Do Patients Tend to Find Positive or Negative Feedback on Social Networks? A Study of The Main Aspects of Modelling Patient Understanding Based on Emotional Variants

Abstract

Information on social networks can have an immediate influence on the public health status regarding a given medication or treatment. That means such positive or negative information about a given drug or service is learned, which frequently affects the emotional state of a patient. Here we draw upon the main question “Do Patients tend to Find Positive or Negative Feedback on Social Networks?”. Using sentiment analysis to comprehend the emotional state of patients is of critical importance, not only analysis of the text is required, a deep study of the main aspects of modeling patient understanding based on the emotional variants approach becomes apparent. Existing feature learning algorithms fail to define a relevant feature that promises the understanding of affect conveyed towards a given target. In this study, the goal is to define a based-emotional embedded-concepts spectrum that consists of defining affective information from selected affect seeds to other biomedical concepts. We develop concept-level emotional classification according to three categorical emotional variants and moral basics: PAIN or AFFECT, MOOD, and POSITIONALITY, where the objective is to predict the predominant concept-level affects. The experimental results on Twitter data show that the proposed strategy achieved significant performance improvements, thus, it might have an impact in real-world scenarios and helps provide situational awareness.