数据集获取

git clone https://github.com/kris96tian/machine_learning_ecg.git

curl -L -o repo.zip https://github.com/kris96tian/machine_learning_ecg/archive/refs/heads/main.zip
unzip repo.zip

源站 README 摘录（使用方式）

Diagnosing ‘silent’ heart attacks using ECG waveforms

Wb-App: https://huggingface.co/spaces/kral2796/heart-attack-prediction

Abstract

This project focuses on the early detection
of silent heart attacks, one of the most
pressing health concerns, through the
innovative application of machine learning
models to analyze electrocardiogram (ECG)
waveforms. Silent heart attacks are a type
of myocardial infarction that often go
undiagnosed due to their asymptomatic
nature, posing a significant health risk due
to the delay in intervention and treatment
which, consequently, leads to a higher risk
of heart failure. The comprehensive
Nightingale Open Science dataset, a
meticulously curated dataset that aligns
ECG waveforms with cardiac ultrasound
data, served as the primary data source for
this project. The primary objective of the
study was the identification of Regional Wall
Motion Abnormalities (RWMA), which are
key indicators of prior myocardial
infarctions. Various machine learning
nmethodologies, including the more
traditional Logistic Regression and Support
Vector Machine (SVM), as well as the
state-of-the-art Recurrent Neural Network
(RNN), were evaluated in this research. The
results demonstrated the immense potential
of machine learning models in the detection
of silent heart attacks through the analysis
of ECG waveforms. Among the models
evaluated, the RNN model stood out as it
outperformed the other models in detecting
RWMA due to its innate ability to capture
sequential and time-dependent data.
However, it is important to note that
challenges such as data imbalance and
overfitting were encountered during this
study, highlighting the need for further
refinement and optimization of the model.

References

Pramanik, Rajiv, Bhumil Shah, Anna Roth, Honga Wei, Ted Castillo, Katie Lin, Sachin Shah, et al.
“Diagnosing ’Silent’ Heart Attack Using ECG Waveforms.”
Nightingale Open Science, 2021. https://doi.org/10.48815/N54W2V.

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。

完整仓库：github.com/kris96tian/machine_learning_ecg