PRICE: A Personalized Recursive Intelligent Cost Estimation Framework for Rare Disease Diagnosis

Area of Science:

• Medical Informatics
• Health Economics
• Artificial Intelligence in Medicine

Background:

• Rare disease diagnosis is complex, costly, and time-consuming.
• Traditional cost-effectiveness analyses use static models, failing to account for patient variability.
• There's a need for dynamic, personalized frameworks for diagnostic cost-effectiveness, especially with AI advancements.

Purpose of the Study:

• Introduce the PRICE (Personalized Recursive Integrated Cost-Effectiveness) framework.
• Develop a dynamic model for evaluating diagnostic strategy cost-effectiveness at the individual patient level.
• Accommodate both expert-driven and AI-assisted diagnostic decision-making.

Main Methods:

• Utilize a tree-based analytical model (PRICE) with a back-propagation algorithm for cost computation.
• Quantify effectiveness using a utility-based approach.
• Incorporate parameters like disease prevalence, test costs, performance metrics, and turnaround time for individualized assessments.

Main Results:

• Applied PRICE to developmental delay and multiple congenital anomalies diagnostic strategies.
• Demonstrated PRICE's ability to model outcomes under varying parameters, supporting patient decision-making.
• Showcased how AI performance influences optimal cost-efficient strategies in AI-delegation scenarios.
• Developed an interactive web tool for real-time visualization and simulation of diagnostic pathways.

Conclusions:

• PRICE is a flexible framework capturing sequential diagnostic workflows, adaptable for AI integration.
• Enables personalized economic and clinical evaluations of diagnostic strategies.
• Promotes informed, individualized decision-making, particularly crucial for rare disease diagnosis.