mlsanityData to 14 Oct 2026

Hierarchical Dual-Strategy Unlearning for Biomedical and Healthcare Intelligence Using Imperfect and Privacy-Sensitive Medical Data

arXiv 2511.19498 ↗First posted 23 Nov 2025cs.LG cs.AI cs.CRcited by 0

On the map: LLM safety and security → Machine unlearning

Abstract (arXiv)

Large language models (LLMs) exhibit exceptional performance but pose substantial privacy risks due to training data memorization, particularly within healthcare contexts involving imperfect or privacy-sensitive patient information. We present a hierarchical dual-strategy framework for selective knowledge unlearning that precisely removes specialized knowledge while preserving fundamental medical competencies. Our approach synergistically integrates geometric-constrained gradient updates to selectively modulate target parameters with concept-aware token-level interventions that distinguish between preservation-critical and unlearning-targeted tokens via a unified four-level medical concept hierarchy. Comprehensive evaluations on the MedMCQA (surgical) and MHQA (anxiety, depression, trauma) datasets demonstrate superior performance, achieving an 82.7% forgetting rate and 88.5% knowledge preservation. Notably, our framework maintains robust privacy guarantees while requiring modification of only 0.1% of parameters, addressing critical needs for regulatory compliance, auditability, and ethical standards in clinical research.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Machine unlearning, one of 12 directions in LLM safety and security. Builds on 33 of its 55 references that are in the map. First posted on arXiv 23 Nov 2025.

Builds on

33 of 55 references.

Closest ancestors · 11

Same direction, newest first

  1. 2025AILS-NTUA at SemEval-2025 Task 4: Parameter-Efficient Unlearning for Large Language Models using Data Chunking4 Mar
  2. 2025A Comprehensive Survey of Machine Unlearning Techniques for Large Language Models22 Feb
  3. 2025MMUnlearner: Reformulating Multimodal Machine Unlearning in the Era of Multimodal Large Language Models16 Feb
  4. 2024Towards Robust Evaluation of Unlearning in LLMs via Data Transformations23 Nov
  5. 2024LLM Unlearning via Loss Adjustment with Only Forget Data14 Oct
  6. 2024Unified Gradient-Based Machine Unlearning with Remain Geometry Enhancement29 Sep
  7. 2024TOFU: A Task of Fictitious Unlearning for LLMs11 Jan
  8. 2023SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generation19 Oct
  9. 2023Tight Bounds for Machine Unlearning via Differential Privacy2 Sep
  10. 2023KGA: A General Machine Unlearning Framework Based on Knowledge Gap Alignment11 May
  11. 2022Knowledge Unlearning for Mitigating Privacy Risks in Language Models4 Oct

Other references · 22

  1. 2022Large Language Models Encode Clinical Knowledgecited by 1,830
  2. 2024DoRA: Weight-Decomposed Low-Rank Adaptationcited by 758
  3. 2024The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearningcited by 509
  4. 2022Deep Bidirectional Language-Knowledge Graph Pretrainingcited by 156
  5. 2022Position: Considerations for Differentially Private Learning with Large-Scale Public Pretrainingcited by 88
  6. 2023DP-OPT: Make Large Language Model Your Privacy-Preserving Prompt Engineercited by 57
  7. 2024Med42 -- Evaluating Fine-Tuning Strategies for Medical LLMs: Full-Parameter vs. Parameter-Efficient Approachescited by 52
  8. 2025Clinical ModernBERT: An efficient and long context encoder for biomedical textcited by 26
  9. 2025Tokens for Learning, Tokens for Unlearning: Mitigating Membership Inference Attacks in Large Language Models via Dual-Purpose Trainingcited by 15
  10. 2023FRAMU: Attention-based Machine Unlearning using Federated Reinforcement Learningcited by 13
  11. 2025ZeroTuning: Unlocking the Initial Token's Power to Enhance Large Language Models Without Trainingcited by 12
  12. 2025ProtoMol: Enhancing Molecular Property Prediction via Prototype-Guided Multimodal Learningcited by 11
10 more
  1. 2025Read Before You Think: Mitigating LLM Comprehension Failures with Step-by-Step Readingcited by 10
  2. 2024Attack-Aware Noise Calibration for Differential Privacycited by 8
  3. 2025MvHo-IB: Multi-View Higher-Order Information Bottleneck for Brain Disorder Diagnosiscited by 8
  4. 2025MHQA: A Diverse, Knowledge Intensive Mental Health Question Answering Challenge for Language Modelscited by 7
  5. 2024Vertical Federated Learning with Missing Features During Training and Inferencecited by 6
  6. 2025Can LLMs Handle WebShell Detection? Overcoming Detection Challenges with Behavioral Function-Aware Frameworkcited by 6
  7. 2024The Limits of Differential Privacy in Online Learningcited by 5
  8. 2023Prior-itizing Privacy: A Bayesian Approach to Setting the Privacy Budget in Differential Privacycited by 2
  9. 2025Rep-GLS: Report-Guided Generalized Label Smoothing for Robust Disease Detectioncited by 1
  10. 2025EVA-Net: Interpretable Anomaly Detection for Brain Health via Learning Continuous Aging Prototypes from One-Class EEG Cohortscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Hierarchy-Aware Multimodal Unlearning for Medical AI
  2. 2025From Learning to Unlearning: Biomedical Security Protection in Multimodal Large Language Models
  3. 2026AMNESIA: A Large Scale Medical Unlearning Benchmark Suite with Disease-Informed Analysis
  4. 2026GRIN+: Towards Fast Yet Effective Machine Unlearning for Imbalanced Medical Data
  5. 2025How to Train Private Clinical Language Models: A Comparative Study of Privacy-Preserving Pipelines for ICD-9 Coding
  6. 2026DISCO-TAB: A Hierarchical Reinforcement Learning Framework for Privacy-Preserving Synthesis of Complex Clinical Data
  7. 2024MedCare: Advancing Medical LLMs through Decoupling Clinical Alignment and Knowledge Aggregation
  8. 2025Leverage Unlearning to Sanitize LLMs