Support Ukraine’s Army
Youtube
Facebook
Artificial intelligence in medicine has harmed dozens of patients
Artificial intelligence is being rapidly integrated into medical devices, offering the promise of more accurate diagnoses and safer interventions. However, reports of system malfunctions during procedures and cases of potential patient harm are also accumulating.
Reuters reports on this issue.
How has artificial intelligence in medicine injured dozens of patients?
In 2021, a Johnson & Johnson subsidiary introduced an update to the TruDi navigation system for sinusitis operations. Prior to this update, the regulator had received unconfirmed reports of seven malfunctions and one case of patient injury. After integrating AI, the number of unconfirmed reports increased to at least one hundred, including incidents during procedures.
Between late 2021 and November 2025, at least ten people may have been injured. In most of the reported cases, the system incorrectly showed surgeons where the instruments were located while they were working inside the skull. Reported consequences include spinal fluid leakage through the nose, skull base perforation, and strokes resulting from the accidental injury of a major artery.
Another example involves the Medtronic LINQ series of implanted cardiac monitors with deep learning algorithms. Reports to the FDA mentioned missed rhythm disturbances or pauses. The manufacturer stated that they reviewed sixteen cases and concluded that the device "missed" only one instance; no patient harm was recorded. The company also linked some of the issues to data display errors or user confusion.
Meanwhile, scientists analyzed market feedback and found a connection between 60 FDA-approved AI devices and 182 product recalls. According to their data, 43% of the recalls occurred within a year of receiving approval, which is approximately twice the rate of devices that underwent similar regulatory procedures.
Read more:
