How has the COVID-19 pandemic affected our collective mental wellbeing? What are the mental health implications of extensive quarantining and restriction of movement?
Government and institutes are keen to learn the answers to these two questions. Policymakers need to be aware of all consequences and potential risks associated with both long and short-term lockdowns to be in a better position to decide what measures are in the best interests of the public.
The word “unprecedented” has entered everyday vernacular for good reason. Never before have we collectively experienced such intense periods of isolation from loved ones, such extensive periods of social distancing, or such severe restrictions of movement.
Given the lack of precedent, policymakers have had to make educated guesses as to the impact of these restrictions. Government bodies, institutes and health organisations around the world have sent out national surveys to track mood and mental wellbeing in their respective countries. But these surveys take time to collect and process. What’s needed are real-time, large-scale studies that can be reacted upon promptly.
In Germany, the Robert Koch Institut (RKI) took a different approach. They used both data from national surveys and data collected from Twitter and processed by Symanto AI technology. The results were potentially revolutionary.
Symanto applied proprietary text analytics (Psychology AI) on over one million Tweets and identified nearly eighty-nine thousand depressive-symptom-related Tweets.
The results showed a strong correlation with the findings of a weekly national survey conducted by RKI. For example, both results reported feelings of worthlessness were the least frequently reported depressive symptoms, while fatigue and energy loss were most frequently reported.
This correlation proves the potential of Psychology AI to revolutionise how governments and institutes monitor population mental wellbeing. You can get a deeper look into how the coronavirus pandemic and fallout are effecting mental wellbeing with the Symanto’s Mental Health Tracker.
Using Twitter and Psychology AI as Mental Health Measurement Tools
This study expands upon existing research that indicates the potential to detect signs of depression in Tweets via behavioural and linguistic analysis.
The 2019 research, conducted by Hospital del Mar Medical Research Institute, concluded that Twitter users with potential depressive disorder modify general characteristics of their language and their interactions on social media.
For the RKI study, Symanto specially devised a Psychology AI module to detect online discussions around depressive symptoms. These symptoms were based on the eight-item Patient Health Questionnaire depression scale (PHQ-8).
Discussion of mental health issues on Twitter
PHQ-8 is an established and valid questionnaire aimed at diagnosing and measuring the severity of depressive symptoms in large clinical studies.
The eight questions ask respondents to mark how often they have been affected by the following problems on a scale of “not at all” to “nearly every day”:
- Lack of interest or pleasure in doing things
- Feeling depressed, hopeless, or down
- Trouble getting to sleep, staying asleep or sleeping too much
- Feeling tired or lacking energy
- Poor control of eating habits (either eating too much or too little)
- Negative thoughts about yourself (e.g. that you have let people down)
- Difficulty concentrating on things, e.g. reading or watching television
- Slow movement or speech to the extent that other people have noticed. Or on the contrary – feeling fidgety and restless and moving more than usual
Using the PHQ-8 as a reference, Symanto queried tweets that contained terms indicative of depressive problems.
The Symanto Psychology AI module was also designed to predict the age and gender of Twitter authors to further enrich the aggregated results.
Using Symanto’s AI module along with freely available data on Twitter as a mental health measurement tool has several key benefits:
- GDPR-compliant: Twitter data is publicly available and easily accessible.
- Far-reaching: Twitter has around 330 million active users per month.
- Cost-saving: Distributing, collecting and analysing national survey data is costly in terms of time and resources. By comparison, the Symanto AI module collects data that is already readily available and has been demonstrated to process hundreds of thousands of data points within a matter of minutes.
- Real-time: Data is continually updated and therefore always relevant. This is particularly useful given the volatility of the pandemic and the measures put in place to control it.
Symanto’s Psychology AI module – especially for emotion and topic detection around mental health struggles – has far-reaching potential applications as a mental health measurement tool. Monitor public mental wellbeing and detect hot spots of improving and declining mental health to help with early intervention. Furthermore, the module can be applied to online conversations in over 15 languages.