The Project
The Project
Scientific basis
Stress, the psychophysical response to overwhelming emotional, cognitive, or social tasks, impacts lives in more ways than we realize. Often, we label ourselves as “stressed,” yet recognizing all the symptoms and understanding its true impact proves challenging. Underestimating stress can lead to chronic dimensions, affecting our overall well-being.
Recent studies reveal that stress is not an isolated symptom but a complex blend of interconnected factors: anxiety, depression, and physical fatigue, as experienced by individuals themselves. Psychologists rely on questionnaires like the Perceived Stress Scale (PSS), Hamilton Depression Rating Scale (HDRS), and Hamilton Anxiety Rating Scale (HAM-A) to assess psychological stress levels and aid in diagnostics.
However, there are limitations. Can a single questionnaire truly capture the depth of someone’s psychological stress? Can the correspondence between interviews, questionnaires, and stress indicators create a reliable database to train an algorithm accurately distinguishing between stressed and non-stressed states?
Yet, a breakthrough emerges—a unique window into our well-being: the way we express ourselves. Research highlights communication parameters that offer insights into stress levels. A groundbreaking study by Mehl and colleagues (2017) reveals that intense social stress can impact the genes responsible for inflammation, potentially leading to severe chronic conditions. Furthermore, a link between individual language use and gene-expression profiles has been discovered.
Building on these groundbreaking findings, the Myndoor team embarked on extensive studies, identifying linguistic markers that detect stress. Over six years, texts and diaries from a diverse global sample were meticulously analyzed, correlated with physiological parameters (heart rate, blood pressure, cortisol levels), and combined with data from psychological questionnaires and cognitive function tests.
The initial correlations between emotional states and communication were published in scientific articles, presented at conferences, and shared in academic circles. Now, we’re taking it further—leveraging artificial intelligence and machine learning. Our goal: a reproducible and predictive system that intertwines the insights of psychologists with cutting-edge technology.
Myndoor technology
Myndoor has pioneered an exclusive, patented Artificial Intelligence model that revolutionizes the way we measure stress levels in textual content. Built upon the advancements of industry leaders, our solution possesses unique characteristics derived from the training of a Natural Language Processing algorithm fueled by a semantic engine.
What sets us apart is the culmination of over 7 years of data collection, through a process known as knowledge graph walking, resulting in a semantic database. This positions our model in line with the latest sentiment analysis technologies, while blazing new trails by enabling the differentiation of information based on semantic context.
Our model is equipped with self-evaluation capabilities, constantly seeking improvement opportunities. This allows our team to actively intervene and retrain the algorithm, refining its performance to perfection.
Currently, our model boasts an impressive accuracy (F1 score) of 83% across all forms of communication. Rest assured, you can trust Myndoor’s AI model to deliver unparalleled insights into stress quantification.
API
Myndoor technology is distributed through a cutting-edge API designed for firms in the realms of corporate welfare, wellness services, and consulting. If you’re seeking innovation and technology, our API is your gateway to revolutionizing your operations. Seamless integration with any back-end component empowers you to analyze both written and oral communication data, tailored to your specific needs.
Rest assured, data privacy is at the core of our design. We strictly adhere to GDPR guidelines, ensuring data anonymity and encryption. Our API operates without storing any records of your data, providing you with peace of mind.
Psychological analysis companies and businesses in the healthcare and wellness sector that leverage semantic analysis tools to understand the emotional states of their users stand to benefit immensely from our technology.
How it works
Our team’s work goes beyond the realm of advanced Deep and Machine Learning techniques. We dive deep into the analysis of data provided by certified professionals, enabling us to gain profound insights into the relationship between lexical and syntactic choices in the text through natural language processing (NLP).
To address the uncertainties arising from the myriad of questionnaires available, we’ve developed an innovative “ensemble” approach. By aggregating questionnaire results, we mitigate errors and inaccuracies inherent in individual scores.
Moreover, variations in the number of questions, diverse response scales, and discrepancies in completion times have spurred the development of our pioneering Artificial Intelligence model.
Today, our model can assess an individual’s stress levels without relying on questionnaires, solely by analyzing their communication style.