Robust Biomedical Publication Type and Study Design Classification with Knowledge-Guided Perturbations

2026-05-12

AI Summary

Existing biomedical literature classification models often rely on superficial lexical cues or dataset-specific artifacts under distribution shifts, limiting their generalization. This work proposes a knowledge-guided semantic perturbation framework that integrates entity masking with domain-adversarial training to selectively suppress non-defining thematic features and enhance reliance on critical methodological signals. By doing so, the approach overcomes the typical trade-off between in-distribution accuracy and out-of-distribution robustness, achieving high classification performance while significantly improving cross-distribution generalization. The effectiveness of feature-level robustness optimization is demonstrated on publication-type and study-design classification tasks, confirming that explicitly guiding models toward semantically meaningful cues yields more reliable and transferable representations.

Abstract

Accurately and consistently indexing biomedical literature by publication type and study design is essential for supporting evidence synthesis and knowledge discovery. Prior work on automated publication type and study design indexing has primarily focused on expanding label coverage, enriching feature representations, and improving in-domain accuracy, with evaluation typically conducted on data drawn from the same distribution as training. Although pretrained biomedical language models achieve strong performance under these settings, models optimized for in-domain accuracy may rely on superficial lexical or dataset-specific cues, resulting in reduced robustness under distributional shift. In this study, we introduce an evaluation framework based on controlled semantic perturbations to assess the robustness of a publication type classifier and investigate robustness-oriented training strategies that combine entity masking and domain-adversarial training to mitigate reliance on spurious topical correlations. Our results show that the commonly observed trade-off between robustness and in-domain accuracy can be mitigated when robustness objectives are designed to selectively suppress non-task-defining features while preserving salient methodological signals. We find that these improvements arise from two complementary mechanisms: (1) increased reliance on explicit methodological cues when such cues are present in the input, and (2) reduced reliance on spurious domain-specific topical features. These findings highlight the importance of feature-level robustness analysis for publication type and study design classification and suggest that refining masking and adversarial objectives to more selectively suppress topical information may further improve robustness. Data, code, and models are available at: https://github.com/ScienceNLP-Lab/MultiTagger-v2/tree/main/ICHI

Problem

• publication type classification
• study design classification
• distributional shift
• robustness
• biomedical literature

Innovation

• knowledge-guided perturbations
• robustness
• entity masking
• domain-adversarial training
• publication type classification