VERI-DPO: Evidence-Aware Alignment for Clinical Summarization via Claim Verification and Direct Preference Optimization

• URL: http://arxiv.org/abs/2603.10494v1
• Date: Wed, 11 Mar 2026 07:41:58 GMT
• Title: VERI-DPO: Evidence-Aware Alignment for Clinical Summarization via Claim Verification and Direct Preference Optimization
• Abstract summary: We introduce VERI-DPO, which uses claim verification to mine preferences and distill them into the summarizer.<n>On MIMIC-III-Ext-VeriFact-BHC (100 ICU patients; patient-level splits), we train a retrieval-augmented verifier to label claim-evidence pairs as Supported, Not Supported, or Not Addressed.<n>On held-out patients, verifier-mined preferences separate candidates by contradiction density, and VERI-DPO reduces Not Supported claim rates from 10.7% to 1.9%.
• Score: 12.675974555403366
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Brief Hospital Course (BHC) narratives must be clinically useful yet faithful to fragmented EHR evidence. LLM-based clinical summarizers still introduce unsupported statements, and alignment can encourage omissions ("say-less" degeneration). We introduce VERI-DPO, which uses claim verification to mine preferences and distill them into the summarizer with Direct Preference Optimization (DPO). On MIMIC-III-Ext-VeriFact-BHC (100 ICU patients; patient-level splits), we train a retrieval-augmented verifier to label claim-evidence pairs as Supported, Not Supported, or Not Addressed via a single-token format. The verifier scores sentence-level claims from sampled BHC candidates and aggregates margins into a coverage-aware utility to mine length-controlled, contradiction-anchored preference pairs. On held-out patients, verifier-mined preferences separate candidates by contradiction density, and VERI-DPO reduces Not Supported claim rates from 10.7% to 1.9% (local verifier judge) and from 11.6% to 6.4% (GPT-4o judge), while improving validity from 76.7% to 82.5% and maintaining informative length.

Related papers

• Symmetry-Aware Likelihood-Orbit Aggregation for Selective Left-Right Claim Verification [14.108900142001628]
  We introduce Relation-Orbit, a closed-form contrast that assigns eight normalized likelihoods to query-supporting and counterfactual roles.<n>On VSR and GQA across four frozen vision-language models, Relation-Orbit yields higher mean test coverage at a 10% selective-risk calibration target.
  arXiv  Detail & Related papers  (2026-09-15T11:19:03Z)
• DiaWhisper-DPO: Role-Attributed Transcription of Clinical Interviews via Failure-Mined Preference Optimization [6.329932257779467]
  Cascaded systems combine speaker diarization with roleassignments, so errors can propagate across stages.<n>We propose an end-to-end model, which we named DiaWhisper, that achieves that fine-tunes Whisper-large-v3 with LoRA and an auxiliary frame-level role head for transcription and attribution.<n>On 29 DAIC-WOZ test sessions, DiaWhisper-DPO 0.973 role accuracy and 0.119 DER, 72% below the strongest cascaded baseline, and reduces seed variation from =.205 to.002.
  arXiv  Detail & Related papers  (2026-09-15T05:32:27Z)
• CliniCARE-Bench: Clinical Calibrated Audit of Medical Reasoning in EHR [15.66578453195388]
  CliniCARE-Bench is a benchmark for retrospective clinical audit.<n>25 clinician-validated scenarios instantiated as 750 patient-specific cases over real-patient-derived MIMIC-IV data.<n>System returns one of four verdicts: Yes, No, Indeterminate: Lack of Data, or Indeterminate: Medically Ambiguous.
  arXiv  Detail & Related papers  (2026-08-07T22:39:13Z)
• Teaching LLMs to Recommend and Defer in Underrepresented Epilepsy Care [5.027305497521]
  We introduce MANANA, a non-trivial prompt-learning framework that learns local prescribing guidance from a small patient-level training set.<n>Manana converts observed prescription errors into auditable prompt memories, instantiated in single-agent and multi-agent variants, and improves over classical ML models.
  arXiv  Detail & Related papers  (2026-06-30T02:09:37Z)
• Guideline-Grounded Evidence Accumulation for High-Stakes Agent Verification [60.18369393468405]
  Existing verifiers usually underperform owing to a lack of domain knowledge and limited calibration.<n>GLEAN compiles expert-curated protocols into trajectory-informed, well-calibrated correctness signals.<n>We empirically validate GLEAN with agentic clinical diagnosis across three diseases from the MIMIC-IV dataset.
  arXiv  Detail & Related papers  (2026-03-03T09:36:43Z)
• Distill and Align Decomposition for Enhanced Claim Verification [51.93960785128124]
  Complex claim verification requires decomposing sentences into verifiable subclaims.<n>We propose a reinforcement learning approach that optimises decomposition quality and verifier alignment.<n>Our framework enables smaller language models to achieve state-of-the-art claim verification.
  arXiv  Detail & Related papers  (2026-02-25T12:32:04Z)
• A Multi-Agent Framework for Medical AI: Leveraging Fine-Tuned GPT, LLaMA, and DeepSeek R1 for Evidence-Based and Bias-Aware Clinical Query Processing [0.4349324020366305]
  Large language models (LLMs) show promise for healthcare question answering, but clinical use is limited by weak verification, insufficient evidence grounding, and unreliable confidence signalling.<n>We propose a multi-agent medical QA framework that combines complementary LLMs with evidence retrieval, uncertainty estimation, and bias checks to improve answer reliability.
  arXiv  Detail & Related papers  (2026-02-15T14:17:27Z)
• Token-Level Inference-Time Alignment for Vision-Language Models [58.41370989069588]
  Vision-Language Models (VLMs) have become essential backbones of modern multimodal intelligence.<n>We present TITA, a lightweight framework that freezes the base VLM and instead trains a reward model to approximate its distribution.<n>During inference, implicit preference signals are extracted as log-probability ratios between the reward model and the target VLM, yielding dense autoregressive feedback.
  arXiv  Detail & Related papers  (2025-10-20T09:58:03Z)
• Aligning Language Models with Clinical Expertise: DPO for Heart Failure Nursing Documentation in Critical Care [4.108872110731109]
  This study applies Direct Preference Optimization to adapt Mistral-7B, a locally deployable language model, using 8,838 heart failure nursing notes.<n> Evaluation across BLEU, ROUGE, BERTScore, Perplexity, and expert qualitative assessments demonstrates that DPO markedly enhances documentation quality.
  arXiv  Detail & Related papers  (2025-10-06T22:04:37Z)
• ELSPR: Evaluator LLM Training Data Self-Purification on Non-Transitive Preferences via Tournament Graph Reconstruction [25.85736569130897]
  Pairwise evaluation of large language models (LLMs) has become the dominant paradigm for benchmarking open-ended tasks.<n>We show that this critical issue stems largely from low-quality data that contains inherently ambiguous preference pairs.<n>We propose ELSPR, a principled graph-theoretic framework that models pairwise preferences as tournament graphs.
  arXiv  Detail & Related papers  (2025-05-23T10:00:03Z)
• An Unsupervised Natural Language Processing Pipeline for Assessing Referral Appropriateness [0.0]
  We propose a fully unsupervised Natural Language Processing pipeline capable of extracting and evaluating referral reasons.<n>Our pipeline leverages Transformer-based embeddings pre-trained on Italian medical texts to cluster referral reasons.<n>It operates in an unsupervised setting and is designed to generalize across different examination types.
  arXiv  Detail & Related papers  (2025-01-24T18:24:16Z)
• Uncertainty-Penalized Direct Preference Optimization [52.387088396044206]
  We develop a pessimistic framework for DPO by introducing preference uncertainty penalization schemes. The penalization serves as a correction to the loss which attenuates the loss gradient for uncertain samples. We show improved overall performance compared to vanilla DPO, as well as better completions on prompts from high-uncertainty chosen/rejected responses.
  arXiv  Detail & Related papers  (2024-10-26T14:24:37Z)
• Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMs [54.05511925104712]
  We propose a simple, effective, and data-efficient method called Step-DPO. Step-DPO treats individual reasoning steps as units for preference optimization rather than evaluating answers holistically. Our findings demonstrate that as few as 10K preference data pairs and fewer than 500 Step-DPO training steps can yield a nearly 3% gain in accuracy on MATH for models with over 70B parameters.
  arXiv  Detail & Related papers  (2024-06-26T17:43:06Z)
• Learning to diagnose cirrhosis from radiological and histological labels with joint self and weakly-supervised pretraining strategies [62.840338941861134]
  We propose to leverage transfer learning from large datasets annotated by radiologists, to predict the histological score available on a small annex dataset. We compare different pretraining methods, namely weakly-supervised and self-supervised ones, to improve the prediction of the cirrhosis. This method outperforms the baseline classification of the METAVIR score, reaching an AUC of 0.84 and a balanced accuracy of 0.75.
  arXiv  Detail & Related papers  (2023-02-16T17:06:23Z)
• Exploiting prompt learning with pre-trained language models for Alzheimer's Disease detection [70.86672569101536]
  Early diagnosis of Alzheimer's disease (AD) is crucial in facilitating preventive care and to delay further progression. This paper investigates the use of prompt-based fine-tuning of PLMs that consistently uses AD classification errors as the training objective function.
  arXiv  Detail & Related papers  (2022-10-29T09:18:41Z)

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.