Summary of Medical-gat: Cancer Document Classification Leveraging Graph-based Residual Network For Scenarios with Limited Data, by Elias Hossain et al.
Medical-GAT: Cancer Document Classification Leveraging Graph-Based Residual Network for Scenarios with Limited Data
by Elias Hossain, Tasfia Nuzhat, Shamsul Masum, Shahram Rahimi, Sudip Mittal, Noorbakhsh Amiri Golilarz
First submitted to arxiv on: 19 Oct 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary A new dataset of biomedical abstracts is presented to improve the classification of cancer-related documents in healthcare management and research. The dataset consists of 1,874 abstracts categorized into thyroid cancer, colon cancer, lung cancer, and generic topics. A Residual Graph Attention Network (R-GAT) model is introduced, which captures semantic information and structural relationships within documents using multiple graph attention layers. The R-GAT model outperforms other techniques, achieving high precision, recall, and F1 scores for different types of cancer and generic topics. Ensemble approaches combining deep learning models are also explored to enhance classification. Feature extraction methods such as TF-IDF, Word2Vec, and tokenizers from BERT and RoBERTA are assessed. The R-GAT model is compared with transformer-based models like BERT, RoBERTa, domain-specific models like BioBERT and Bio+ClinicalBERT, deep learning models (CNNs, LSTMs), and traditional machine learning models (Logistic Regression, SVM). |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper presents a new dataset of biomedical abstracts that can help improve the classification of cancer-related documents. The dataset is used to train a special type of AI model called a Residual Graph Attention Network. This model is good at understanding the relationships between words in documents and is able to classify documents into different categories, such as thyroid cancer or colon cancer. The model performed well compared to other techniques used for this task. |
Keywords
» Artificial intelligence » Attention » Bert » Classification » Deep learning » Feature extraction » Graph attention network » Logistic regression » Machine learning » Precision » Recall » Tf idf » Transformer » Word2vec
