AI-assisted software for chronic wound detection: Development, validation, and certification strategy.

AIMS: This study evaluates the feasibility of an artificial intelligence (AI)-assisted software tool for early identification and classification of chronic wounds and for supporting decisions on whether professional clinical review is needed. METHODS: A pilot assistive software solution was developed by integrating wound image analysis with patient-reported metadata. The system combines You Only Look Once (YOLO) for wound localization and U-Net for tissue segmentation. A standardized wound imaging and annotation protocol was designed in collaboration with the University Hospital Bulovka. Diagnostic performance will be evaluated using sensitivity, specificity, positive predictive value and negative predictive value. RESULTS: A Python-based prototype and a multimodal software architecture were developed. More than 500 standardized wound images have been collected to date, and the dataset is being expanded iteratively according to model performance, class balance, annotation quality and validation requirements. A regulatory certification strategy was outlined under the Medical Device Regulation and the AI Act. CONCLUSION: The study establishes a technical and regulatory foundation for future AI-based wound care tools intended to support earlier triage and professional assessment of chronic wounds.