LLM-Delphi per-statement run logs (2,040 runs)

Complete per-statement run logs for the study "A Scalable and Auditable Multi-Model LLM-Delphi Framework for Reproducible Clinical Consensus Evaluation." The archive contains 2,040 statement-level runs (panel sizes 9, 17, and 25 × iterations IT1–IT5 × 136 statements), stored as JSON and Markdown (4,080 files). Unpack llm-delphi-run-logs.tar.gz to 9models/, 17models/, and 25models/. File naming and schema are documented in the companion software repository: https://github.com/appletea1011/LLM-delphi (software DOI: 10.5281/zenodo.21995773). The logs include the full text of clinical statements reproduced from Salminen et al., Obesity Surgery 2024;34:30–42, https://doi.org/10.1007/s11695-023-06913-8, licensed under CC BY 4.0. Statement wording is unchanged. API credentials are not stored in the logs.