# AI-Powered Detection of Inappropriate Language in Medical School Curricula

arXiv: https://arxiv.org/abs/2508.19883 · HTML: https://mlsanity.com/paper/2508.19883/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 7 of its 40 references that are in the map.
- First posted on arXiv 27 Aug 2025.

## Builds on (most cited first, 7 of 7)

- [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://mlsanity.com/paper/1910.01108/) (2019)
- [BioBERT: a pre-trained biomedical language representation model for biomedical text mining](https://mlsanity.com/paper/1901.08746/) (2019)
- [BioSentVec: creating sentence embeddings for biomedical texts](https://mlsanity.com/paper/1810.09302/) (2018)
- [A Network Perspective on Stratification of Multi-Label Data](https://mlsanity.com/paper/1704.08756/) (2017)
- [Reducing Biases towards Minoritized Populations in Medical Curricular Content via Artificial Intelligence for Fairer Health Outcomes](https://mlsanity.com/paper/2407.12680/) (2024)
- [Fairness via AI: Bias Reduction in Medical Information](https://mlsanity.com/paper/2109.02202/) (2021)
- [Hidden or Inferred: Fair Learning-To-Rank with Unknown Demographics](https://mlsanity.com/paper/2407.17459/) (2024)

## Abstract (arXiv)

The use of inappropriate language -- such as outdated, exclusionary, or non-patient-centered terms -- medical instructional materials can significantly influence clinical training, patient interactions, and health outcomes. Despite their reputability, many materials developed over past decades contain examples now considered inappropriate by current medical standards. Given the volume of curricular content, manually identifying instances of inappropriate use of language (IUL) and its subcategories for systematic review is prohibitively costly and impractical. To address this challenge, we conduct a first-in-class evaluation of small language models (SLMs) fine-tuned on labeled data and pre-trained LLMs with in-context learning on a dataset containing approximately 500 documents and over 12,000 pages. For SLMs, we consider: (1) a general IUL classifier, (2) subcategory-specific binary classifiers, (3) a multilabel classifier, and (4) a two-stage hierarchical pipeline for general IUL detection followed by multilabel classification. For LLMs, we consider variations of prompts that include subcategory definitions and/or shots. We found that both LLama-3 8B and 70B, even with carefully curated shots, are largely outperformed by SLMs. While the multilabel classifier performs best on annotated data, supplementing training with unflagged excerpts as negative examples boosts the specific classifiers' AUC by up to 25%, making them most effective models for mitigating harmful language in medical curricula.