WER is Unaware: Assessing How ASR Errors Distort Clinical Understanding in Patient-Facing Dialogue

This repository hosts the code, models, and datasets accompanying the paper. The work investigates how Automatic Speech Recognition (ASR) errors distort clinical meaning in patient-facing dialogue — and shows that traditional metrics like Word Error Rate (WER) fail to capture real clinical risk. The project includes scripts for aligning ground-truth utterances to ASR-generated utterances using an LLM-based semantic aligner, and optimizing an LLM-as-a-Judge for clinical impact assessment using GEPA through DSPy.

📝 Abstract

As Automatic Speech Recognition (ASR) is increasingly deployed in clinical dialogue, standard evaluations still rely heavily on Word Error Rate (WER). This paper challenges that standard, investigating whether WER or other common metrics correlate with the clinical impact of transcription errors. We establish a gold-standard benchmark by having expert clinicians compare ground-truth utterances to their ASR-generated counterparts, labeling the clinical impact of any discrepancies found in two distinct doctor-patient dialogue datasets. Our analysis reveals that WER and a comprehensive suite of existing metrics correlate poorly with the clinician-assigned risk labels (No, Minimal, or Significant Impact). To bridge this evaluation gap, we introduce an LLM-as-a-Judge, programmatically optimized using GEPA to replicate expert clinical assessment. The optimized judge (Gemini-2.5-Pro) achieves human-comparable performance, obtaining 90% accuracy and a strong Cohen's κ of 0.816. This work provides a validated, automated framework for moving ASR evaluation beyond simple textual fidelity to a necessary, scalable assessment of safety in clinical dialogue.

🔍 Overview

We introduce (available here):

Clinician-annotated clinical-impact dataset: llm_judge/dataset/primock_data_final_outcomes.csv
Semantic LLM-based aligner: alignment/aligner/ (see alignment/README.md for usage)
LLM-as-a-Judge optimized with GEPA/MIPRO: llm_judge/ (artifacts in llm_judge/results/)
Evaluations of ASR metrics (code under alignment/scripts/ and alignment/results/)

🛠️ Environment Setup

Install Python 3.10+ and uv (recommended): https://github.com/astral-sh/uv
Install dependencies: uv sync
Environment:
- OpenRouter (default for LLM calls): OPENROUTER_API_KEY (required), OPENROUTER_MODEL optional
- Gemini (optional): GCP_PROJECT_ID, GCP_LOCATION
- Bedrock (optional): AWS_REGION

Example: run aligner evaluation

uv run python alignment/scripts/run_evaluation.py --case-id sample --asr-system demo

Example: run judge (GEPA)

uv run python -m llm_judge.cli.run_gepa \
  --data-path llm_judge/dataset/primock_data_final_outcomes.csv \
  --provider openrouter \
  --task-model meta-llama/llama-3.3-70b-instruct \
  --reflection-model anthropic/claude-4-sonnet \
  --output llm_judge/results/clinical_judge_gepa.json

📁 Folder Structure

alignment/ — semantic alignment toolkit (aligner code, scripts, sample data, sample results).
llm_judge/ — clinical impact judge (signatures, metrics, providers, optimizers, CLI, bundled dataset, saved judges).

Important Files

alignment/data/ — example ASR transcripts and ground-truth alignments.
alignment/results/ — sample alignment evaluations.
llm_judge/dataset/ — clinical-impact dataset.
llm_judge/results/ — optimized judges (GEPA, MIPROv2).

📦 Coming Soon

Additional dataset metadata and documentation
Evaluations of 20+ ASR metrics, showing their poor correlation with clinical safety

📄 Paper

Preprint available on arXiv: https://arxiv.org/abs/2511.16544

📚 Citation

@misc{ellis2025werunawareassessingasr,
      title={WER is Unaware: Assessing How ASR Errors Distort Clinical Understanding in Patient Facing Dialogue}, 
      author={Zachary Ellis and Jared Joselowitz and Yash Deo and Yajie He and Anna Kalygina and Aisling Higham and Mana Rahimzadeh and Yan Jia and Ibrahim Habli and Ernest Lim},
      year={2025},
      eprint={2511.16544},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2511.16544}, 
}

4.3 KiB Raw Blame History