Dataset Structure

Interactive file-tree of the MIMIC-IV ECG dataset directory. Click any folder node to expand or collapse its children.

Directory & File Hierarchy
Root / Directory
CSV Metadata File (click to expand columns ▾)
WFDB Waveform File (.hea / .dat)
Other File
Column name (monospace, hover for type)

Dataset Summary

MIMIC-IV ECG Diagnostic Electrocardiogram Matched Subset v1.0 — Beth Israel Deaconess Medical Center

Record Coverage

How many records from the canonical list appear in each data source.

Records per Source

Temporal Distribution

ECG study counts over time. Note: MIMIC-IV dates are shifted for de-identification — use relative patterns only.

ECG Studies by Year
ECG Studies by Hour of Day
ECG Studies by Day of Week
ECG Studies by Month

Studies per Patient

Distribution of how many ECG studies each patient has.

Studies per Patient Histogram
Summary Statistics

Measurement Quality

Missing value rates and summary statistics for machine_measurements.csv columns.

Missing % per Column (all columns)
Numeric Field Summary Statistics

Interval Distributions

Histograms of RR, PR, QRS, and QT intervals derived from machine measurements.

RR Interval (ms)
PR Interval (ms)
QRS Duration (ms)
QT Interval (ms)

Electrical Axis Distributions

Distribution of P-wave, QRS, and T-wave electrical axes (degrees, −180 to +180).

P-wave Axis (degrees)
QRS Axis (degrees)
T-wave Axis (degrees)

Machine Report Phrases

Most frequent diagnostic phrases from machine-generated ECG reports (report_0 … report_17 columns).

Top report phrases

Carts & Equipment

ECG cart usage, bandwidth settings, and filter configurations.

Top ECG Carts by Study Count
Bandwidth Settings
Filter Settings

Waveform Header Stats

Properties extracted from a random sample of .hea header files.

Header Property Breakdown

Lead Completeness

How many records are missing each standard lead, based on full .hea file scan.

Per-Lead Absence Rate
Incomplete Records (first 50)

Signal Quality

Flat-line, dropout, and clipping rates detected from a sampled set of .dat waveform files.

Flat Lead Rate per Lead (%)
Dropout Rate per Lead (%)
Clipping Rate per Lead (%)
Flagged Records (first 200)