Fine-Grained Classification of Pressure Ulcers and Incontinence-Associated Dermatitis Using Multimodal Deep Learning: Algorithm Development and Validation Study
Feingranulare Klassifikation von Dekubitus und inkontinenzassoziierter Dermatitis mit multimodalem Deep Learning: Entwicklung und Validierung eines Algorithmus
JMIR AI · 01.05.2025 · 4 · e67356-e67356
DOI: 10.2196/67356 ↗Auf einen Blick
Ein multimodales Deep-Learning-Modell unterschied zwischen Dekubitus und inkontinenzassoziierter Dermatitis mit hoher Genauigkeit. Es übertraf dabei Wundexpert:innen und Pflegefachpersonen auf dem Testdatensatz.
Fragestellung
Kann ein multimodales Deep-Learning-Framework Dekubitus und inkontinenzassoziierte Dermatitis sowie deren Schweregrade genau klassifizieren?
Methode
Entwicklungs- und Validierungsstudie mit 1.555 annotierten Wundbildern und Konsens von 4 Wundexpert:innen. Verglichen wurden 4 Modelle mit Bilddaten und kategorialen Patientendaten.
Ergebnisse
TinyViT erreichte in der binären Klassifikation von Dekubitus vs. IAD den besten F1-Score von 93,23 %. Für feingranulare Kategorien war TinyViT bei PU-Kategorien mit 75,43 % am besten, ConvNeXtV2 bei IAD-Kategorien mit 53,20 %.
Einschränkungen
Die Angaben beziehen sich auf Testdaten; Validierung in realen Versorgungsumgebungen wird als künftiger Schritt genannt.
Schlussfolgerung der Autor:innen
Das multimodale Framework differenziert Dekubitus und IAD wirksam und erzielt hohe Genauigkeit. Die Autor:innen sehen es als potenzielles Decision-Support-Tool für die klinische Praxis.
Kernaussagen
- Binäre Klassifikation: TinyViT mit F1-Score von 93,23 %
- Feingranular: PU 75,43 %; IAD 53,20 %
- Multimodale Daten verbesserten die binäre Klassifikation
Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.
Kurz gefasst
Fachartikel in der Zeitschrift JMIR AI, online erschienen am 01.05.2025. Verfasst von Alexander Brehmer und 10 weiteren Autor:innen. Beteiligte Einrichtung: Institute for Artificial Intelligence in Medicine, Essen University Hospital, Girardetstr. 2, Essen, 45131, Germany, 0201 72377829. Die Arbeit wurde bisher 7-mal zitiert (Crossref) und nennt 40 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar.
Themen: Inkontinenz, Dekubitus
Abstract (Originalsprache)
Background
Pressure ulcers (PUs) and incontinence-associated dermatitis (IAD) are prevalent conditions in clinical settings, posing significant challenges due to their similar presentations but differing treatment needs. Accurate differentiation between PUs and IAD is essential for appropriate patient care, yet it remains a burden for nursing staff and wound care experts.
Objective
This study aims to develop and introduce a robust multimodal deep learning framework for the classification of PUs and IAD, along with the fine-grained categorization of their respective wound severities, to enhance diagnostic accuracy and support clinical decision-making.
Methods
We collected and annotated a dataset of 1555 wound images, achieving consensus among 4 wound experts. Our framework integrates wound images with categorical patient data to improve classification performance. We evaluated 4 models—2 convolutional neural networks and 2 transformer-based architectures—each with approximately 25 million parameters. Various data preprocessing strategies, augmentation techniques, training methods (including multimodal data integration, synthetic data generation, and sampling), and postprocessing approaches (including ensembling and test-time augmentation) were systematically tested to optimize model performance.
Results
The transformer-based TinyViT model achieved the highest performance in binary classification of PU and IAD, with an F1-score (harmonic mean of precision and recall) of 93.23%, outperforming wound care experts and nursing staff on the test dataset. In fine-grained classification of wound categories, the TinyViT model also performed best for PU categories with an F1-score of 75.43%, while ConvNeXtV2 showed superior performance in IAD category classification with an F1-score of 53.20%. Incorporating multimodal data improved performance in binary classification but had less impact on fine-grained categorization. Augmentation strategies and training techniques significantly influenced model performance, with ensembling enhancing accuracy across all tasks.
Conclusions
Our multimodal deep learning framework effectively differentiates between PUs and IAD, achieving high accuracy and outperforming human wound care experts. By integrating wound images with categorical patient data, the model enhances diagnostic precision, offering a valuable decision-support tool for health care professionals. This advancement has the potential to reduce diagnostic uncertainty, optimize treatment pathways, and alleviate the burden on medical staff, leading to faster interventions and improved patient outcomes. The framework’s strong performance suggests practical applications in clinical settings, such as integration into hospital electronic health record systems or mobile applications for bedside diagnostics. Future work should focus on validating real-world implementation, expanding dataset diversity, and refining fine-grained classification capabilities to further enhance clinical utility.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Institute for Artificial Intelligence in Medicine, Essen University Hospital, Girardetstr. 2, Essen, 45131, Germany, 0201 72377829.
- Center for Virtual and Extended Reality in Medicine, University Medicine Essen, Essen, Germany.
- Faculty of Computer Science, University of Duisburg-Essen, Essen, Germany.
- Department of Nursing Development and Nursing Research, University Hospital Essen, Essen, Germany.
- Department of Clinical Nursing Research and Quality Management, Hospital of the Ludwig Maximilian University, Munich, Germany.
- Sciendis GmbH, Leipzig, Germany.
Zitieren
APA
Brehmer, A., Seibold, C., Egger, J., Majjouti, K., Tapp-Herrenbrück, M., Pinnekamp, H., Priester, V., Aleithe, M., Fischer, U., Hosters, B., & Kleesiek, J. (2025). Fine-Grained Classification of Pressure Ulcers and Incontinence-Associated Dermatitis Using Multimodal Deep Learning: Algorithm Development and Validation Study. JMIR AI, 4, e67356-e67356. https://doi.org/10.2196/67356
Vancouver
Brehmer A, Seibold C, Egger J, Majjouti K, Tapp-Herrenbrück M, Pinnekamp H, et al. Fine-Grained Classification of Pressure Ulcers and Incontinence-Associated Dermatitis Using Multimodal Deep Learning: Algorithm Development and Validation Study. JMIR AI. 2025;4:e67356-e67356. doi:10.2196/67356
BibTeX
@article{Brehmer2025,
author = {Brehmer, Alexander and Seibold, Constantin and Egger, Jan and Majjouti, Khalid and Tapp-Herrenbrück, Michaela and Pinnekamp, Hannah and Priester, Vanessa and Aleithe, Michael and Fischer, Uli and Hosters, Bernadette and Kleesiek, Jens},
title = {Fine-Grained Classification of Pressure Ulcers and Incontinence-Associated Dermatitis Using Multimodal Deep Learning: Algorithm Development and Validation Study},
journal = {JMIR AI},
year = {2025},
volume = {4},
pages = {e67356--e67356},
doi = {10.2196/67356}
}
EUVIMED