{
  "abstract": "The main objective of this paper is to propose a general framework for prediction based on different sources of multimodal data in the healthcare domain. We evaluated whether frozen medical large language model (LLM) representations can serve as a shared embedding space for multimodal primary diagnosis category prediction. We propose a backbone-flexible probing framework that serializes structured electronic health record (EHR) variables and leakage-pruned discharge-note sections into a shared frozen-LLM representation space and recovers primary diagnosis categories with linear probes, without fine-tuning the LLM. On a MIMIC-IV cohort of 13,645 admissions (seven diagnosis categories), we probed inputs of different modalities (Structured-only, Unstructured-only, and Combined) across transformer layers and six frozen backbones, compared against XGBoost and three text baselines (CAML, LAAT, PLM-ICD) under a matched split and context budget, and tested robustness across datasets on MIMIC-III. The Combined terminal-layer probe reached 88.86% top-1 accuracy on MIMIC-IV, significantly outperforming CAML and PLM-ICD and statistically comparable to LAAT; Combined accuracy ranged from 87.69% to 89.25% across six backbones. On MIMIC-III, in-domain Combined probing reached 74.18% accuracy and significantly exceeded all three baselines, and in a transfer-learning test a 2.8M-parameter adapter on frozen MIMIC-IV representations reached 92.51% accuracy, exceeding adapted LAAT (85.54%). Frozen LLM embeddings unify structured and narrative EHR information: the Combined configuration was strongest in every setting, and its representations transferred across datasets through a compact adapter. Multimodal probing of frozen LLM representations provides a practical, backbone-flexible approach for studying EHR modalities and adapting clinical representations across datasets.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2606.28798",
  "author_total": 4,
  "authors": [
    {
      "name": "Chengyuan Liu"
    },
    {
      "name": "Xinyue Zhang"
    },
    {
      "name": "Yao Li"
    },
    {
      "name": "Guanting Chen"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/4n3r64b1h38fg.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2606.28798",
  "fulltext_url": null,
  "id": "4n3r64b1h38fg",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2606.28798",
  "language": null,
  "licence": "LicenseRef-arXiv-nonexclusive-1.0",
  "licence_label": "arXiv licence",
  "licence_source": "article",
  "licence_url": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.AI"
    ],
    [
      "arxiv",
      "stat.AP"
    ],
    [
      "openalex",
      "1702|Machine Learning in Healthcare"
    ]
  ],
  "openalex": "W7166651509",
  "pdf_url": "https://arxiv.org/pdf/2606.28798",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-19",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/4n3r64b1h38fg.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:23:33Z",
      "native_id": "2606.28798",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2606.28798"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.AP",
      "subfield": "statistics"
    }
  ],
  "title": "A General Multimodal Probing Framework for Data Fusion and Prediction with Frozen Large Language Models: Evidence from Multimodal EHR Data",
  "title_text": "A General Multimodal Probing Framework for Data Fusion and Prediction with Frozen Large Language Models: Evidence from Multimodal EHR Data",
  "type": "preprint",
  "updated_at": "2026-10-30T12:23:33Z",
  "url": "https://research.draconitic.com/paper/4n3r64b1h38fg",
  "venue": "arXiv",
  "versions": []
}