(PI-020) Closing the Loop: A Responsible Artificial Intelligence Solution for Evidence-Based Wound Care Decision Support

Tania Santos, MSc, Nurse; Lourdes Muñoz-Hidalgo, MSc, Nurse; Kátia Furtado, PhDs, Nurse; Elaine Song, MD, PhD, MBA; Tiffany Hamm, RN, CWON, CWS; Jeffrey Mize, RRT, CHT, CWCA

Introduction:

Introduction:
Clinicians encounter an average of 11 clinical questions daily, most of which remain unanswered¹. While artificial intelligence (AI) holds promise², unconstrained generative AI for diagnoses or treatment plans presents significant risks of patient harm and malpractice liability. In typical clinical decision-making, AI achieves an accuracy of only ~52%, frequently failing to identify life-threatening conditions³. To mitigate these risks and provide reliable, evidence-based answers in wound care, we aimed to develop a responsible, AI-powered, point-of-care clinical intelligence solution leveraging a Portuguese wound care society’s digital clinical pathways.

Methods:

Method:
Using Design Thinking⁴ , the solution was developed as a module within a decision support platform.

      ●  One year after the wound care society turned their printed pocket guide library into a comprehensive, multi-media digital clinical pathways library, an evaluation revealed that while 86% of members found the pathways helpful, access was cumbersome⁵.
● To solve these navigation issues, a proprietary closed-loop AI model was utilized to retrieve, aggregate, and generate answers strictly from the Society's interactive digital pathway library.
User feedback optimized the interface across multiple iteration

Results:

Results:

The resulting mobile-responsive module successfully leveraged AI-powered search and generative answers that are vetted, evidence-based, and reliable. Members can instantly found trusted answers, linked back to the peer-reviewed pathways. Unlike open-AI models, this closed-loop solution effectively curbs hallucinations, providing only referenced and society-vetted answers. Additionally, the platform optionally collects de-identified pathway data to generate insights on cost-effective interventions for specific patient populations.

Discussion:

Discussion: A responsible AI-powered clinical intelligence solution for wound care was successfully developed, with the potential to significantly enhance point-of-care decision-making for the society’s members. Generative AI should supplement - not replace - human judgment, requiring responsible use and robust oversight.