{
  "abstract": "BackgroundMuch of the symptom burden in colorectal cancer (CRC) patients is documented in unstructured discharge-note narrative, and manual extraction is not scalable. Whether large language models (LLMs) outperform rule-based and named entity recognition (NER) methods has not been rigorously benchmarked.\n\nObjectiveTo benchmark rule-based, NER, and zero-shot LLM methods for extracting 46 cancer-related symptoms from CRC discharge notes against an adjudicated ground truth.\n\nMethodsWe analyzed 2,704 discharge notes from CRC patients in MIMIC-IV. A 46-symptom target list was built from the Memorial Symptom Assessment Scale and the EORTC QLQ-CR29. Four approaches -- dictionary-based rule matching, pretrained clinical NER, and zero-shot Claude Haiku and Gemini 3.5 Flash -- plus two hybrid variants (LLM output with post-hoc rule-based negation filtering) were evaluated against a 200-note gold standard adjudicated by two raters (pooled kappa=0.71, macro kappa=0.49), using Macro/Micro F1, precision, and recall.\n\nResultsGemini 3.5 Flash performed best (Macro F1=0.70, Micro F1=0.86, Macro Precision=0.74), followed by Claude Haiku (Macro F1=0.63, Macro Recall=0.71); both substantially outperformed rule-based (Macro F1=0.44) and NER (Macro F1=0.38) methods. Post-hoc negation filtering paradoxically degraded LLM performance (Gemini+Hybrid Macro F1=0.58; Claude+Hybrid Macro F1=0.54) by overriding correct predictions through rigid, fixed-window matching.\n\nConclusionsZero-shot LLMs substantially outperform rule-based and NER approaches for CRC symptom extraction; post-hoc negation correction should not be applied to LLM outputs without syntactic scope validation. Implications for Practice: Zero-shot LLM extraction offers a scalable, accurate alternative to manual chart review and traditional NLP pipelines for oncology symptom surveillance, without institution-specific rule development or model training.",
  "abstract_shown": true,
  "affiliations": [
    "University of Michigan"
  ],
  "article_type": "PUBLISHAHEADOFPRINT",
  "arxiv": null,
  "author_total": 4,
  "authors": [
    {
      "affs": [
        0
      ],
      "name": "Y. Lee"
    },
    {
      "name": "I. Dinov"
    },
    {
      "name": "X. Hu"
    },
    {
      "name": "Y. Jiang"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/awvdtqz4xtvmt.bib",
  "cited_by": 1,
  "doi": "10.64898/2026.09.15.26362961",
  "fulltext_url": "https://www.medrxiv.org/content/10.64898/2026.09.15.26362961v1.full",
  "id": "awvdtqz4xtvmt",
  "issns": [],
  "landing_url": "https://www.medrxiv.org/content/10.64898/2026.09.15.26362961v1",
  "language": null,
  "licence": "CC-BY-NC-ND",
  "licence_label": "CC BY-NC-ND",
  "licence_source": "article",
  "licence_url": "https://creativecommons.org/licenses/by-nc-nd/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "medrxiv",
      "health informatics"
    ],
    [
      "openalex",
      "1312|Biomedical Text Mining and Ontologies"
    ]
  ],
  "openalex": "W7213787178",
  "pdf_url": "https://www.medrxiv.org/content/10.64898/2026.09.15.26362961v1.full.pdf",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-15",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/awvdtqz4xtvmt.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:18:58Z",
      "native_id": "10.64898/2026.09.15.26362961",
      "source": "medrxiv",
      "url": "https://www.medrxiv.org/content/10.64898/2026.09.15.26362961v1"
    }
  ],
  "subjects": [
    {
      "field": "medicine",
      "origin": "medrxiv:health informatics",
      "subfield": "epidemiology-public-health"
    }
  ],
  "title": "Leveraging Large Language Models for Colorectal Cancer Symptom Extraction from MIMIC-IV Clinical Notes",
  "title_text": "Leveraging Large Language Models for Colorectal Cancer Symptom Extraction from MIMIC-IV Clinical Notes",
  "type": "preprint",
  "updated_at": "2026-10-30T12:18:58Z",
  "url": "https://research.draconitic.com/paper/awvdtqz4xtvmt",
  "venue": "medRxiv",
  "versions": []
}