霍奇金淋巴瘤社交媒体帖子实体提取频率数据集

数据集获取（Figshare）

• 源站标题：Frequency of extracted class and its attributes.
• DOI：10.1371/journal.pdig.0000765.t003
• 发布日期：2025-03-19
• 文件数：1 个
• 文件总体积：13.82 KB
• 源站页面：https://plos.figshare.com/articles/dataset/Frequency_of_extracted_class_and_its_attributes_/28626983

命令行下载

curl -L -o "Table 3.xls" "https://ndownloader.figshare.com/files/53113259"

文件清单

• Table 3.xls（13.82 KB）

文件总体积 13.82 KB（源站 API 实测）。

数据加载示例（图像类）

from PIL import Image
import glob, os

files = (glob.glob(os.path.join(path, "**", "*.png"), recursive=True)
       + glob.glob(os.path.join(path, "**", "*.jpg"), recursive=True)
       + glob.glob(os.path.join(path, "**", "*.tif"), recursive=True))
print("图像文件数:", len(files))
img = Image.open(files[0]); print("尺寸/模式:", img.size, img.mode)

# torchvision Dataset 方式:
# from torchvision import datasets
# ds = datasets.ImageFolder(path)  # 要求 子目录=类别

目录组织与标注格式以源站说明和下载后实际文件为准。

数据说明

Background

The use of social media platforms in health research is increasing, yet their application in studying rare diseases is limited. Hodgkin’s lymphoma (HL) is a rare malignancy with a high incidence in young adults. This study evaluates the feasibility of using social media data to study the disease and treatment characteristics of HL.

Methods

We utilized the X (formerly Twitter) API v2 developer portal to download posts (formerly tweets) from January 2010 to October 2022. Annotation guidelines were developed from literature and a manual review of limited posts was performed to identify the class and attributes (characteristics) of HL discussed on X, and create a gold standard dataset. This dataset was subsequently employed to train, test, and validate a Named Entity Recognition (NER) Natural Language Processing (NLP) application.

Results

After data preparation, 80,811 posts were collected: 500 for annotation guideline development, 2,000 for NLP application development, and the remaining 78,311 for deploying the application. We identified nine classes related to HL, such as HL classification, etiopathology, stages and progression, and treatment. The treatment class and HL stages and progression were the most frequently discussed, with 20,013 (25.56%) posts mentioning HL’s treatments and 17,177 (21.93%) mentioning HL stages and progression. The model exhibited robust performance, achieving 86% accuracy and an 87% F1 score. The etiopathology class demonstrated excellent performance, with 93% accuracy and a 95% F1 score.

Discussion

The NLP application displayed high efficacy in extracting and characterizing HL-related information from social media posts, as evidenced by the high F1 score. Nonetheless, the data presented limitations in distinguishing between patients, providers, and caregivers and in establishing the temporal relationships between classes and attributes. Further research is necessary to bridge these gaps.

Conclusion

Our study demonstrated potential of using social media as a valuable preliminary research source for understanding the characteristics of rare diseases such as Hodgkin’s Lymphoma.

许可

CC BY 4.0

引用

学术使用请引用 DOI 10.1371/journal.pdig.0000765.t003（source: https://plos.figshare.com/articles/dataset/Frequency_of_extracted_class_and_its_attributes_/28626983）。