Abstract

A review that we conducted revealed that neither conventional nor modern methods for improving mental health assessment and treatment are adequate in the IT industry. We aimed to develop a significantly improved framework for identifying and addressing mental health issues, starting within the IT sector. Currently, the surveying approach stands as the most innovative method, boasting an accuracy rate of approximately 81%. In contrast, other electronic methods such as facial recognition and sentiment analysis, due to their low precision or implementation challenges, are not viable options. Leveraging the OSMI 2018 dataset and a similar surveying methodology, we constructed and evaluated nine machine learning models. Among these, in terms of accuracy, the Random Forest model performed exceptionally well of approximately 96%. In light of this achievement, we formulated a 15-item questionnaire that proves highly effective in assessing whether a tech professional exhibits signs of mental instability, both in their interactions with colleagues and in their personal life.