Evaluation of Reasoning Models for Musculoskeletal Physiotherapy Clinical Reasoning

Data Sheet 1 presents results from a mixed-methods pilot study evaluating reasoning models for clinical decision-making in musculoskeletal physiotherapy, specifically for low-back pain. The study was authored by Ricardo Knauer and uploaded to figshare in March 2026. It includes a human evaluation of large language models on dimensions like reliability, correctness, and patient-centeredness.

Use Cases

• Benchmarking language model performance on clinical reasoning tasks based on the described evaluation framework.
• Analyzing model weaknesses in logical coherence and patient-centeredness as identified in the qualitative analysis.
• Informing prompting strategies for clinical applications based on the study's guidance for enhancing model performance.
• Comparing test-retest reliability of different reasoning models as reported in the results.

Strengths

• Dataset is openly licensed under CC-BY-4.0, permitting broad reuse.
• Study employs a multidimensional evaluation framework assessing conceptual reasoning, completeness, correctness, relevance, and usefulness.
• File size is 700.1 KB, indicating a focused and manageable document for analysis.

Limitations

• Dataset is a single 700.1 KB DOCX file, indicating a very limited scope and likely a pilot study report rather than a primary data corpus.
• Row count and column-level documentation are absent; the data structure and variables must be inferred from the document.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Ricardo Knauer via figshare

Collection Method

Mixed-methods pilot study involving human evaluation of reasoning models.

Freshness

Last updated 2026-03-18 06:42:19