After two years of health crisis, 41% of employees say they are in psychological distress due to successive protocols, change in work pace or lack of prospects. According to the latest OpinionWay survey for Human Footprint [cabinet spécialisé dans les risques psychosociaux et la qualité de vie au travail], 34% of employees are in burn-out, including 13% in severe burn-out, i.e. 2.5 million people. According to the Health Watch Institute, 480,000 people in France are in psychological distress at work and burnout concerns 7%or 30,000.
What are we talking about when we talk about burnout?
According to the INRS, burnout is characterized by intense emotional fatigue and a professional “feeling of non-accomplishment”. This is an excess of work that drains the employee of all his energy. According to the WHO, it is a “pathology resulting from chronic occupational stress that has not been properly managed” – with three components: “feelings of loss or exhaustion”, “mental distance heightened with work or feelings of work-related negativity or cynicism”, and “decreased job efficacy”.
To avoid burnout or get out of it, it is essential to be able to detect it as quickly as possible; especially through several warning signs, emotional or physical. But health professionals are still struggling to spot this syndrome early. First of all, as Emmanuelle Wyart, professional coach and psychosocial risk preventer, writes in “Burn-out: it’s not your fault but it may be your luck”, because it’s about a pathology comprising “a rich and polymorphic symptomalogy”. Thus, many symptoms can be the expression of other illnesses or syndromes (depression or anxiety, among others), and “do not on their own qualify burnout”. Then, because current diagnostic tools are “incomplete”.
Diagnostic tools not yet reliable enough
The tools that exist today are thus far from benefiting from a medical consensus. The most used, the tests MBI (Maslach Burnout Inventory), OLBI (Oldenburg Burnout Inventory), CBI (Copenhagen Burnout Inventory), SMBM (Shirom Melamed Burnout Measure) and BM (Burn-out Measure), include dozens of questions that explore emotional exhaustion, relationship to others and the degree accomplishment at work. They aim to assess the severity of the syndrome. Examples of standard questions and answers: “I feel exhausted at the end of my working day: never / sometimes / every day”; or even “Working with people throughout the day requires a lot of effort: never / sometimes / every day””.
But if these tests are validated by bodies in psychology, “many criticisms are addressed to them, and they are not unanimous within the scientific community”, observes Emmanuelle Wyart. In particular because they are not that reliable, the results being often biased. “These questionnaires have only (closed) scale questions, and no free-text questions. In addition, some people are reluctant to tick ‘never’ or ‘everyday’ answers, or are tempted to lie to influence the results”, explains Mascha Kurpicz-Brikiprofessor of data engineering at the Bernese University of Applied Sciences in Biel, Switzerland.
With a team of specialists in AI and applied psychology, this Swiss researcher has developed a method based on the automatic analysis of texts, to overcome this limitation. “Text analysis, for example in the form of interview transcripts, has been shown to hold promise, but is often not carried out in practice due to the extra time required for manual text evaluation But with AI, we can detect burnout more easily and quickly, with greater efficiency and reliability,” she notes.
A more efficient and faster “screening” of burn-out thanks to AI?
The texts analyzed by the Swiss researchers (anonymously) came here from the English-speaking community website Reddit and its many forums, or “subreddits” (and not only those dedicated to burn-out). Mascha Kurpicz-Briki’s team thus put together a dataset consisting of 13,568 samples describing first-hand experiences, of which 352 are related to burnout and 979 to depression. She then resorted to machine learning. Concretely, algorithms automatically analyze the texts and identify whether the language is burn-out or not. For this, the collected texts were labeled and classified into two groups, between those referring to this syndrome, and those which are not linked to it. Then models were trained, via several different configurations, to determine whether a text contains “indications on burn-out” or not.
According to the Swiss scientist, this method of diagnosis by Natural Language Processing (NLP, natural language processing) successfully identified 93% of burnout cases. But the data used are still few in number for the results to be completely certain. “As a next step, we want to apply our method in collaboration with clinical partners, in order to further develop and validate such methods. In addition, we will need a more diverse training dataset, including different groups of the population, because we have so far worked with entirely anonymous data. Due to the difficulty of finding sufficient data, we have used English texts in this study. Our medium-term objective is to apply our methods to texts in German and French, from patients whose burnout is confirmed”, says Mascha Kurpicz-Briki.
For the researcher, this technology could be used in a clinical context, to “create new tools” for psychology/psychiatry: “For example, interview transcripts could be analyzed, or new forms of inventories including questions in free text could be defined and validated using this method”. Could we go so far as to imagine doing the same by analyzing words orally, through the patient’s voice? “We are investigating the possibility of automatically transcribing spoken texts, for example from interviews with patients, and then analyzing them. However, this can be difficult in the case of specific dialects, which can be complicated to transcribe,” she says.
An “augmented intelligence” rather than artificial
But beware: there is no question of automating everything. Mascha Kurpicz-Briki thus specifies that the collaboration of AI experts and medical experts remains essential. Initially to verify the conclusions of this study on real cases of burnout and on a representative sample of the population, but even beyond. “Our work is oriented towards the augmented intelligence approach rather than towards artificial intelligence: instead of replacing humans (health professionals), such a tool will have to support them in their daily work. In the context of our work, it would help the clinical practitioner by providing insights into the patient, like a decision aid,” she explains.
Because the researcher is not afraid to admit it: she and her team are unable to know exactly which words or which turns of phrase are retained by their AI system as symptoms of burnout. It is thus a method comparable to a “black box”, which it would be better, therefore, not to use without a minimum of hindsight. “It is very difficult to assess how AI arrives at such and such a decision. This is the challenge we are trying to meet, but in absolute terms, this tool cannot be automated”, she concludes.
Such technology could make more “intelligent” the “psy chatbots” which have been multiplying for 4 or 5 years. owlie, Woebot, wysa and Tess are conversational agents of “psy support”, who ask you questions in order to determine your moral and mental state, and to give you advice. Or refer you to a human specialist. These chatbots already detect key words for this, such as “stress”, “insomnia”, “anger”, via an AI. Machine learning and natural language processing allow them to “help users manage their emotions”, and “relieve their depression”. But their effectiveness remains limited at present. Hence the interest of the method developed by Mascha Kurpicz-Briki and his team. A tool that will never, of course, replace human shrinks, but which could allow them to better connect with their patients, as far upstream as possible.
The same scheme would be applicable in the case of a chatbot that would detect the emotions of a human orally, as is beginning to do Ellie. This “artificial avatar” designed at the Institute of Creative Technologies of the USC (University of Southern California) by researchers in computer science and psychologists, is thus able to discuss with patients, but also to detect their emotions by analyzing their facial expressions and words. Much like the Owlie or Woebot chatbots, Ellie’s AI starts with general questions and then moves on to more specific clinical questions.
Throughout the discussion, Ellie uses machine vision to interpret verbal and facial cues from the patient’s facial expressions and tone of voice, to tailor her questions, but also her responses — which can be nodding head, smile or utter a “hum” when the subject is telling a sensitive story. These subtle “reactions” thus make it possible to create a closer “relationship” with the patient, and to invite him to share more information. If Mascha Kurpicz-Briki’s technique could be applied to oral discussions, it is therefore difficult not to imagine its contribution to projects such as that of USC. But here too, the human would remain at the center. As Gale Lucas, the researcher behind Ellie, explains to us, AI can only be tools at the service of human therapists: “With our virtual avatars, we don’t want to replace humans. Robots can detect better than we non-verbal information is the big potential of these technologies. They can be used to identify people in distress, but the final treatment will always lie with human therapists.”
Note that in Walloon Belgium, other researchers are working on a solution using AI to more easily detect burnout. Not based on the text, but on the analysis of data from the patient’s smartphone and connected objects. “By observing behavioral data produced by the phone and other connected devices, we can easily detect and recognize key symptoms such as behavior change, physical exhaustion and even possible unusual emotional reactions” , explains Antoine Sepulcher, founder of NOÖSa NOÖS platform, dedicated to mental health and well-being, which is collaborating on this project with specialists from the UCLouvain artificial intelligence research laboratory.
The behaviors analyzed via connected objects and other smartphones include a change in the way the user markets, in the intonation of his voice, in his sleep pattern, in his searches on Google, in his eating habits, or his musical choices again. A potentially effective method, but quite risky on the side protection of personal (health) dataand even ethics… This is why, here again, the researchers insist on the importance of leaving “the human being at the heart of the solution”.
We would like to thank the writer of this article for this incredible web content
A more efficient and faster “screening” of burn-out thanks to AI? – CNET France
Our social media profiles here as well as additional related pages here.https://yaroos.com/related-pages/