Related Experiment Video
Updated: Aug 6, 2026

Virtual Agent for Real-Time Motivational Interviewing by Integrating Adaptive Nonverbal Behavior and Language Models
Published on: December 23, 2025
AI-driven troubleshooting for TrueBeam systems: Development and testing of a GPT-4.1 Chatbot
Cory Knill1, Sean Devan1, Charles Matrosic1
1Department of Radiation Oncology, University of Michigan, Ann Arbor, MI, United States of America.
A large language model (LLM) chatbot was developed to aid linear accelerator troubleshooting, improving information access during critical patient care. While effective, it requires careful implementation to mitigate safety and workflow risks.
Area of Science:
- Medical Physics
- Artificial Intelligence in Healthcare
- Radiation Oncology Technology
Background:
- Troubleshooting linear accelerator (linac) faults is critical and cognitively demanding.
- Accessing historical troubleshooting data is often slow and relies on individual experience.
Purpose of the Study:
- To develop and evaluate a large language model (LLM)-powered chatbot for streamlining TrueBeam linac troubleshooting.
- To enhance decision-making by providing rapid access to relevant historical information.
Main Methods:
- Extracted and structured eight years of troubleshooting records from five TrueBeam linacs.
- Indexed 1394 logs (5.4 MB) using a GPT-4.1-based environment.
- Evaluated chatbot performance on recall, real-time troubleshooting, aggregation, safety, and temporal filtering, with physicist scoring.
Main Results:
- Chatbot indexing completed in 16 minutes with a mean response time of 7.5 ± 2.5 seconds.
- Demonstrated proficiency in retrieving prior events, summarizing institutional knowledge, and identifying unavailable information.
- Identified weaknesses including procedural misordering, unclear responsibilities, verbosity, and inconsistent temporal filtering.
Conclusions:
- A GPT-4.1-based Retrieval-Augmented Generation (RAG) chatbot can quickly provide institutional knowledge for linac troubleshooting.
- Potential to reduce cognitive burden during machine faults, but safety and workflow risks necessitate guardrails and risk evaluation.
- Recommended as a decision-support tool, not a replacement for expert judgment, prior to clinical deployment.
Related Concept Videos
Non-equilibrium in the Cell
Machines: Problem Solving II
Machines: Problem Solving I
The toggle clamp system is a machine structure consisting of movable, pin-connected multi-force members that form a stabilized system to transmit forces. The...
Transmission Shafts: Problem Solving
Next, use bending moment diagrams for the shaft to...