Dataset License
This dataset is licensed under the Open Data Commons Open Database License v1.0 (ODbL 1.0 license).
Further details:
ODbL 1.0
This dataset is derived from the MIMIC III Waveform Database:
Moody, B., Moody, G., Villarroel, M., Clifford, G. D., & Silva, I. (2020). MIMIC-III Waveform Database (version 1.0). PhysioNet.
https://doi.org/10.13026/c2607m
The MIMIC III Waveform Database is licensed under the ODbL 1.0 license.
The MIMIC-III database is described in:
Johnson, A. E. W., Pollard, T. J., Shen, L., Lehman, L. H., Feng, M., Ghassemi, M., Moody, B., Szolovits, P., Celi, L. A., & Mark, R. G. (2016). MIMIC-III, a freely accessible critical care database. Scientific Data, 3, 160035.
https://doi.org/10.1038/sdata.2016.35
It is available on PhysioNet:
https://physionet.org/
Goldberger, A., Amaral, L., Glass, L., Hausdorff, J., Ivanov, P. C., Mark, R., ... & Stanley, H. E. (2000). PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals. Circulation [Online]. 101 (23), pp. e215–e220.
The following annotations of AF and non-AF were used to create the dataset:
Bashar, Syed Khairul (2020): Atrial Fibrillation annotations of electrocardiogram from MIMIC III matched subset. figshare. Dataset.
https://doi.org/10.6084/m9.figshare.12149091.v1
Bashar, S.K., Ding, E., Walkey, A.J., McManus, D.D. and Chon, K.H., 2019. Noise Detection in Electrocardiogram Signals for Intensive Care Unit Patients. IEEE Access, 7, pp.88357-88368.
https://doi.org/10.1109/ACCESS.2019.2926199
This annotation information is reproduced under the terms of the
CC BY 4.0 licence
""",elem_id=None)
with gr.Tab("Analyze Dataset"):
info_ds = gr.Textbox(label="Dataset Info", interactive=False, lines=5)
preview_ds = gr.Dataframe(label="Data Preview", interactive=False)
plot_ds = gr.Plot(label="ECG Signal Plot")
btn_ds = gr.Button("Analyze Dataset")
def analyze_dataset(file_path):
import pandas as pd
import matplotlib.pyplot as plt
if file_path is None:
return "No file uploaded. Please upload a CSV file first.", None, None
try:
df = pd.read_csv(file_path)
except Exception as e:
return f"Failed to read file. Make sure the file is a valid CSV. Error: {e}", None, None
info_lines = []
info_lines.append(f"Shape: {df.shape}")
info_lines.append(f"Columns: {list(df.columns)}")
info_lines.append(f"Missing: {df.isnull().sum().to_dict()}")
duration = None
sampling_rate = 125
n_samples = len(df)
duration = n_samples / sampling_rate
info_lines.append(f"Sampling rate: {sampling_rate} Hz")
info_lines.append(f"Data duration: {duration:.2f} seconds ({duration/60:.2f} minutes)")
preview = df.head(10)
fig = None
ecg_col = None
for col in df.columns:
if 'ecg' in col.lower():
ecg_col = col
break
if ecg_col is None:
return "The uploaded CSV does not contain an 'ecg' column. Please upload a CSV file with an 'ecg' feature/column.", preview, None
if not pd.api.types.is_numeric_dtype(df[ecg_col]):
return f"The selected signal column ('{ecg_col}') is not numeric. Please upload a valid ECG CSV.", preview, None
plot_samples = min(sampling_rate*10, len(df))
try:
fig, ax = plt.subplots()
ax.plot(df[ecg_col].values[:plot_samples])
ax.set_title(f"First 10 Seconds Signal Plot: {ecg_col}")
ax.set_xlabel("Sample")
ax.set_ylabel("Amplitude")
except Exception as e:
return f"Failed to plot ECG signal: {e}", preview, None
return "\n".join(info_lines), preview, fig
btn_ds.click(analyze_dataset, inputs=[shared_file_path], outputs=[info_ds, preview_ds, plot_ds])
with gr.Tab("Analyze Model"):
all_model_labels = [m[2] for m in MODEL_PAIRS] + [s[0] for s in SINGLE_MODELS]
pilih_model = gr.Dropdown(all_model_labels, label="Select Model", value=all_model_labels[0])
hasil = gr.Textbox(label="Prediction Result", interactive=False)
tombol = gr.Button("Predict")
def handle_predict(file_path, model_label):
if file_path is None or str(file_path).strip() == "":
return "No file uploaded. Please upload a CSV file first."
if not model_label:
return "No model selected. Please select a model."
try:
import pandas as pd
df = pd.read_csv(file_path)
ecg_col = None
for col in df.columns:
if 'ecg' in col.lower():
ecg_col = col
break
if ecg_col is None:
return "The uploaded CSV does not contain an 'ecg' column. Please upload a CSV file with an 'ecg' feature/column for prediction."
if not pd.api.types.is_numeric_dtype(df[ecg_col]):
return f"The selected signal column ('{ecg_col}') is not numeric. Please upload a valid ECG CSV."
class DummyFile:
def __init__(self, name):
self.name = name
dummy_file = DummyFile(file_path)
result, _ = analisis_sinyal(dummy_file, model_label)
if str(result).strip() == '0':
return 'Non-AF'
elif str(result).strip() == '1':
return 'AF'
else:
return str(result)
except Exception as e:
return f"Prediction failed: {e}"
tombol.click(handle_predict, inputs=[shared_file_path, pilih_model], outputs=[hasil])
def clear_file():
return None, None, None
clear_btn.click(clear_file, inputs=None, outputs=[file_upload, shared_file_path, hasil])
demo.launch(css=css)