Work in Progress: AI-Assisted pH-Based Colorimetric Monitoring of Chronic Wound Healing Using Fluorescent Hydrogel Dressings
Chronic wounds, such as diabetic foot ulcers (DFUs), require continuous monitoring to prevent complications like infection and amputation. Traditional treatment methods include frequent dressing changes and subjective visual inspections which are often invasive, painful, and unreliable. As a non-invasive alternative, pH-sensitive fluorescent silk fibroin (SF) hydrogel dressings are gaining attention due to their ability to reflect wound pH, a key biomarker of healing (pH 4–6 for healthy skin; pH 7–8 for chronic, non-healing wounds), through fluorescence changes along with subtle colorimetric shifts. However, accurately interpreting these alterations without specialized equipment remains a challenge. To address this, we developed a hybrid machine learning model combining ResNet18 for image feature extraction with a Random Forest Classifier (RFC) for pH prediction. Trained on 2,112 images captured under simulated pH conditions (5–8), the model achieved 83% accuracy and an AUC-ROC of 0.96, with strong recall for pH 7–8. This enables accurate, real-time wound assessment using standard imaging devices, offering a scalable solution for point-of-care DFU monitoring.
