BMD-HS数据集
包含超过800个心音录音，分为6类瓣膜疾病和健康样本，用于自动化心血管疾病诊断。
基本信息
资源简介
BMD-HS数据集
【数据集背景】
本数据集源站标题为《sani002/BMD-HS-Dataset》。源站首次发布：2024-08-18。
【数据内容】
源站原始描述：Organised data # BMD-HS Dataset: Heart Sound Recordings for Automated Cardiovascular Disease Diagnosis The BMD-HS dataset is a groundbreaking collection of heart sound recordings, meticulously curated to enhance automated cardiovascular disease (CVD) diagnosis. This dataset contains over 800 recordings, classified into six categories, including common valvular diseases: Aortic Stenosis (AS), Aortic Regurgitation (AR), Mitral Regurgitation (MR), Multi Disease (MD), and Mitral Stenosis (MS), along…
【适用方向】
音频数据可用于声学信号分析建模，如语音与呼吸音相关研究。
【获取与许可】
源站页面：https://github.com/sani002/BMD-HS-Dataset
使用许可：NOASSERTION。
【使用注意】
· 源站最后修改于 2026-03-19，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/sani002/BMD-HS-Dataset.git
curl -L -o repo.zip https://github.com/sani002/BMD-HS-Dataset/archive/refs/heads/main.zip
unzip repo.zip
源站 README 摘录（使用方式）
BMD-HS Dataset: Heart Sound Recordings for Automated Cardiovascular Disease Diagnosis
The BMD-HS dataset is a groundbreaking collection of heart sound recordings, meticulously curated to enhance automated cardiovascular disease (CVD) diagnosis. This dataset contains over 800 recordings, classified into six categories, including common valvular diseases: Aortic Stenosis (AS), Aortic Regurgitation (AR), Mitral Regurgitation (MR), Multi Disease (MD), and Mitral Stenosis (MS), along with healthy (Normal) samples.
Key Features:
- Multi-label annotations: Allows nuanced classification by capturing unique disease states, including both single and multi-valvular diseases.
- De-identified metadata: Includes non-identifying demographic and clinical metadata to provide additional diagnostic context.
- Diverse demographic representation: Recorded at the National Institute of Cardiovascular Disease, Dhaka, the dataset includes a gender-balanced collection of heart sounds, ensuring its relevance for healthcare in Bangladesh and similar regions.
- Balanced class representation: Recordings were collected from 20 healthy subjects and 20 subjects for each valvular disease class, addressing class imbalance issues.
- Rich metadata: Annotations include disease presence, severity, and de-identified demographic information (e.g., age bands), enabling in-depth research and potential discovery of new correlations.
- Multi-disease data: Includes patients with multiple valvular diseases, offering a complex dataset that reflects real-world scenarios where patients often suffer from more than one cardiovascular condition.
Dataset Structure
1. Train Folder
- Files: Contains 872
.wavaudio files. - Details: Recordings were taken from 59 patients in 8 different positions, with each recording lasting 20 seconds. The recordings were sampled at a frequency of 4 kHz.
2. Train.csv
- Purpose: Contains training labels and corresponding recording file names for each patient.
- Columns:
- patient_id: File names in the train folder.
- AS: Label for Aortic Stenosis (0 = absent, 1 = present).
- AR: Label for Aortic Regurgitation (0 = absent, 1 = present).
- MR: Label for Mitral Regurgitation (0 = absent, 1 = present).
- MS: Label for Mitral Stenosis (0 = absent, 1 = present).
- MD: Label for multi-disease pa
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。
