人工临床推理肺炎诊断数据集
人工创建数据集，结合表格和文本数据，用于自动化诊断肺炎与上呼吸道感染。
基本信息
资源简介
人工临床推理肺炎诊断数据集
【数据集背景】
本数据集源站标题为《prabaey/bn-text》。源站首次发布：2024-03-13。
【数据内容】
源站原始描述：Code for "Clinical Reasoning over Tabular Data and Text with Bayesian Networks" # Clinical Reasoning over Tabular Data and Text with Bayesian Networks Summary This repository contains the code for Clinical Reasoning over Tabular Data and Text with Bayesian Networks. Should you reuse any ideas from this repository or the accompanying paper, please cite our work as follows: In this paper, we propose two approaches to integrate text in Bayesian networks, to enable joint inference of tabular data an…
【适用方向】
多模态数据适合跨模态对齐、融合建模与联合表征学习方向的研究。
【获取与许可】
源站页面：https://github.com/prabaey/bn-text
【使用注意】
· 源站最后修改于 2024-09-13，引用时请注意版本时效。
下载信息
注册下载
需要注册 Kaggle 账号并登录后下载，适合需要跟踪下载记录和使用 API 的用户。
暂未开放公开下载
无需注册即可直接获取公开样本或文档，适合快速预览和评估数据集质量。
免登录有偿下载
公开数据集受托下载与技术交付服务。
提供高速下载与技术交付服务（收技术服务费，非数据销售）
暂未开放当前数据集主要来源为 Kaggle 公开托管，完整影像包建议通过原始链接或 Kaggle API 获取。
使用方式
数据集获取
git clone https://github.com/prabaey/bn-text.git
curl -L -o repo.zip https://github.com/prabaey/bn-text/archive/refs/heads/main.zip
unzip repo.zip
源站 README 摘录（使用方式）
Clinical Reasoning over Tabular Data and Text with Bayesian Networks
Summary
This repository contains the code for Clinical Reasoning over Tabular Data and Text with Bayesian Networks. Should you reuse any ideas from this repository or the accompanying paper, please cite our work as follows:
@InProceedings{prabaey_clinical_reasoning,
author="Rabaey, Paloma
and Deleu, Johannes
and Heytens, Stefan
and Demeester, Thomas",
title="Clinical Reasoning over Tabular Data and Text with Bayesian Networks",
booktitle="Artificial Intelligence in Medicine",
year="2024",
publisher="Springer Nature Switzerland",
pages="229250",
}
In this paper, we propose two approaches to integrate text in Bayesian networks, to enable joint inference of tabular data and unstructured text. One approach uses a generative text model, while the other uses a discriminative one. We also compare with two types of baselines: a normal Bayesian network trained on only the tabular data and a feed-forward neural network.
<p float=“left”>
We evaluate our models using simulation results for a primary care use case (diagnosis of pneumonia), for which we generate our own dataset. This dataset is based on both expert knowledge (used to build a ground truth Bayesian network) and ChatGPT for generation of a realistic textual description that fits the tabular portion of the sample. An example of what the samples in this dataset look like is shown below.
<p float=“center”>
Files
notebooks:
data_generation.ipynb: illustration of the full simulation data generation processBN_baselines.ipynb: training and evaluation of two Bayesian network baselines ($BN$ and $BN^{++}$)FF_baseline.ipynb: training and evaluation of discriminative feed-forward neural network baseline ($\text{ff-discr-text}$)gen_model.ipynb: training and evaluation of Bayesian network with text generator $(\text{bn-gen-text})$, and its ablated version $(\text{bn-gen-text}^{-})$discr_model.ipynb: training and evaluation of Bayesian network with text discriminator $(\text{bn-discr-text})$, and its ablated version $(\text{bn-discr-text}^{-})$
data directory:
train_4000_final.p: simulated train dataset (4000 samples)test_1000_final.p: simulated test dataset (1000 samples)train_4000_final_fever_pain.p: version of train dataset with obse
数据加载示例（表格/文本类）
import pandas as pd, glob, os
files = (glob.glob(os.path.join(path, "**", "*.csv"), recursive=True)
+ glob.glob(os.path.join(path, "**", "*.tsv"), recursive=True)
+ glob.glob(os.path.join(path, "**", "*.xlsx"), recursive=True))
print("数据文件:", files)
df = pd.read_csv(files[0])
print(df.shape); print(df.columns.tolist()); print(df.head(3))
数据缺失？
依托陕西、四川两大基地，我们与超过十家三甲医院建立直接合作关系，覆盖合规授权、采集治理、专业标注、数据交付的全流程，为AI医疗团队提供即拿即用的高质量临床数据。
⚡️ 需要数据支持或标注服务？立即联系我们获取专业方案。
