mlsanityData to 7 Oct 2026

BadCLM: Backdoor Attack in Clinical Language Models for Electronic Health Records

arXiv 2407.05213 ↗First posted 6 Jul 2024cs.CL cs.AIcited by 5

On the map: LLM safety and security → Safety alignment, refusal and guardrails

Abstract (arXiv)

The advent of clinical language models integrated into electronic health records (EHR) for clinical decision support has marked a significant advancement, leveraging the depth of clinical notes for improved decision-making. Despite their success, the potential vulnerabilities of these models remain largely unexplored. This paper delves into the realm of backdoor attacks on clinical language models, introducing an innovative attention-based backdoor attack method, BadCLM (Bad Clinical Language Models). This technique clandestinely embeds a backdoor within the models, causing them to produce incorrect predictions when a pre-defined trigger is present in inputs, while functioning accurately otherwise. We demonstrate the efficacy of BadCLM through an in-hospital mortality prediction task with MIMIC III dataset, showcasing its potential to compromise model integrity. Our findings illuminate a significant security risk in clinical decision support systems and pave the way for future endeavors in fortifying clinical language models against such vulnerabilities.

Key facts. Cited by 5 papers in the map (data through 7 Oct 2026). Placed in Safety alignment, refusal and guardrails, one of 12 directions in LLM safety and security. Builds on 31 of its 55 references that are in the map. First posted on arXiv 6 Jul 2024.

Builds on

31 of 55 references.

References · 31

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  3. 2019HuggingFace's Transformers: State-of-the-art Natural Language Processingcited by 2,201
  4. 2019BioBERT: a pre-trained biomedical language representation model for biomedical text miningcited by 1,935
  5. 2020Don't Stop Pretraining: Adapt Language Models to Domains and Taskscited by 1,548
  6. 2017BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chaincited by 1,052
  7. 2019Publicly Available Clinical BERT Embeddingscited by 995
  8. 2017Multitask learning and benchmarking with clinical time series datacited by 287
  9. 2020MACER: Attack-free and Scalable Robust Training via Maximizing Certified Radiuscited by 114
  10. 2024Model Tailor: Mitigating Catastrophic Forgetting in Multi-modal Large Language Modelscited by 54
  11. 2022A Multimodal Transformer: Fusing Clinical Notes with Structured EHR Data for Interpretable In-Hospital Mortality Predictioncited by 48
  12. 2022A Study of the Attention Abnormality in Trojaned BERTscited by 47
19 more
  1. 2019Using Clinical Notes with Time Series Data for ICU Managementcited by 38
  2. 2023Knowledge Distillation on Graphs: A Surveycited by 38
  3. 2023Attention-Enhancing Backdoor Attacks Against BERT-based Modelscited by 32
  4. 2023NTK-SAP: Improving neural network pruning by aligning training dynamicscited by 30
  5. 2021How to Leverage Multimodal EHR Data for Better Medical Predictions?cited by 26
  6. 2024Task-Agnostic Detector for Insertion-Based Backdoor Attackscited by 23
  7. 2023Learning to Segment from Noisy Annotations: A Spatial Correction Approachcited by 16
  8. 2021TopoTxR: A Topological Biomarker for Predicting Treatment Response in Breast Cancercited by 13
  9. 2024Beyond Single-Event Extraction: Towards Efficient Document-Level Multi-Event Argument Extractioncited by 12
  10. 2023Robust Nonparametric Regression under Poisoning Attackcited by 11
  11. 2021DeepMTL Pro: Deep Learning Based MultipleTransmitter Localization and Power Estimationcited by 10
  12. 2019Transfer Learning for Scientific Data Chain Extraction in Small Chemical Corpus with BERT-CRF Modelcited by 8
  13. 2023Balanced Training for Sparse GANscited by 8
  14. 2022Quantifying Uncertainty In Traffic State Estimation Using Generative Adversarial Networkscited by 7
  15. 2023BadSAM: Exploring Security Vulnerabilities of SAM via Backdoor Attackscited by 7
  16. 2023Generalized Universal Domain Adaptation with Generative Flow Networkscited by 7
  17. 2023RGMComm: Return Gap Minimization via Discrete Communications in Multi-Agent Reinforcement Learningcited by 7
  18. 2020Multiscale Detection of Cancerous Tissue in High Resolution Slide Scanscited by 6
  19. 2022Attention Hijacking in Trojan Transformerscited by 6

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 1; 2025: 3; 2026: 1.

5 papers.

  1. 2024Backdooring Vision-Language Models with Out-Of-Distribution Datacited by 31
  2. 2025Backdoor Attack on Vision Language Models with Stealthy Semantic Manipulationcited by 13
  3. 2025Can Transformer Memory Be Corrupted? Investigating Cache-Side Vulnerabilities in Large Language Modelscited by 3
  4. 2025SCOUT: A Defense Against Data Poisoning Attacks in Fine-Tuned Language Modelscited by 1
  5. 2026Silent Sabotage During Fine-Tuning: Few-Shot Rationale Poisoning of Compact Medical LLMscited by 1