ECG-Grounding is a dataset created by LANSG to provide evidence-driven interpretations of electrocardiograms. It contains 30,000 instruction pairs annotated with heartbeat-level physiological features, aiming to improve the trustworthiness of medical AI. The dataset was last updated on December 11, —.
Use Cases
- Training models for evidence-based ECG diagnosis based on annotated physiological features.
- Developing AI systems that ground diagnoses in measurable heartbeat-level features.
- Improving the interpretability and trustworthiness of medical AI for cardiology.
- Benchmarking models on high-granularity ECG grounding tasks.
Strengths
- Contains 30,000 instruction pairs.
- Provides heartbeat-level physiological feature annotations.
- Described as the first high-granularity ECG grounding dataset.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
- Data may reflect temporal or source bias inherent to its collection method.
Provenance
- Source
- LANSG
- Collection Method
- Annotated instruction pairs.
- Time Range
- null
- Freshness
- Last updated 2025-12-11 03:40:36.
- Geography
- null