The increased amount of data revealed on advanced diagnostic imaging studies in recent years (e.g., particularly those modalities using cross-sectional imaging such as MRI and CT), has created awareness among radiologists of the large number of radiological findings that may be unrelated to the primary focus of a study ordered by the patient's treating physician. These unrelated findings are referred to as "incidental."
A new study published in "NEJM Catalyst Innovations in Care Delivery" reviews the use of artificial intelligence (AI) at Northwestern Medicine in Chicago to identify and alert patients and their ordering physicians to incidental lung findings in radiology reports. Although many incidental findings are benign, some can develop into malignancies. Thus, timely follow-up to these findings can improve a patient’s long-term health. And they portend a new era of patient engagement in radiology.
Realizing the importance to patients of following up on these findings, as well as the professional liability risks from not following up on clearly reported results, a team at Northwestern created a system using AI to address certain incidental findings. An AI system is integrated into the electronic health record (EHR) that evaluates nearly every imaging study ordered within the hospital system. Using AI to identify incidental findings, actionable Best Practice Advisory (BPA) notifications alert the ordering physician or the patient's primary care physician to such incidental findings. Those physicians receive communications highlighting the incidental finding within EHR where they can order follow-up studies as they determine appropriate.
The Catalyst study reviewed the first year of implementation of the AI system. Out of more than 460,000 imaging studies screened in the study, 23,000 were flagged as containing incidental lung follow-up recommendations, representing 5% of the studies – 68 findings per day that required follow-up.
I love a comment from the study’s lead author Jane Domingo, MBA, MS, CCC-SLP, program manager, System Clinical Performance Improvement Office at Northwestern Medicine, that reflects the potential for using AI to assure follow-up. “As patients, we often believe that ‘no news is good news,' and if we don’t hear from our physician after a test, we tend to assume everything is fine and don’t proactively follow-up. By building patient notification into our process, we are adding another layer of safety by creating informed patients who can advocate for themselves.”
Systems that can alert patients of incidental findings - and recent innovations in patient-centered reporting of radiology findings we have recently discussed here - are examples of accelerated patient engagement emerging in today's diagnostic radiology departments.
An artificial intelligence natural language processing (NLP) system was developed to identify radiology reports containing [incidental] lung- and adrenal-related findings requiring follow-up.