Innovative Two-Phase Deep Learning System for Automated Classification of Pressure Injuries
-
By
-
Ting-Yu Lai
-
Yi-Jiun Chou
-
Chun-You Liu
-
Chien-Wei Chen
-
Ching-Ting Lin
-
Wei-Chun Wang
-
Yimin Hsu
-
Ming-Li Hsieh
-
Shih-Sheng Chang
-
March 27, 2026
Clinical Scorecard: Innovative Two-Phase Deep Learning System for Automated Classification of Pressure Injuries
At a Glance
| Category | Detail |
|---|
| Condition | Pressure Injury (PI) |
| Key Mechanisms | Two-stage deep learning approach combining object detection (YOLOv9) and image classification (DenseNet161) |
| Target Population | Hospitalized patients, particularly in intensive care units |
| Care Setting | Clinical workflows for nursing assessments |
Key Highlights
- Developed an AI framework for automatic PI staging from clinical images
- Achieved an overall accuracy of 0.775 in staging on an independent test set
- Demonstrated improved diagnostic accuracy and reduced subjectivity in PI staging
- Framework provides standardised and reproducible PI staging
- Potential integration into nursing workflows as a clinical decision-support tool
Guideline-Based Recommendations
Diagnosis
- Utilize AI-driven classification systems for objective PI staging
Management
- Implement two-stage deep learning frameworks in clinical practice to enhance PI assessment
Monitoring & Follow-up
- Regularly evaluate the performance of AI models in real-world clinical settings
Risks
- Challenges in handling intra-wound heterogeneity and variability in image quality
Patient & Prescribing Data
Patients with pressure injuries in clinical settings
AI system may assist in providing consistent and objective staging for treatment planning
Clinical Best Practices
- Incorporate AI tools to support nursing assessments of pressure injuries
- Ensure high-quality image acquisition for accurate AI analysis
- Train clinical staff on the use of AI systems for enhanced decision-making
Related Resources & Content
