An explainable predictive model for the detection of mental health conditions in the workplace

Mental Health is a ubiquitous contemporary issue causing significant mortality and morbidity in today’s society. A strong need exists to develop prompt and robust diagnosis and treatment options. Further, with altered workplace conditions (especially in the Information Technology field) and more individuals working from home, a safe, supportive and flexible work environment is paramount. This is where the real potential of Machine Learning (ML) and Artificial Intelligence (AI) comes into play. The intersection of ML, AI, and Mental Health (MH) in Personalized Healthcare (PH) is a rapidly growing field that is opening up many new opportunities in terms of early diagnosis, prompt interventions, and personalized healthcare plans on a patient-specific basis. This research seeks to apply ML to predict the likelihood of an individual developing an MH condition. The work also considers specific mental states, such as anxiety disorder and mood disorder, based on a series of variables containing information about their workplace conditions. Three predictive models are developed, each with a strong level of performance, particularly for predicting the presence of a general mental health disorder and mood disorder. The research finds that the employee’s ease of leave accessibility, the overall importance an employer places on staff MH, and the level of company-wide support in addressing MH concerns are critical factors influencing employees’ mental health.