Discussing the role of artificial intelligence (AI) in improving outcomes within research and practice development, Hamid Mojibian (Yale School of Medicine, New Haven, USA) explicated the ways in which AI is changing interventional radiology (IR) at the 2023 Global Embolization Oncology Symposium (GEST, 18–21 May, New York City, USA).
“AI is here” was Mojibian’s introductory statement, pointing to the definite AI integrations that are already occurring in radiology. Expanding on this, he described an AI-based workflow that can identify critical findings through image acquisition, automatic AI application and prioritisation of key findings. These findings can then trigger an alert which connects patients with the most appropriate care teams, and subsequently automate patient follow-up.
Using pulmonary embolism (PE) as an example, Mojibian draws parallel comparison between the workflow with and without AI—with more steps and a less direct pathway to treatment in the non-AI roadmap, Mojibian highlighted how AI can enable a “more effective” care system.
Following this, he discussed disease discovery modules that utilise natural language processing (NLP) to analyse radiology reports. Potential uses for this technology span from uterine fibroids, spine compression fractures and detecting inferior vena cava (IVC) filters that are eligible for removal—Mojibian stated in a single month they identified 91 filters that were eligible. Furthermore, he added that in deep venous thrombosis cases, NLP systems have the potential to increase patient recruitment for randomised trials.
Continuing to outline the effectiveness of AI’s employment within healthcare, Mojibian explained a prospective research workflow enabled by the technology. Initiated by a notification, he described how AI can then enable patient stratification and inclusion through integration with electronic medical records. He opined that this system improves efficiency and maximises collaboration throughout the research process, while also incorporating advanced image processing which can benefit classification, segmentation and quality control.
Of his concluding thoughts Mojibian noted that AI will not replace clinicians, but clinicians who do not take up AI technology will be replaced by those who do. Now is the time to plan and act, he stated, emphasising that greater awareness is needed to spot opportunities to lead and co-innovate with fellow healthcare professionals.