使用链接电子健康记录报告医疗相关感染

数据集获取（Figshare）

• 源站标题：Using linked electronic health records to report healthcare-associated infections
• DOI：10.1371/journal.pone.0206860
• 发布日期：2018-11-07
• 文件数：3 个
• 文件总体积：55.95 KB
• 源站页面：https://plos.figshare.com/articles/dataset/Using_linked_electronic_health_records_to_report_healthcare-associated_infections/7311155

命令行下载

curl -L -o "pone.0206860.s001.docx" "https://ndownloader.figshare.com/files/13505768"
curl -L -o "pone.0206860.s002.docx" "https://ndownloader.figshare.com/files/13505771"
curl -L -o "pone.0206860.s003.xlsx" "https://ndownloader.figshare.com/files/13505774"

文件清单

• pone.0206860.s001.docx（14.06 KB）
• pone.0206860.s002.docx（16.31 KB）
• pone.0206860.s003.xlsx（25.58 KB）

文件总体积 55.95 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
Reporting of strategic healthcare-associated infections (HCAIs) to Public Health England is mandatory for all acute hospital trusts in England, via a web-based HCAI Data Capture System (HCAI-DCS).
Aim
Investigate the feasibility of automating the current, manual, HCAI reporting using linked electronic health records (linked-EHR), and assess its level of accuracy.
Methods
All data previously submitted through the HCAI-DCS by the Oxford University Hospitals infection control (IC) team for methicillin-resistant and methicillin-susceptible Staphylococcus aureus (MRSA, MSSA), Clostridium difficile, and Escherichia coli, through March 2017 were downloaded and compared to outputs created from linked-EHR, with detailed comparisons between 2013–2017.
Findings
Total MRSA, MSSA, E. coli and C. difficile cases entered by the IC team vs linked-EHR were 428 vs 432, 795 vs 816, 2454 vs 2450 and 3365 vs 3393 respectively. From 2013–2017, most discrepancies (32/37 (86%)) were likely due to IC recording errors. Patient and specimen identifiers were completed for >98% of cases by both methods, with very high agreement (>97%). Fields relating to the patient at the time the specimen was taken were complete to a similarly high level (>99% IC, >97% linked-EHR), and agreement was fairly good (>80%) except for the main and treatment specialties (57% and 54% respectively) and the patient category (55%). Optional, organism-specific data-fields were less complete, by both methods. Where comparisons were possible, agreement was reasonably high (mostly 70–90%).
Conclusion
Basic factual information, such as demographic data, is almost-certainly better automated, and many other data fields can potentially be populated successfully from linked-EHR. Manual data collection is time-consuming and inefficient; automated electronic data collection would leave healthcare professionals free to focus on clinical rather than administrative work.

许可

CC BY 4.0

引用

学术使用请引用 DOI 10.1371/journal.pone.0206860（source: https://plos.figshare.com/articles/dataset/Using_linked_electronic_health_records_to_report_healthcare-associated_infections/7311155）。