PrivMedChat: End-to-End Differentially Private RLHF for Medical Dialogue Systems

• URL: http://arxiv.org/abs/2603.03054v1
• Date: Tue, 03 Mar 2026 14:53:20 GMT
• Title: PrivMedChat: End-to-End Differentially Private RLHF for Medical Dialogue Systems
• Authors: Sudip Bhujel,
• Abstract summary: PrivMedChat is an end-to-end framework for differentially private RLHF.<n>We present PrivMedChat, an end-to-end framework for differentially private RLHF.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models are increasingly used for patient-facing medical assistance and clinical decision support, but adapting them to clinical dialogue often requires supervision derived from doctor-patient conversations that may contain sensitive information. Conventional supervised fine-tuning and reinforcement learning from human feedback (RLHF) can amplify memorization risks, enabling empirical membership inference and extraction of rare training-set content. We present PrivMedChat, an end-to-end framework for differentially private RLHF (DP-RLHF) for medical dialogue. Our design enforces differential privacy at every training stage that directly accesses dialogue-derived supervision: (i) Differential Private Stochastic Gradient Descent (DP-SGD) for medical SFT and (ii) DP-SGD for reward model learning from preference pairs. To limit additional privacy expenditure during alignment, we apply DP-SGD to the PPO actor and critic when operating on dialogue-derived prompts, while the reward model remains fixed after DP training. We also introduce an annotation-free preference construction strategy that pairs physician responses with filtered non-expert generations to produce scalable preference data without clinician labeling. Experiments on medical dialogue benchmarks show that PrivMedChat at $\varepsilon=7$ achieves the highest ROUGE-L of 0.156 among all DP models, reduces clinical hallucinations to 1.4% and harmful advice to 0.4%, and obtains the highest overall score of 2.86 in a 3-model LLM-jury evaluation, while producing membership-inference signals that are near chance (AUC 0.510-0.555). We open-source our code at https://github.com/sudip-bhujel/privmedchat.

Related papers

• A Federated and Parameter-Efficient Framework for Large Language Model Training in Medicine [59.78991974851707]
  Large language models (LLMs) have demonstrated strong performance on medical benchmarks, including question answering and diagnosis.<n>Most medical LLMs are trained on data from a single institution, which faces limitations in generalizability and safety in heterogeneous systems.<n>We introduce the model-agnostic and parameter-efficient federated learning framework for adapting LLMs to medical applications.
  arXiv  Detail & Related papers  (2026-01-29T18:48:21Z)
• Note2Chat: Improving LLMs for Multi-Turn Clinical History Taking Using Medical Notes [17.99778043736069]
  We propose a note-driven framework that trains LLMs to conduct structured history taking and diagnosis by learning from medical notes.<n>We convert real-world medical notes into high-quality doctor-patient dialogues using a decision tree-guided generation and refinement pipeline.<n>We also propose a novel single-turn reasoning paradigm that reframes history taking as a sequence of single-turn reasoning problems.
  arXiv  Detail & Related papers  (2026-01-29T11:05:46Z)
• How to Train Private Clinical Language Models: A Comparative Study of Privacy-Preserving Pipelines for ICD-9 Coding [0.33148826359547523]
  Large language models trained on clinical text risk exposing sensitive patient information.<n>Despite rapid progress in DP optimisation, it remains unclear which privacy-preserving strategy actually works best.<n>Knowledge distillation from DP-trained teachers outperforms both direct DP-SGD and DP-synthetic data training.
  arXiv  Detail & Related papers  (2025-11-18T21:51:04Z)
• Differential privacy enables fair and accurate AI-based analysis of speech disorders while protecting patient data [10.6135892856374]
  This study is the first to investigate differential privacy in pathological speech data.<n>It focuses on the trade-offs between privacy, diagnostic accuracy, and fairness.<n>Our results establish that DP can balance privacy and utility in speech disorder detection.
  arXiv  Detail & Related papers  (2024-09-27T18:25:54Z)
• RuleAlign: Making Large Language Models Better Physicians with Diagnostic Rule Alignment [54.91736546490813]
  We introduce the RuleAlign framework, designed to align Large Language Models with specific diagnostic rules. We develop a medical dialogue dataset comprising rule-based communications between patients and physicians. Experimental results demonstrate the effectiveness of the proposed approach.
  arXiv  Detail & Related papers  (2024-08-22T17:44:40Z)
• Dr-LLaVA: Visual Instruction Tuning with Symbolic Clinical Grounding [53.629132242389716]
  Vision-Language Models (VLM) can support clinicians by analyzing medical images and engaging in natural language interactions. VLMs often exhibit "hallucinogenic" behavior, generating textual outputs not grounded in contextual multimodal information. We propose a new alignment algorithm that uses symbolic representations of clinical reasoning to ground VLMs in medical knowledge.
  arXiv  Detail & Related papers  (2024-05-29T23:19:28Z)
• Towards Adapting Open-Source Large Language Models for Expert-Level Clinical Note Generation [19.08691249610632]
  This study presents a comprehensive domain- and task-specific adaptation process for the open-source LLaMA-2 13 billion parameter model.<n>Our process incorporates continued pretraining, supervised fine-tuning, and reinforcement learning from both AI and human feedback.<n>Our resulting model, LLaMA-Clinic, can generate clinical notes comparable in quality to those authored by physicians.
  arXiv  Detail & Related papers  (2024-04-25T15:34:53Z)
• Preserving privacy in domain transfer of medical AI models comes at no performance costs: The integral role of differential privacy [5.025818976218807]
  We evaluate the efficacy of DP-enhanced domain transfer (DP-DT) in diagnosing cardiomegaly, pleural effusion, pneumonia, atelectasis, and in identifying healthy subjects. Our results show that DP-DT, even with exceptionally high privacy levels, performs comparably to non-DP-DT.
  arXiv  Detail & Related papers  (2023-06-10T18:41:50Z)
• An Experimental Study on Private Aggregation of Teacher Ensemble Learning for End-to-End Speech Recognition [51.232523987916636]
  Differential privacy (DP) is one data protection avenue to safeguard user information used for training deep models by imposing noisy distortion on privacy data. In this work, we extend PATE learning to work with dynamic patterns, namely speech, and perform one very first experimental study on ASR to avoid acoustic data leakage.
  arXiv  Detail & Related papers  (2022-10-11T16:55:54Z)
• Semi-Supervised Variational Reasoning for Medical Dialogue Generation [70.838542865384]
  Two key characteristics are relevant for medical dialogue generation: patient states and physician actions. We propose an end-to-end variational reasoning approach to medical dialogue generation. A physician policy network composed of an action-classifier and two reasoning detectors is proposed for augmented reasoning ability.
  arXiv  Detail & Related papers  (2021-05-13T04:14:35Z)
• DeepEnroll: Patient-Trial Matching with Deep Embedding and Entailment Prediction [67.91606509226132]
  Clinical trials are essential for drug development but often suffer from expensive, inaccurate and insufficient patient recruitment. DeepEnroll is a cross-modal inference learning model to jointly encode enrollment criteria (tabular data) into a shared latent space for matching inference.
  arXiv  Detail & Related papers  (2020-01-22T17:51:25Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.