AI implementation improves when radiology staff concerns are considered
Incorporating artificial intelligence applications into clinical workflows holds promise for improving efficiency, reducing burnout, addressing staffing issues and more. But although the potential benefits of AI are numerous, its implementation is not without headaches for healthcare staffers.
A new paper published this week in Clinical Radiology details the experience of a large healthcare institution in London that integrated an AI chest radiography triage application to flag suspicious exams for further evaluation. Authors of the paper sought to better understand how the implementation affected the roles of numerous staff positions, both clinical and administrative, and whether the resultant changes were perceived as positive or negative.
"Successful deployment of AI hinges on a thorough understanding of these human factors (as well as the technical landscape) to ensure continued, smooth and sustainable operation of services,” Susan Cheng Shelmerdine, from the Department of Clinical Radiology at Great Ormond Street Hospital for Children in London, and co-authors noted.
The tool would mark chest radiographs as suspicious, prompting techs to request a second read, which had to be completed within 30 minutes of radiologists being alerted. If the reader agreed with AI’s suspicion, techs would notify the CT department and attempt to schedule the patient for a same-day chest CT scan. A certain number of openings were left in the CT schedule to accommodate patients who were sent for same-day imaging.
Staff involved in the process completed surveys related to their workflow, perceptions of AI and how its integration affected patient care at three different timepoints throughout the integration process—before implementation, one month after and then again seven months later—to determine how the changes were received by employees.
Initially, most respondents were optimistic about how AI could increase workflow efficiency. The later surveys revealed a shift in attitudes, highlighting greater confidence in AI’s ability to improve patient care.
Concerns about how job roles might be affected were noted in the early responses, but these feelings later transitioned to frustration with challenges related to the technical aspects of implementation. Later during the post-implementation phase, staff warmed up to the software, acknowledging its positive impact on patient care, though logistical issues with workflows remained a point of contention.
The staff’s initial beliefs that AI might improve efficiency did not translate to their workflows, but It did streamline diagnostic processes for patients, which was perceived as its greatest benefit. When asked whether they would be comfortable if they were the patient receiving an AI read on their imaging, around half of the group reported they would accept the additional interpretation.
“This study underscores the necessity of engaging with all staff members to fully grasp the multifaceted impacts of AI integration,” the authors wrote. “A clear understanding of the care pathway, including patient flow, department logistics, and the minimization of additional burdens on staff, is crucial for the successful adoption of AI tools.”
The group suggested that any organization looking to integrate AI into their workflows take time to consider staff opinions and prepare to adapt strategies accordingly to minimize frustration during rollout.
In addition to her background in journalism, Hannah also has patient-facing experience in clinical settings, having spent more than 12 years working as a registered rad tech. She began covering the medical imaging industry for Innovate Healthcare in 2021.