Feature Engineering for Recognizing Adverse Drug Reactions from Twitter Posts

Abstract

Social media platforms are emerging digital communication channels that provide an easy way for common people to share their health and medication experiences online. With more people discussing their health information online publicly, social media platforms present a rich source of information for exploring adverse drug reactions (ADRs). ADRs are major public health problems that result in deaths and hospitalizations of millions of people. Unfortunately, not all ADRs are identified before a drug is made available in the market. In this study, an ADR event monitoring system is developed which can recognize ADR mentions from a tweet and classify its assertion. We explored several entity recognition features, feature conjunctions, and feature selection and analyzed their characteristics and impacts on the recognition of ADRs, which have never been studied previously. The results demonstrate that the entity recognition performance for ADR can achieve an F-score of 0.562 on the PSB Social Media Mining shared task dataset, which outperforms the partial-matching-based method by 0.122. After feature selection, the F-score can be further improved by 0.026. This novel technique of text mining utilizing shared online social media data will open an array of opportunities for researchers to explore various health related issues.

TwitPersonality: Computing Personality Traits from Tweets Using Word Embeddings and Supervised Learning

“…It consists of classifying the sentiment polarity of a sentence, as positive, negative or neutral. Dai et al [49] explored numerous feature engineering techniques for detecting adverse drug reactions from Twitter posts. Studies on demographics are also worth of note, in particular for those sites (e.g., Twitter) where demographic information is mostly unavailable.…”

BIGODM System in the Social Media Mining for Health Applications Shared Task 2019

“…In order to extract features for training our classifiers, we first pre-processed tweets to replace URLs, dosages and Twitter specific characters with the corresponding symbols, and modified the numeral parts in each token to one as proposed in our previous work Dai et al (2016). The preprocessed tweet was then processed by a tweet tokenizer (Owoputi et al, 2013) to generate tokens.…”

“… Domain knowledge features: The presence of adverse drug reaction (ADR) or drug mentions were engineered as two binary features with the value of either 0 or 1. The occurrences of ADR and drug names were recognized by using the ADR mention recognizer developed in our previous work Dai et al (2016) and Wang et al (2018).…”

Natural Language Analysis of Online Health Forums

“…In a recent paper, Korkontzelos et al [9] analysed the effect of sentiment analysis features in ADR classification, which made use of rules such as "negation" to improve the performance of their system. Dai et al [3] also investigated features to use for finding ADR in Twitter posts.…”