A digital platform supported by Artificial Intelligence promotes emotional health in the workplace
11% of mental illnesses causing occupational incapacity are attributed to working conditions. Factors such as work-related stress are precursors of so-called common mental disorders such as anxiety, depression, burnout, mood disorders and substance abuse, which together constitute one of the main causes of common sick leave.
These data show the need for action to promote mental health in the workplace. However, interventions to prevent occupational risks tend to focus on safety, hygiene and ergonomic factors, relegating psychosocial factors to a secondary role.
Two teams from the University of Barcelona (UB) are involved in a new consortium to develop a digital work ecosystem based on scientific evidence with the aim of reducing the psychological gap and promoting emotional health in the workplace. The initiative will incorporate cutting-edge methodologies in psychological assessment, occupational health psychology, psychometrics, and artificial intelligence (AI) to improve the health of workers, increase productivity levels and boost the economy.
“The main aim of the project is to create an integrated platform of solutions and resources for the evaluation and support of mental health management in the workplace, covering all areas of prevention, care, and return to work for employees,” explains David Gallardo-Pujol, a lecturer at the Faculty of Psychology and researcher at the UB’s Institute of Neurosciences.
Dr. Gallardo is in charge of coordinating the university’s contribution to this project, which also includes researchers from the UB’s Human Analysis and Intelligent Systems (HuPBA) group, led by Sergio Escalera, a lecturer from the Faculty of Mathematics and Computer Science. The company MetrikaMind leads the public-private consortium together with the company HappyForce and the University of Murcia.
Assessing and predicting employment-related mental health
One of the innovative factors of the project is the recurrent and automated assessment of mental health and its effects on functionality in terms of work performance, with an approach that combines psychology, data science, and AI. To this end, the project will establish metrics that reflect the correlation between the psychosocial resources and capabilities of workers and the psychosocial demands of their jobs, as well as their value as a predictor of productivity and absenteeism due to common mental disorders.
“The goal is to assess the perceived level of mental capacity of workers in Spanish companies in relation to the demands of the job and the organisation, as well as the availability of resources and skills to cope with these,” researchers emphasise. Among the benefits of this approach, it will help to improve the adaptation of the job to the characteristics of employees, reducing absenteeism and enhancing productivity.
A questionnaire with honesty filters
One of the keys to this approach will be the use of a valid, reliable, open-access psychometric questionnaire with psychometric honesty filters to achieve an objective assessment and thus overcome one of the problems of questionnaires related to mental health and well-being: the lack of honesty and subjectivity in the answers. This situation occurs, for example, when workers do not want to perform certain activities or, conversely, when they try to make a good impression on the evaluators.
Explainable and highly predictive algorithms
The information collected through the questionnaires and supported by pioneering psychometric and AI machine learning techniques —which will be led by UB researchers— will allow researchers to make predictions about the evolution of the mental health of workers and thus “support decision-making to improve productivity and mental health, as well as ensuring regulatory compliance regarding occupational risk prevention services and the identification of company liability, if applicable,” researchers point out.
In this sense, a key element of the project is the development of ‘explainable’, fair and highly predictive algorithms. “It is not only critical to achieve high accuracy in the recognition of, for example, anxiety and depression from predictive models, but also to understand the behaviour of deep learning AI and the decisions it makes in order to produce such predictions. Such explainability will allow experts to reach a justified decision in relation to diagnosis,” says Dr. Gallardo.
This intersection between this wealth of information and AI-based predictive models may even lead to “changes in the way we analyse and interpret behavioural data in psychology,” note the researchers. In addition, an open data repository will be created, which will be made available to researchers under confidentiality agreements.
A system capable of adapting to other cultural contexts
During the three years of the project, the platform and the questionnaire will be developed and pilot tested, which will include a validation with a minimum of 5,000 workers, to end up with a system ready for market introduction. The results can also be used as an indicator for the implementation of policies to improve mental health, or to monitor them. Moreover, it is a “pioneering initiative at a global level that can be easily adapted to other cultural contexts, facilitating the internationalisation of the final product,” researchers underline.
The consortium, with reference number SCPP2200C010001, has obtained 218.125 euros in funding, which has been awarded to the UB within the framework of the 2022 call for public-private collaboration projects of the 2021-2023 Spanish National Plan for Scientific, Technical and Innovation Research, as part of the Recovery, Transformation and Resilience Plan.