Scholarlink

We presents a novel methodology for classifying pulmonary diseases using Vietnamese clinical text data, tackling issues of data complexity and inconsistency in electronic medical records, along with the inadequate applicability of English-centric models. Clinical data from An Giang Regional General Hospital were transformed into feature vectors using Embedding-based and Large Language Models (LLMs). The experimental outcomes demonstrate that the XGBoost model, when combined with Term Frequency–Inverse Document Frequency (TF-IDF) and Bag of Words (BoW) representations, achieved the highest classification accuracies of 87.66% and 87.63%, respectively, thereby outperforming the other evaluated models. Significant contributions encompass the development of a standardized Vietnamese clinical dataset for 12 pulmonary diseases and an illustration of the potential for integration into hospital information systems to improve diagnostic precision and healthcare efficiency.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Efficient Gradient Boosting Integrated with Advanced Text Vectorization for Pulmonary Disease Classification

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Phuoc-Hai Huynh

摘要

We presents a novel methodology for classifying pulmonary diseases using Vietnamese clinical text data, tackling issues of data complexity and inconsistency in electronic medical records, along with the inadequate applicability of English-centric models. Clinical data from An Giang Regional General Hospital were transformed into feature vectors using Embedding-based and Large Language Models (LLMs). The experimental outcomes demonstrate that the XGBoost model, when combined with Term Frequency–Inverse Document Frequency (TF-IDF) and Bag of Words (BoW) representations, achieved the highest classification accuracies of 87.66% and 87.63%, respectively, thereby outperforming the other evaluated models. Significant contributions encompass the development of a standardized Vietnamese clinical dataset for 12 pulmonary diseases and an illustration of the potential for integration into hospital information systems to improve diagnostic precision and healthcare efficiency.