Word2Vec and Doc2Vec in Unsupervised Sentiment Analysis of Clinical Discharge Summaries

TL;DR

This paper investigates the use of Word2Vec and Doc2Vec for unsupervised sentiment analysis of clinical discharge summaries, aiming to detect biases related to diseases without relying on annotated data.

Contribution

It demonstrates the application of unsupervised Word2Vec and Doc2Vec methods for sentiment analysis in clinical texts, highlighting their complementary strengths.

Findings

Word2Vec and Doc2Vec complement each other in sentiment detection.

Unsupervised methods can effectively analyze clinical sentiment without annotated datasets.

The approach reveals potential biases in clinical discharge summaries.

Abstract

In this study, we explored application of Word2Vec and Doc2Vec for sentiment analysis of clinical discharge summaries. We applied unsupervised learning since the data sets did not have sentiment annotations. Note that unsupervised learning is a more realistic scenario than supervised learning which requires an access to a training set of sentiment-annotated data. We aim to detect if there exists any underlying bias towards or against a certain disease. We used SentiWordNet to establish a gold sentiment standard for the data sets and evaluate performance of Word2Vec and Doc2Vec methods. We have shown that the Word2vec and Doc2Vec methods complement each other results in sentiment analysis of the data sets.