AI in Hospitals: MRI Checks Cut from 400 Minutes to Under 5

At the radiology department of Singapore's National University Hospital, staff no longer have to comb through roughly 160 MRI record sets a day by hand. According to senior radiographer Teo, AI searches MRI referral forms, device-related records and past radiology reports all at once, leaving only the patients whom clinicians need to check directly. The work that once took 400 minutes now takes less than 5 minutes. Identifying patients with metal implants before a scan must not go wrong, so whether safety has improved along with the time savings is the central question in this case.

That such cases are being presented together is no coincidence. Hospitals in many countries face the same problem: aging populations are driving up demand for care while medical staff remain scarce. The 49th IHF World Hospital Congress opened on the 20th at the COEX Magok Convention Center in Seoul, where hospital officials shared their experiences of using AI and data to tackle practical problems on the ground. One session on the practical results of data and AI drew more than 200 attendees and filled the hall.

Hong Kong's Hospital Authority has taken things a step further by trying to read risks in advance. By analyzing 21 clinical variables, including vital signs and test results, the system flags the risk of patient deterioration within the next 48 hours. When an alert sounds, a nurse examines the patient first, and a doctor decides on tests and treatment. After six months of operation, the system identified an average of about 35 patients a day, and average hospital stays fell by 1.5 days. However, these figures are operational results published by the hospital, and no data verified by an external body has been made public. The hospital is gradually expanding the programme to other hospitals.

Samsung Electronics has looked beyond hospital walls. In his keynote, Vice President Park Heon-soo said that linking health data collected through wearables and home appliances with hospital care would allow clinicians to see a patient's condition in greater depth and intervene at the right time. His argument is that prevention is not the job of hospitals alone. He said the company is pursuing plans to apply digital care programmes to hospital operations and to reflect information patients contribute in clinical decisions. However, the presentation did not specify any concrete products or adoption timelines.

So the question is who will benefit from this trend, and when. The most immediate beneficiaries are radiology departments, where long scan waits and record-keeping burdens are common, and patients whose risk signals are caught late and who end up in intensive care. The Singapore and Hong Kong cases are already running in clinical settings, for periods ranging from several months to more than a year, but whether other hospitals will see the same results has not yet been confirmed. Lee Wang-jun, chair of the organizing committee and incoming IHF president, said AI will bring fundamental change to hospitals and healthcare systems, but stressed that hospital leaders must use it safely and responsibly. In other words, verification must come before speed.