The technology looks for illnesses beyond those that were the primary subject of the test
Spotted: A large network of 11 hospitals, health care facilities, and clinical research departments in Chicago has developed an AI-powered diagnostic support tool which is now included in its electronic health records system. Northwestern Medicine screened more than 460,000 lung radiography images in the year-long pilot of the technology. The goal was to flag signs of conditions beyond those that were the subject of the original referral. Five per cent of the images were marked as needing further investigation – approximately 68 per day.
With medical experts focused on the condition needing immediate attention, secondary health concerns as revealed by diagnostic imaging can easily shift further down the list of priorities. Radiology teams make notes of anything of concern, but, until now, there has been no way to automate the required follow-up actions. The AI system helps take some of the burden off caregivers by automating alerts when an image reveals something additional that needs review.
Patients that use the online care system also receive an alert, and for individuals who are not online, a dedicated team of follow-up nurses ensure that they, too, receive the relevant information. The algorithm learned on exceptionally high-quality data because the hospital system used its own clinical teams to label and sort images.
The system is available online for other facilities to learn from and adapt. Development of the programme includes expanding the AI to other areas of care that use diagnostic imaging, such as for thyroid and ovarian conditions.
AI is growing rapidly as a tool for healthcare teams, with Springwise spotting a diagnostic imaging system that helps prioritise care and an app that supports clinicians in managing in-hospital patient nutrition needs.
Written by: Keely Khoury