Integrating Chatbots in Radiology Workflow: A Primer

Purpose: The purpose of this work is to describe the potential benefits and challenges of integrating chatbots into the radiology workflow.
1. To describe chatbot systems: types, building, and deployment into radiology workflows.
2. To illustrate our experience integrating chatbots in our radiology department - How do we do it?
3. To evaluate chatbot models in radiology workflow.

Methods/Materials: We develop an implementation guideline for chatbots system in radiology workflow, including:
1. Identifying the specific areas of the radiology workflow that would benefit from integrating a chatbot.
2. Designing the chatbot's conversational interface and determining the appropriate natural language processing (NLP) and machine learning (ML) techniques to be used.
3. Training the chatbot on a dataset of radiology-related information.
4. Integrating the chatbot into the radiology workflow.
5. Evaluating the chatbot's performance and effectiveness in the radiology workflow.
6. Examplify this guideline developing and evaluating three chatbots deployed in our workflow, including CDS Chatbot for thoracic imaging in ER using the ACR appropriateness criteria, Chatbot for Standardized Reporting for Lung Cancer Staging, and Chatbot as Clinical Decision Support tool on Thyroid Nodule Assessment.

Results: We developed a completed guideline for chatbot development and integration into radiology workflow and illustrated it with real-life examples. We described the development and deployment of three chatbots in the radiology workflow. The results showed high accuracy in identifying the appropriate imaging study for thoracic conditions in the ER, a significant reduction in time spent on report generation, and improved diagnostic accuracy for thyroid nodules management. These chatbots implementation provided cost savings, improved patient outcomes, and more efficient workflow for the radiology department.

Conclusions: The integration of chatbots into the radiology workflow has shown several potential benefits. Our work found that a chatbot-based system significantly increased radiologist productivity, improved patient satisfaction, reduced errors, and increased quality in radiology.