SynthClinicalNotes: Synthetic Inpatient Trajectories for Progress Note Generation

Intel Labs created a fully synthetic, privacy-safe benchmark dataset for evaluating long-context clinical reasoning. The dataset contains multi-day inpatient trajectories, including labs, vitals, and daily progress notes. It was released in version v1.0 and last updated on May 15, -2025.

Use Cases

• Benchmarking long-context language models based on the described multi-day inpatient trajectories.
• Evaluating clinical progress note generation models based on the synthetic labs and vitals data.
• Training privacy-safe medical AI models based on the fully synthetic nature of the dataset.
• Researching clinical reasoning tasks based on the structured trajectory and note data.

Strengths

• Dataset is fully synthetic, designed to be privacy-safe.
• Created by Intel Labs, a known research organization.
• Specifically designed for benchmarking long-context clinical tasks.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and sample data are unknown, which may limit suitability assessment.
• License details are not fully specified in the provided metadata.

Provenance

Source

Intel Labs

Collection Method

Synthetically generated.

Time Range

null

Freshness

Last updated 2025-05-15 22:14:12.

Geography

null