AI provides a picture of mental wellbeing from the bushfires and through COVID-19
It was an undeniably challenging start to 2020 for many Australians. The Black Summer of December and January destroyed environments, homes, businesses and communities, while in March the COVID-19 pandemic forced daily life as we knew it to a halt.
Lifeline has seen a significant increase in calls for help, with usage up by 25 per cent after 90,000 Australians phoned the crisis support service each month in March, April and May.
As we continue to respond to and be impacted by events that re-shape our nation, ensuring Australians mental health and emotional wellbeing is crucial.
Roughly 6,000 tweets expressing users’ feelings are posted every second on Twitter, providing a publicly accessible dataset that is analysed by an artificial intelligence-powered tool called WeFeel to show sentiment of people around the world in real-time.
Designed in 2014 by CSIRO’s Data61 in collaboration with the Black Dog Institute and Amazon Web Services, WeFeel analyses the words from millions of tweets to display a real-time view of users’ emotions.
It provides valuable insights on how a population’s emotions fluctuate over time due to changes in social, economic and environmental factors such as weather, time of day, news of a natural disaster or announcements about the economy.
WeFeel has revealed that in early January, New South Wales users experienced a significant spike of ’sadness’ in response to the news that two people had been killed by the South Coast bushfires, a result approximately three times higher than the average.
Tweets expressing ‘fear’ peaked in mid-March after a rise in COVID-19 cases, while users voiced their ‘surprise’ in early May at how well lockdown had worked.
The one day closure of Epping Boys High School on Friday 6th March saw a slew of ‘joyful’ posts, as did the arrival of much needed rain in Brisbane in February and the Perth Wildcats winning the NBL championship in early March.
Victorians were incredibly concerned by the arrival of the coronavirus in Australia in February, with a noticeable peak in ‘sadness’ and ‘fear’, while a climate change rally in mid-January, the East Gippsland bushfire of late February, and thousands protesting in response to the shutdown in early May inciting reactions dominated by ‘anger’.
“WeFeel could greatly assist in providing that real time data on how a nation is reacting to both natural and unnatural events, a bit like an organisational pulse survey,” explains Fiona Martin, CSIRO’s resident psychologist and member of the Health and Wellbeing team.
“WeFeel can help plan a quick response in terms of resource allocation when there is a notable increase in emotions that could impact people’s mental health. The sooner resources can be deployed, the better and more effective the outcomes.”
WeFeel looks for up to 600 specific words in a stream of around 27 million tweets per day and maps them to a hierarchy of emotions which include love, joy, surprise, anger, sadness and fear.
It does this by using natural language processing (NLP), a subset of artificial intelligence (AI), techniques to look at the words people type in their posts before mapping them in a hierarchy, or ‘wheel of emotions’.
WeFeel enables the exploration of emotional trends on a minute by minute time scale across locations and genders around the globe.
For Australians, these insights could ultimately predict when potentially life-saving services may be required in different geographical areas across the nation and help inform government decision making.
With untreated mental illness in Australia costing the economy up to $180 billion per year and the impact of COVID-19 expected to increase this figure, WeFeel could be a crucial tool to ensure a healthy future for all Australians.