Enhancing Wound Care Management with Artificial Intelligence to Automate Tissue Identification

Introduction: Accurate assessment and documentation of wound tissue composition are crucial to monitoring wound healing progress or deterioration. 1 However, clinician evaluation of tissue types is subjective and varies due to knowledge and skills in quantifying tissues, leading to inconsistent, variable assessments.1 One approach to addressing this challenge is to use deep learning to train an artificial intelligence model (SmartTissueTM) using tens of thousands of labelled images to predict tissue types. 2 This study presents clinician evaluations of tissue types in contrast to the AI’s predictions.Methods:An observational cross-sectional study involving wound care clinicians was conducted to evaluate the performance of a deep learning-based tissue segmentation model. Clinicians were asked to assess and identify the tissue types in 9 different wound images, which varied in complexity and patient skin tone. Their assessments were compared with the automated segmentation results produced by the AI model. The study aimed to understand how clinician estimations of tissue types vary compared to the AI-model and stratify the findings based on years of experience, training, wound complexity, and tissue type.Results:Descriptive data will be presented, including the clinician's background, years of experience, and levels of specialized wound experience. Clinician confidence identifying the selected tissue types and confidence in quantifying these tissue types will also be reported. The range, average, and inter and intra-rater reliability of clinicians’ ratings of tissue types will be analyzed and compared to the predicted automated AI-model segmentations.Discussion: The study highlights the significant variation in clinician estimations of wound tissue types, influenced by experience, training, and wound complexity. Consistent, reliable and clinically acceptable data on wound tissue is important for wound prognostics. 3  AI models can reduce clinician documentation time, while increasing the quality of data. Further analysis on the impact of the AI model in research and practice is required to support implementation and evaluation of clinical, operational and financial impacts.References:1. Bates-Jensen, B. (2022). Assessment of the patient with a wound. In L. L. McNichol, C. R. Ratliff, & S. S. Yates (Eds.), Wound, ostomy, and continence nurses society core curriculum: Wound management (2nd ed., pp. 55–91). Wolters Kluwer. 2. Gupta, R., Goldstone, L., Eisen, S., Ramachandram, D., Cassata, A., Fraser, R. D. J., Ramirez-GarciaLuna, J. L., Bartlett, R., & Allport, J. (2022). Towards an AI-based Objective Prognostic Model for Quantifying Wound Healing [Preprint]. https://doi.org/10.36227/techrxiv.21067261.v1 3. Gupta, R., Goldstone, L., Eisen, S., Ramachandram, D., Cassata, A., Fraser, R. D. J., Ramirez-GarciaLuna, J. L., Bartlett, R., & Allport, J. (2022). Towards an AI-based Objective Prognostic Model for Quantifying Wound Healing [Preprint]. https://doi.org/10.36227/techrxiv.21067261.v1