Verification error and supervision error – new types of medical errors resulting from the use of artificial intelligence systems occurring during the provision of health services

Abstract

Medical devices based on artificial intelligence have been used to provide health services for several years. Changes in the diagnostic and treatment process have made it necessary to distinguish new types of medical error - verification error as a subcategory of diagnostic error and supervision error as a subcategory of therapeutic error. Artificial intelligence is only a tool used by the physician. A doctor is obliged to verify the correctness of the diagnosis made by the AI. If the medical knowledge allows the physician to conclude that the results of the analysis performed by artificial intelligence are incorrect, and the physician nevertheless makes a diagnosis in accordance with the AI findings, a verification error will occur. When a surgeon is able to use a fully autonomous robot, the activities he or she undertakes will change completely. In such a case, physician will only be obliged to supervise the surgical robot and take over the surgical procedure if any malfunction in its operation is observed. If physician does not stop the autonomous robot although, according to medical knowledge, it does not work properly or the surgeon could perform the surgical procedure with his own hands more precisely, or the surgeon’s procedure would result in less damage to health, he will make a supervision error.