AI & Robotics

Coronavirus stress deciphered – thanks to AI

The Fraunhofer Institute for Industrial Engineering (IAO) has deciphered the long-term psychological consequences of the coronavirus pandemic using Artificial Intelligence. Thanks to applied machine-learning methods, the scientists were able to correlate and analyse massive volumes of data..

Social distancing, curfews, mandatory mask-wearing, working from home, isolation: the COVID-19 pandemic seriously took its toll on people’s social lives. Every lockdown inevitably has an impact on our psyche – and thus also on our health and well-being. 

To find out which specific groups of people are particularly affected by the effects of the pandemic, the University Hospital Dresden joined forced with the Fraunhofer IAO and interviewed 275 German respondents in a project entitled “WIBCE – what I am experiencing during the coronavirus pandemic”. Apart from their health-related data, such as belonging to a particular risk group, their behaviour (for example, the number and type of contacts, the care of children or the use of certain information sources) and a psychological assessment were also collated. 

Assessing long-term consequences

The unique aspect of WIBCE: the respondents had several question and answer sessions with the researchers between April and August 2020. The results therefore reflect a snapshot of life during the pandemic, but also represent an ongoing record. “It was really only thanks to our AI methodology that we were able to evaluate such large amounts of data with reasonable effort,” explains Doris Janssen, Project Manager at the Fraunhofer IAO. In total, over 1,430 surveys were conducted over the entire period. Typical machine-learning (ML) methods were used for data exploration: unsupervised clustering methods and supervised classification models of the family ensemble. Correlation analysis between the individual variables and statistical comparisons between the individual clusters using bootstrap confidence intervals completed the data analysis. 

Doris Janssen, Project Manager at the Fraunhofer IAO | © Fraunhofer IAO

Do you belong to the “untroubled” or “concerned” group?

Thanks to AI, the researchers were able to identify two clusters among the respondents that differed in age, psychological protection and stress factors: 

1. The “untroubled” group is coping relatively well with the restrictions caused by the COVID-19 pandemic. Their resilience had scarcely changed compared to the pre-pandemic phase. Around two-thirds of the respondents belong to this cluster.

2. By contrast, one third of the respondents belong to the “concerned” group, which is coping with the pandemic situation in a more fearful and depressed manner. This group also includes people who have not contracted the coronavirus. Nonetheless, they are experiencing massive psychological stress during the pandemic, well above the pre-pandemic reference value.  

Young + low income = high psychological stress

One thing is clear after evaluating all the data from the machine learning methods: younger people – especially those with lower incomes – are struggling with relatively high levels of psychological stress during the pandemic. And this, despite the fact that they are exposed to a relatively low objective health risk. In the long term, it is also completely irrelevant whether or not they become infected with the virus during the survey period. The Fraunhofer IAO concludes that politics and society must take the worries and concerns of this population group seriously and put in place appropriate means of support.


Link

Download the complete report with all the key interim results from the WIBCE project as a PDF document on the IAO website: http://publica.fraunhofer.de/documents/N-630598.html

Text: Susanne Widrat