Between Myths and Metaphors: Rethinking LLMs for SRH in Conservative Contexts

2025-10-31

AI Summary

In conservative cultural contexts (e.g., Pakistan), sexual and reproductive health (SRH) communication relies heavily on indirect language, posing significant challenges for large language models (LLMs), including semantic drift, metaphor misinterpretation, and lexical ambiguity. To address this, we developed a two-dimensional framework for analyzing indirect communication—grounded in clinical observation, clinician–patient interviews, and focus groups—to systematically identify myth-based narratives, figurative expressions, and pragmatic features in SRH dialogues. We empirically evaluated five state-of-the-art LLMs across real-world, low-resource healthcare settings, characterizing their semantic parsing failures. Building on these findings, we propose a culture-adapted fine-tuning strategy and human–AI interaction design principles, substantially improving model accuracy in detecting and responding to implicit, culturally embedded expressions. Our work establishes a theoretical foundation and practical methodology for enhancing cross-cultural robustness and localized deployment of health AI systems.

Technology Category

• Natural Language Processing: Safety and Robustness
• Machine Learning: Large Multimodal Models (LMMs)
• Humans and AI: AI for Accessibility

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Low-resource countries represent over 90% of maternal deaths, with Pakistan among the top four countries contributing nearly half in 2023. Since these deaths are mostly preventable, large language models (LLMs) can help address this crisis by automating health communication and risk assessment. However, sexual and reproductive health (SRH) communication in conservative contexts often relies on indirect language that obscures meaning, complicating LLM-based interventions. We conduct a two-stage study in Pakistan: (1) analyzing data from clinical observations, interviews, and focus groups with clinicians and patients, and (2) evaluating the interpretive capabilities of five popular LLMs on this data. Our analysis identifies two axes of communication (referential domain and expression approach) and shows LLMs struggle with semantic drift, myths, and polysemy in clinical interactions. We contribute: (1) empirical themes in SRH communication, (2) a categorization framework for indirect communication, (3) evaluation of LLM performance, and (4) design recommendations for culturally-situated SRH communication.

Problem

• LLMs struggle with indirect SRH communication in conservative contexts
• Semantic drift and cultural myths hinder automated health assessments
• Existing models fail to interpret nuanced clinical interactions accurately

Innovation

• Analyzed indirect SRH communication themes in Pakistan
• Evaluated five LLMs performance on cultural interpretations
• Proposed design framework for culturally-situated health communication