{
  "id": "2505.05763",
  "version": "v1",
  "published": "2025-05-09T03:53:10.000Z",
  "updated": "2025-05-09T03:53:10.000Z",
  "title": "BMMDetect: A Multimodal Deep Learning Framework for Comprehensive Biomedical Misconduct Detection",
  "authors": [
    "Yize Zhou",
    "Jie Zhang",
    "Meijie Wang",
    "Lun Yu"
  ],
  "categories": [
    "cs.LG",
    "cs.CL"
  ],
  "abstract": "Academic misconduct detection in biomedical research remains challenging due to algorithmic narrowness in existing methods and fragmented analytical pipelines. We present BMMDetect, a multimodal deep learning framework that integrates journal metadata (SJR, institutional data), semantic embeddings (PubMedBERT), and GPT-4o-mined textual attributes (methodological statistics, data anomalies) for holistic manuscript evaluation. Key innovations include: (1) multimodal fusion of domain-specific features to reduce detection bias; (2) quantitative evaluation of feature importance, identifying journal authority metrics (e.g., SJR-index) and textual anomalies (e.g., statistical outliers) as dominant predictors; and (3) the BioMCD dataset, a large-scale benchmark with 13,160 retracted articles and 53,411 controls. BMMDetect achieves 74.33% AUC, outperforming single-modality baselines by 8.6%, and demonstrates transferability across biomedical subfields. This work advances scalable, interpretable tools for safeguarding research integrity.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2025-05-09T03:53:10.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "multimodal deep learning framework",
      "comprehensive biomedical misconduct detection",
      "identifying journal authority metrics",
      "reduce detection bias",
      "academic misconduct detection"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 0,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}