Sentiment Analysis and Topic Modeling on Polycystic Ovary Syndrome from Online Forum Using Deep Learning Approach

Twitter and other internet sites offer an easy way for people apprehensive about communicating and exchanging information regarding health issues. These discussion boards are sources of information about people's opinions, which can be used to learn new information and analyse user behaviour. Polycystic ovary syndrome (PCOS) is a topic that has been discussed on Twitter. There has been a rise in the number of women diagnosed with Polycystic Ovary Syndrome in recent days. People use online forums like Twitter to express their positive and negative opinions, as well as share health-related concerns, complaints, symptoms, and needs. This study used to find useful recommendations for enhancing health-care programmes and understanding people's concerns by analysing these tweets. In this research article, sentimental analysis was uncovered many concerns associated to Polycystic Ovary Syndrome utilising Twitter data. The Latent Dirichlet Allocation (LDA), latent semantic analysis (LSA) and probabilistic latent semantic analysis (pLSA) methods are used to process of modelling the index of words in a topic. The results of evaluating the indexing process in subject modelling demonstrate that LDA performs better than LSA and pLSA in topic modelling of Polycystic Ovary Syndrome tweets, with a topic coherence of 0.1479, which is higher than LSA and pLSA. Then, Machine Learning techniques such as long short-term memory (LSTM), Naive Bayes (NB) and support vector machine (SVM) are used to classify the PCOS tweets. From the experimental results, it is inferred that the LSTM model provides an accuracy of 97%, which was higher than other machine-learning algorithms.