiety were least stigmatising (139/634, 21.9%). A stigmatising tweet depended on perceived intention and context but some words (e.g. “psycho”) felt stigmatising irrespective of context. Discussion: The anonymity of social media seemingly increased stigma, but COVID-19 lockdowns improved mental health literacy. This is the first study to qualitatively investigate service users' views of stigma towards various mental health conditions on Twitter and we show stigma is common, particularly towards schizophrenia. Service user involvement is vital when designing solutions to stigma.

show abstract

“…Sentiment analysis describes the affective or emotional tone presented in the text [ 33 ] based on psychological evidence of the emotional meaning of constituent words or phrases [ 34 , 35 ]. It has been used in several health-related cases, such as in detecting language associated with depressive symptoms [ 36 , 37 ], extracting opinions on health care–related topics [ 38 ], and identifying mental health stigma in social media data [ 39 ]. The score derived from this analysis identifies text with positive, neutral, or negative tones on a continuous scale, where scores closer to −1 are very negative, scores closer to +1 are very positive, and a score of 0 is neutral.…”

Section: Methodsmentioning

confidence: 99%

Health Tracking via Mobile Apps for Depression Self-management: Qualitative Content Analysis of User Reviews

Polhemus¹,

Simblett²,

Dawe-Lane³

et al. 2022

JMIR Hum Factors

Self Cite

View full text Add to dashboard Cite

Background Tracking and visualizing health data using mobile apps can be an effective self-management strategy for mental health conditions. However, little evidence is available to guide the design of mental health–tracking mechanisms. Objective The aim of this study was to analyze the content of user reviews of depression self-management apps to guide the design of data tracking and visualization mechanisms for future apps. Methods We systematically reviewed depression self-management apps on Google Play and iOS App stores. English-language reviews of eligible apps published between January 1, 2018, and December 31, 2021, were extracted from the app stores. Reviews that referenced health tracking and data visualization were included in sentiment and qualitative framework analyses. Results The search identified 130 unique apps, 26 (20%) of which were eligible for inclusion. We included 783 reviews in the framework analysis, revealing 3 themes. Impact of app-based mental health tracking described how apps increased reviewers’ self-awareness and ultimately enabled condition self-management. The theme designing impactful mental health–tracking apps described reviewers’ feedback and requests for app features during data reporting, review, and visualization. It also described the desire for customization and contexts that moderated reviewer preference. Finally, implementing impactful mental health–tracking apps described considerations for integrating apps into a larger health ecosystem, as well as the influence of paywalls and technical issues on mental health tracking. Conclusions App-based mental health tracking supports depression self-management when features align with users’ individual needs and goals. Heterogeneous needs and preferences raise the need for flexibility in app design, posing challenges for app developers. Further research should prioritize the features based on their importance and impact on users.

show abstract

Identifying schizophrenia stigma on Twitter: a proof of principle model using service user supervised machine learning

Cited by 17 publications

References 30 publications

Codeveloping and Evaluating a Campaign to Reduce Dementia Misconceptions on Twitter: Machine Learning Study

Codeveloping and Evaluating a Campaign to Reduce Dementia Misconceptions on Twitter: Machine Learning Study

Investigating mental health service user views of stigma on Twitter during COVID-19: a mixed-methods study

Health Tracking via Mobile Apps for Depression Self-management: Qualitative Content Analysis of User Reviews

Contact Info

Product

Resources

About