Evaluating sentiment analysis models in healthcare: addressing bias and enhancing interpretability
Frontiers in Public Health · 19.03.2026 · 13
DOI: 10.3389/fpubh.2025.1663871 ↗Kurz gefasst
Fachartikel in der Zeitschrift Frontiers in Public Health, online erschienen am 19.03.2026. Verfasst von Chenxu Wang und 2 weiteren Autor:innen. Beteiligte Einrichtung: Department of Emergency, First Clinical Medical College, Nanchang University, Nanchang, China. Hauptthemen laut MEDLINE: Delivery of Health Care, Models, Statistical, Artificial Intelligence. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 47 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Introduction
Advancing trustworthy AI applications in healthcare necessitates systems that are not only high-performing but also capable of explaining decisions and addressing biases, particularly in critical tasks like sentiment analysis on clinical narratives and patient feedback. Conventional sentiment analysis methods, while effective in general applications, struggle with domain shift, linguistic variability, and ambiguous labeling in healthcare, limiting their interpretability and fairness in clinical contexts. To overcome these limitations, a novel sentiment analysis framework is proposed to improve both accuracy and interpretability.
Methods
This framework employs a formal probabilistic modeling approach that incorporates fine-grained sentiment granularity and domain-aware priors. Central to the framework is the Sentiment Modulated Encoding Network (SMEN), a transformer-based architecture featuring a gating mechanism that dynamically enhances sentiment-relevant features across network layers, enabling rich sentiment representation learning without external resources. Additionally, the Context Polarity Decoupling Scheme (CPDS) disentangles sentiment from domain-specific artifacts through a multi-stage adversarial and contrastive training process, accompanied by a polarity explanation module that provides token-level interpretability. Results and Discussion Together, SMEN and CPDS form a robust system capable of producing domain-invariant and explainable sentiment predictions. Experimental results on multiple healthcare datasets demonstrate superior generalization and more transparent model attributions compared to existing approaches. This research contributes to the development of explainable and bias-resistant AI tools for healthcare and highlights potential avenues for interdisciplinary exploration at the interface of affective computing and clinical informatics.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Emergency, First Clinical Medical College, Nanchang University, Nanchang, China.
- Inner Mongolia University of Technology, Hohhot, China.
Zitieren
APA
Wang, C., Miao, Z., & Zeng, H. (2026). Evaluating sentiment analysis models in healthcare: addressing bias and enhancing interpretability. Frontiers in Public Health, 13. https://doi.org/10.3389/fpubh.2025.1663871
Vancouver
Wang C, Miao Z, Zeng H. Evaluating sentiment analysis models in healthcare: addressing bias and enhancing interpretability. Frontiers in Public Health. 2026;13. doi:10.3389/fpubh.2025.1663871
BibTeX
@article{Wang2026,
author = {Wang, Chenxu and Miao, Zhuang and Zeng, Haoran},
title = {Evaluating sentiment analysis models in healthcare: addressing bias and enhancing interpretability},
journal = {Frontiers in Public Health},
year = {2026},
volume = {13},
doi = {10.3389/fpubh.2025.1663871}
}
EUVIMED