Large language models like XLNet show promise in automating CPT code assignment for interventional radiology procedures.
Key Details
- 1Assigning CPT codes for interventional radiology is time-consuming and prone to error, with coding taking an average of 12 minutes per procedure and up to 80% of bills containing errors.
- 2Researchers used XLNet (an open-source LLM) to predict CPT codes from post-procedure reports, focusing on embolization and catheterization cases.
- 3The study analyzed two datasets: one with 1,600 embolization reports (17 CPT codes) and one combining embolization and catheter cases (5,600 reports; 42 CPT codes).
- 4XLNet performed best on vascular embolization and occlusion (code 37243) and central venous-access procedures (code 36597), with some codes being more challenging.
- 5The open-source and local deployment aspects of XLNet support data privacy and accessibility for institutions of all sizes.
Why It Matters
Automating CPT coding in interventional radiology using AI could reduce administrative burden, minimize errors, improve billing accuracy, and facilitate secure, institution-specific solutions. Such advancements may streamline workflows and improve financial outcomes in radiology departments.
Source
Radiology Business
Related News
•Radiology Business
UCLA Appoints Inaugural Associate Dean for Health AI Strategy
UCLA has appointed Katherine P. Andriole as its first associate dean for Health AI Strategy and Innovation, with an initial focus on radiology.
•AuntMinnie
AI Models Reveal Racial Disparities in Breast Cancer Patterns
Machine learning models reveal significant racial disparities and key predictors in breast cancer incidence across diverse groups.
•AuntMinnie
AI Algorithm Streamlines and Standardizes Shoulder Ultrasound Acquisition
A multitask AI system demonstrated high accuracy in standardizing and guiding shoulder musculoskeletal ultrasound imaging.