Evaluating few-shot prompting for spectrogram-based lung sound classification using a multimodal language model
PLOS Digital Health · 07.01.2026 · 5 (1) · e0001179
DOI: 10.1371/journal.pdig.0001179 ↗Kurz gefasst
Fachartikel in der Zeitschrift PLOS Digital Health, online erschienen am 07.01.2026. Verfasst von Nicholas Dietrich und 2 weiteren Autor:innen. Beteiligte Einrichtung: Temerty Faculty of Medicine, University of Toronto, 1 King's College Cir, Toronto, Ontario, Canada. Die Arbeit nennt 56 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Traditional deep learning models for lung sound analysis require large, labeled datasets, whereas multimodal large language models (LLMs) may offer a flexible, prompt-based alternative. This study aimed to evaluate the utility of a general-purpose multimodal LLM, GPT-4o, for lung sound classification from mel-spectrograms and assess whether a few-shot prompt approach improves performance over zero-shot prompting. Using the ICBHI 2017 Respiratory Sound Database, 6898 annotated respiratory cycles were converted into mel-spectrograms. GPT-4o was prompted to classify each spectrogram using both zero-shot and few-shot strategies. Model outputs were evaluated against ground truth labels using performance metrics including accuracy, precision, recall, and F1-score. Few-shot prompting improved overall accuracy (0.363 vs. 0.320) and yielded modest gains in precision (0.316 vs. 0.283), recall (0.300 vs. 0.287), and F1-score (0.308 vs. 0.285) across labels. McNemar’s test indicated a statistically significant difference in performance between prompting strategies (p < 0.001). Model repeatability analysis demonstrated high agreement (κ = 0.76–0.88; agreement: 89–96%), indicating excellent consistency. GPT-4o demonstrated limited but statistically significant performance gains using few-shot prompting for lung sound classification. While current performance remains insufficient for clinical deployment, this prompt-based approach provides a baseline for spectrogram-based multimodal tasks and a foundation for future exploration of prompt-based multimodal inference.
Autor:innen und Institutionen
- Temerty Faculty of Medicine, University of Toronto, 1 King's College Cir, Toronto, Ontario, Canada.
- Faculty of Engineering, McMaster University, Hamilton, Ontario, Canada.
- Translational Medicine, Hospital for Sick Children Research Institute, Toronto, Ontario, Canada.
ORCID-Profile: Nicholas Dietrich, David McShannon
Zitieren
APA
Dietrich, N., McShannon, D., & Rzepka, M. F. (2026). Evaluating few-shot prompting for spectrogram-based lung sound classification using a multimodal language model. PLOS Digital Health, 5(1), e0001179. https://doi.org/10.1371/journal.pdig.0001179
Vancouver
Dietrich N, McShannon D, Rzepka MF. Evaluating few-shot prompting for spectrogram-based lung sound classification using a multimodal language model. PLOS Digital Health. 2026;5(1):e0001179. doi:10.1371/journal.pdig.0001179
BibTeX
@article{Dietrich2026,
author = {Dietrich, Nicholas and McShannon, David and Rzepka, Mark F},
title = {Evaluating few-shot prompting for spectrogram-based lung sound classification using a multimodal language model},
journal = {PLOS Digital Health},
year = {2026},
volume = {5},
number = {1},
pages = {e0001179},
doi = {10.1371/journal.pdig.0001179}
}
EUVIMED