MODELLING AND APPLICATION OF BIOMARKERS FOR AFFECTIVE STATE MINING USIING SOFT COMPUTING TECHNIQUES SOFT COMPUTING

Abstract

The craving to succeed in this fast-paced life is impacting the stress levels of everyone. Stress is a psychological condition in which a person feels overwhelmed with pressure. The body reacts to these changes with physical, mental, and emotional responses, these changes, the feelings that a person experience is their affective state. These changes can be observable or non-observable from the outside view. The measures of these changes are biomarkers, which can be of physiological, visual, electrochemical, or psycholinguistic in nature. Medical practitioners can measure and understand these biomarkers and can analyze the psychological and emotional state of a person. Analyzing the biomarkers manually is a tedious and time-consuming task for trained medical practitioners, delaying early identification and timely intervention. With the availability of IoT based sensors for healthcare, these biomarkers can be monitored using various wearable devices, implants, and cameras. Motivated by the need to design a model for affective state recognition using sensor-based bio-signals, this research proffers multi-model deep learning-based models for affective state recognition for identifying psychological as well as emotional state, using different fusion strategies to combine different modalities of biomarkers. Upsurge in IoMT devices has made the remote healthcare a reality. The bio-signals of patients can be monitored from remote location by medical professionals, making it possible for everyone to have access to healthcare. Though the benefits offered are unparalleled and promise useful decision support information, but all this is challenged by a lot of noise created owing to the large volume and variety of information sent at almost light speed. One of the most significant threats that the IoMT poses is of data security & privacy. As these devices capture and transmit data in real-time with no standard data protocols and data ownership regulations, it makes it highly susceptible to hacks and frauds. To resolve these issues, two novel methods has been proposed to reduce the size of the data using genetically optimized fuzzy c means clustering, and Federated transfer learning approach for proving privacy to the subjects, by training the model in decentralized environment. The preliminary results have shown promising results for both resolution techniques. This research also establishes the relation between cause and its effect on human affective state, by proposing a causal affective theory, and validating it with the help of a case study on Indian students during COVID-19.