Sentiment Analysis Tool for Pharmaceutical Industry & Healthcare

Abstract

Sentiment analysis (SA) is broadly used to analyze people's opinions about a product or an event to identify breakpoints in public opinion. Particularly, pharmaceutical companies use SA to ensure they gain a competitive edge through better understanding of patients' experiences allowing for more personalization and high responsiveness to consumers on social media. Patients selfreports on social media, frequently capture varied elements ranging from medical issues, product accessibility issues to potential side effects. The exploitation of such overwhelming unstructured data on social media through SA is of critical importance, however, the generalpurpose sentiment analysis tools are not adapted and do not incorporate the specific lexicon used in the life sciences/pharma context which reduces the ability of such tools to accurately detect the meanings of the sentiments expressed towards treatments/scientific studies and pharma companies at large. Indeed, those tools involve generalized dictionaries, techniques and methods that face several challenges to detecting the suitable sentiment polarity. In this paper, we develop a dedicated research tool that extract in real time the sentiments and emotions conveyed by users on social media regarding pharmaceutical industries. Our focus is to improve the analyze of opinion about a target in pharmaceutical industries based on hybrid approaches used in single unified system throughout sentiment analysis process to seek the relevant sentiment polarity, which is tailored specifically to detect positive, negative or neutral opinions of patients and consumer health. This paper proposes a first architecture validated by a real and industry case.

The impact of social media messages on Parkinson’s disease treatment: detecting genuine sentiment in patient notes

“…Many Distributions combined different pre-trained embeddings to enhance sentiment classification model performance, [6] proposed Sentiment Information Ex-tractor based on Bi-directional Long Short Term Memory structure that applied to join the results of various subextractors. In particular scope, few approaches have been proposed for mining sentiment from related-medication text [7]. Patients notes on social networks and the use of informal medical language pose additional text challenges, including non-standard format, wrongly spelt, and abbreviation forms, as well as typos in social media messages.…”