Natural Language Processing for Understanding Chronic Illness Patients\' Narratives

• Ano de defesa: 2024
• Autor(a) principal: Ito, Viviane
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Análise de Sentimento; Clinical Natural Language Processing; Heart Failure; Insuficiência Cardíaca; Modelagem de Tópicos; Natural Language Processing; Processamento de Linguagem Natural; Processamento de Linguagem Natural Clínico; Sentiment Analysis; Topic Modeling
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/8/8139/tde-06012025-122305/

Natural Language Processing (NLP) has become increasingly prevalent in healthcare due to the digitization of medical records and advancements in computational techniques. Initially used for tasks such as automated coding and information extraction from electronic health records, NLP applications have expanded to include disease prediction, decision support, and patient experience analysis. However, applying NLP to healthcare data presents challenges, including deciphering complex medical language, ensuring patient privacy, and maintaining data integrity across various sources. This study focuses on analyzing the linguistic patterns in interviews with heart failure (HF) patients, an area that has received limited attention. While prior research has applied NLP to structured data and qualitative analysis of patient feedback, there is a gap in understanding the linguistic aspects of HF patients\' experiences. This study aims to fill this gap by applying NLP methods to transcribed interviews of 266 HF patients from the Heart Institute at the University of São Paulo, Brazil. The objectives are to characterize the linguistic patterns patients use to describe their experiences and to demonstrate how NLP methods can be applied to study narratives about chronic illnesses. My methods include Topic Detection with BERTopic, Sentiment Analysis, and Emotion Detection, complemented by Inferential Analyses to test interactions between extracted variables and patients narratives. The findings aim to uncover relevant linguistic information to better characterize the experiences of HF patients and provide healthcare professionals with insights into handling the sensitive aspects of the disease. This study aims to contribute to the broader understanding of the social and cultural dimensions of heart failure and the crucial role language plays in patient care and communication. The methods applied to this research can also be adapted to other studies on Chronic Illness narratives