胸部X光AI性能差异中技术获取参数主导人口统计因素：数据和分析输出

数据集获取（Figshare）

• 源站标题：Data and analysis outputs for: Technical Acquisition Parameters Dominate Demographic Factors in Chest X-ray AI Performance Disparities
• DOI：10.6084/m9.figshare.31797865.v1
• 发布日期：2026-03-18
• 文件数：20 个
• 文件总体积：10.87 MB
• 源站页面：https://figshare.com/articles/dataset/Data_and_analysis_outputs_for_Technical_Acquisition_Parameters_Dominate_Demographic_Factors_in_Chest_X-ray_AI_Performance_Disparities/31797865

命令行下载

curl -L -o "analysis_anova_prediction_scores.csv" "https://ndownloader.figshare.com/files/62907520"
curl -L -o "analysis_cohens_d_all_comparisons.csv" "https://ndownloader.figshare.com/files/62907523"
curl -L -o "analysis_deviance_decomposition.csv" "https://ndownloader.figshare.com/files/62907526"
curl -L -o "analysis_intersectional_sensitivity.csv" "https://ndownloader.figshare.com/files/62907529"
curl -L -o "analysis_intersectional_summary.csv" "https://ndownloader.figshare.com/files/62907532"
curl -L -o "analysis_sensitivity_ranges.csv" "https://ndownloader.figshare.com/files/62907535"
curl -L -o "analysis_table1_by_view_type.csv" "https://ndownloader.figshare.com/files/62907538"
curl -L -o "analysis_validation_framework.csv" "https://ndownloader.figshare.com/files/62907544"

文件清单

• analysis_anova_prediction_scores.csv（1.26 KB）
• analysis_cohens_d_all_comparisons.csv（4.50 KB）
• analysis_deviance_decomposition.csv（2.84 KB）
• analysis_intersectional_sensitivity.csv（3.87 KB）
• analysis_intersectional_summary.csv（751 B）
• analysis_sensitivity_ranges.csv（1.68 KB）
• analysis_table1_by_view_type.csv（798 B）
• analysis_validation_framework.csv（1.05 KB）
• model_densenet121_res224_chex_predictions.csv（1.96 MB）
• model_densenet121_res224_chex_results_by_age.csv（342 B）
• model_densenet121_res224_nih_predictions.csv（1.97 MB）
• model_densenet121_res224_nih_results_by_age.csv（327 B）
• model_densenet121_res224_rsna_predictions.csv（1.97 MB）
• model_densenet121_res224_rsna_results_by_age.csv（343 B）
• model_densenet121-all_results_by_age.csv（330 B）
• model_densenet121-all_results_by_sex.csv（196 B）
• model_densenet121-all_rsna_predictions.csv（1.96 MB）
• model_padchest_predictions.csv（1.90 MB）
• README.txt（2.45 KB）
• view_type_lookup.csv（1.09 MB）

文件总体积 10.87 MB（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

Processed prediction data and analysis outputs accompanying the manuscript:
“Technical Acquisition Parameters Dominate Demographic Factors in Chest X-ray AI Performance Disparities: A Multi-Dataset External Validation Study” (PLOS Digital Health, PDIG-D-26-00087).
Contains per-image pneumonia prediction scores from five DenseNet-121 models evaluated on the RSNA Pneumonia Detection Challenge dataset (n=26,684), along with view type metadata (AP/PA) and revision analysis outputs including formal ANOVA decomposition, Cohen’'s d effect sizes, intersectional analysis, and validation framework classification.
Raw DICOM images are not included due to licensing; they are available from:

• RSNA: https://www.kaggle.com/c/rsna-pneumonia-detection-challenge
• NIH: https://nihcc.app.box.com/v/ChestXray-NIHCC
  Analysis code: https://doi.org/10.5281/zenodo.19081166
  GitHub: https://github.com/hayden-farquhar/AI-model-fairness-study

许可

CC BY 4.0

引用

学术使用请引用 DOI 10.6084/m9.figshare.31797865.v1（source: https://figshare.com/articles/dataset/Data_and_analysis_outputs_for_Technical_Acquisition_Parameters_Dominate_Demographic_Factors_in_Chest_X-ray_AI_Performance_Disparities/31797865）。