A Smart Federated Learning Approach for Mental Health Detection

Mental health has emerged as an important part of health care and self-care following the recent COVID-19 pandemic. With everyone being locked in their homes, it has led to isolation and disconnect from human presence and this has caused people to realize the importance of mental well-being and fitness along with physical fitness. Over time, various forms of mental illnesses, such as depression, anxiety disorders, schizophrenia, and bipolar disorder, have been identified and recognized, highlighting the diverse and concerning nature of these conditions. Out of all the illnesses, the most common is depression which has been identified in mostly adults and senior citizens and rarely in children. With this study, we aim to make mental health detection easier and recognize early signs by harnessing data from a multitude of sources including wearable devices, smartphone apps, electronic health records, and social media. Our main objective of using federated learning was to protect the patient’s privacy which might have been compromised using other predictive learning techniques.