AI OrthoTriage Abstract The goal of AI OrthoTriage is to establish a standardized benchmark for the use of AI in assessing joint and orthopedic conditions. Not only does musculoskeletal and joint pain negatively affect over 1.7 billion people globally, but patients relying on AI for their medical diagnoses is also increasing rapidly. A recent
OpenAI report found that about 40 million people use ChatGPT every day to ask health-related questions. AI OrthoTriage is simple: it uses two different scores and weighs both to return three different benchmarks to the user. Green (66-100) indicates the user can refer to the advice AI gave them with confidence, yellow (32-65) indicates they should be cautious and refer to their physician before proceeding and a red (0-33) means the advice AI gave is not clinically credible.
The first score we assess is called the Safety Score, which measures the accuracy of AI’s recommendations to objective, clinical, ESI guidelines. The second score is called the LBI Score, which factors in different literacy levels and different ways to phrase health scenarios. Together, the two scores are used to return a green, yellow and red benchmark to the user.
This dual-score framework is feasible because it leverages existing clinical guidelines and real-world language variation without requiring complex infrastructure or proprietary data. By making AI recommendations transparent and explicitly tied to safety and bias checks, AI OrthoTriage empowers users to understand when to trust AI and seek a physician, thereby strengthening public confidence in responsible medical AI use.
Log in or sign up for Devpost to join the conversation.