Evaluating the Impact of Implementing AI-Assisted Contouring on Radiotherapy Staff and Workflows
Author(s)
Catriona Inverarity, PhD1, Juan I. Baeza, PhD2, Jon Hindmarsh, PhD2, Benjamin Caswell-Midwinter, PhD1, Babak Jamshidi, PhD1, Angela A. Kehagia, MD PhD1, Anna Barnes, PhD1.
1King's Technology Evaluation Centre (KiTEC), King's College London, London, United Kingdom, 2Department of Public Services Management & Organisation, King's College London, London, United Kingdom.
1King's Technology Evaluation Centre (KiTEC), King's College London, London, United Kingdom, 2Department of Public Services Management & Organisation, King's College London, London, United Kingdom.
OBJECTIVES: This multi-centre, multi-vendor evaluation aimed to provide an holistic assessment of the impact of AI-assisted contouring in radiotherapy departments. The qualitative research sought to describe the impact on staff and workflows, complementing and adding insight to the quantitative data that was collected in parallel.
METHODS: Semi-structured interviews were conducted with 32 staff from five NHS radiotherapy departments. Interviews were structured across three domains: workforce dynamics, trust in the AI contouring system, and training and communication regarding the technology. Transcribed responses were coded using NVivo prior to reflexive thematic analysis. Interview findings were contextualised with quantitative data regarding acceptability of the AI-generated contours and the staff involved in reviewing contours, as well as pathway maps describing manual and AI-enabled contouring workflows.
RESULTS: Staff unanimously reported time savings with AI contouring compared to manual methods. Time released by AI-enabled contouring was variously cited as allowing time for training, service development and improvement, supporting work-life balance and managing increased demand on radiotherapy services. A change in the staff mix involved in contouring was also observed in AI pathways, with less reliance on clinical oncologists for contour review (100% for manual contouring versus 39% AI-assisted contouring cases). This confers greater operational freedom for optimising workflows and staffing. AI contouring tools were widely embraced and accepted by staff, although some variation in the acceptability of contours between different anatomies was observed. Engagement and discussion across stakeholders were seen as important in achieving support and buy-in from staff.
CONCLUSIONS: AI-assisted contouring was overwhelmingly embraced by departments. Expansion of professional roles involved in AI-contour review confers additional flexibility to staffing and resourcing. Departments reported using pre-determined dates for processes on the treatment planning, so local pathway audit before and after adoption could help capitalise on time released by allowing departments to optimise AI-assisted pathways according to local requirements.
METHODS: Semi-structured interviews were conducted with 32 staff from five NHS radiotherapy departments. Interviews were structured across three domains: workforce dynamics, trust in the AI contouring system, and training and communication regarding the technology. Transcribed responses were coded using NVivo prior to reflexive thematic analysis. Interview findings were contextualised with quantitative data regarding acceptability of the AI-generated contours and the staff involved in reviewing contours, as well as pathway maps describing manual and AI-enabled contouring workflows.
RESULTS: Staff unanimously reported time savings with AI contouring compared to manual methods. Time released by AI-enabled contouring was variously cited as allowing time for training, service development and improvement, supporting work-life balance and managing increased demand on radiotherapy services. A change in the staff mix involved in contouring was also observed in AI pathways, with less reliance on clinical oncologists for contour review (100% for manual contouring versus 39% AI-assisted contouring cases). This confers greater operational freedom for optimising workflows and staffing. AI contouring tools were widely embraced and accepted by staff, although some variation in the acceptability of contours between different anatomies was observed. Engagement and discussion across stakeholders were seen as important in achieving support and buy-in from staff.
CONCLUSIONS: AI-assisted contouring was overwhelmingly embraced by departments. Expansion of professional roles involved in AI-contour review confers additional flexibility to staffing and resourcing. Departments reported using pre-determined dates for processes on the treatment planning, so local pathway audit before and after adoption could help capitalise on time released by allowing departments to optimise AI-assisted pathways according to local requirements.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
HTA133
Topic
Economic Evaluation, Health Technology Assessment, Medical Technologies
Topic Subcategory
Systems & Structure
Disease
No Additional Disease & Conditions/Specialized Treatment Areas, Oncology