Development and Assessment of Artificial Intelligence (AI)-Enhanced Pretreatment Peer-review Process to Improve Patient Safety in Radiation Oncology

RT is a critical component of cancer treatment, used by approximately 50 percent of patients with cancer, equating to around 600,000 individuals annually in the United States. While technological advancements in treatment have improved safety, the complexity of these treatments has introduced new pathways for errors, affecting about 3 percent of patients receiving RT and posing considerable risks. Reducing variability among providers during critical pretreatment planning steps presents a significant opportunity to enhance safety and minimize errors. Pretreatment peer-review, a quality check process, by multidisciplinary teams has proven effective in reducing these errors but is often unavailable in rural RT centers.

This research aims to strengthen and automate the peer-review process using AI and machine learning (ML) to improve error detection and address complex treatment factors. With an infrastructure already in place to develop and test these interventions, the proposed work has the potential to significantly improve safety and quality in RT care delivery.

The specific aims of the research are as follows:

• Develop and assess the effectiveness of innovative AI/ML algorithms focused on RT providers’ variability in key treatment planning steps (e.g., defining target volumes, prescribed doses) during pretreatment peer-review processes to improve RT providers’ performance of peer-review.
• Assess the impact of our intervention into the RT work system on patient safety.

The research will be conducted in several phases, beginning with design cycles to develop user-friendly AI/ML algorithms and visual summaries aimed at improving RT providers' peer-review performance. Researchers will measure improvements by tracking errors in treatment plans, the time and effort required for pretreatment peer-reviews, and the usability of the visual tools. A clinical trial will then compare RT cases before and after changes to the peer-review process. Patient safety will be assessed by monitoring missed errors and using statistical models to track safety improvements over time.

This research uses AI/ML to improve patient safety in RT through an innovative pretreatment peer-review process. It is the first RT clinical trial to focus on enhancing safety using AI/ML, with an approach that brings together expertise from multiple medical fields. The study has the potential to improve RT providers’ performance by reducing variability in treatment planning, increasing patient safety by minimizing errors, and creating user-friendly AI/ML tools to support providers. By creating a toolkit for broader use and sharing findings to guide clinical practices and policies, this work could bring robust peer-review processes to underserved patients, preventing thousands of errors each year and greatly enhancing safety and quality in RT care, especially for underserved communities.