Barriers and facilitators to the adoption of artificial intelligence in radiation oncology: A New Zealand study

• Main Author: Koki Victor Mugabe
• Format: Article
• Language: English
• Published: Elsevier 2021-06-01
• Series: Technical Innovations & Patient Support in Radiation Oncology
• Subjects: Artificial intelligence; Data science; Automation; Expertise; Workforce planning
• Online Access: http://www.sciencedirect.com/science/article/pii/S2405632421000184

Introduction: Advances in computing capabilities and automated data collection have led to an increase in the use of Artificial Intelligence (AI) in radiation therapy. This has implications to workflow and workforce planning in radiation oncology departments. A survey was conducted in New Zealand to determine the likelihood of departments adopting AI into their practice. Survey responses were used to determine barriers and facilitators to the adoption of AI. Materials and Methods: An online electronic survey was sent to all ten radiation therapy centres in New Zealand. The survey was sent to radiation oncologists, medical physicists and senior radiation therapists involved in treatment planning. Descriptive analysis, factor analysis, analysis of variance and hierarchical multiple regression were used to analyse the data. Results: AI usage was low across the country and there was middling expertise. Most respondents found AI had a lot of perceived benefits. On the whole, respondents reported a high likelihood to adopt AI. There were significant differences on the Expertise factor between the staff groupsp=0.016 with radiation therapists reporting more expertise than oncologists. Innovation factors (Perceived Benefit) on their own accounted for over 51% of total variance and was the biggest predictor of likelihood to adopt AI(p<0.001). Organisational factors (Expertise) was a moderate predictor(p<0.059). Conclusion: The survey results have been used to investigate the barriers and facilitators to the adoption of AI. These results demonstrate that respondents are likely to adopt AI in their practice. Perceived benefits were a facilitator as high scores were correlated with high likelihood of adoption of AI. Low expertise on the other hand was a barrier to adoption as the low scores were linked to lower likelihood of adoption.