Studies show that one in 20 patients experiences a medical error while receiving health care, and medication dispensing errors are the most common; the patient is given the wrong medication or the wrong dose. According to the World Health Organization, these errors affect about 1.3 million people a year in the United States alone, and cause at least one death a day. Hospitals have implemented several measures to reduce these errors, including color-coding drugs or using scanners to verify the authenticity of the drug and dosage. However, mistakes still happen constantly. “Some studies show that 90% of anesthesiologists have made medication errors during their career, and this is what prompted me to think about finding a technical solution based on artificial intelligence,” says Dr. Kelly Michaelsen, assistant professor in the Department of Anesthesiology at the University of Washington. Leveraging artificial intelligence to reduce medical errors
Kelly Michaelsen has thought about developing an intelligent system that can detect medical errors related to medicines before they occur. She believes that the types of drugs used in operating rooms are relatively limited, which makes it possible to train an artificial intelligence model to recognize them and work as a doctor's assistant. Kelly Michaelsen focused on a specific type of error known as “bottle switching,” which accounts for about 20% of medication errors. This error occurs when an incorrect drug is used due to confusion in the labels, all injectable drugs come in vials with labels, and are transferred to syringes with similar labels on the medicine cart inside the operating room. But sometimes, you choose the wrong vial or put the wrong labels on the syringe, and the patient is injected with the wrong medicine. In a famous incident, a 75-year-old woman died after being injected with a drug that causes muscle paralysis instead of a sedative, as a result of an error in choosing a vial of the drug. Hence, Kelly Michaelsen's idea took off to develop smart glasses with an AI-powered camera, which can read the labels attached to medicine vials and syringes and check their match. “The system reads the labels, detects any conflict, and triggers an alert at the right moment,”says Kelly Michaelsen. Device development and testing
A new medical innovation.. Artificial intelligence-powered glasses help reduce medical errors John Wiederspan Photo Source: NBC News The construction of this device took more than three years, and he trained his system using realistic footage and analog clips conducted in an environment simulating operating rooms. The study, during which the system was tested, published late last year in the journal npj Digital Medicine, showed that the system was able to monitor bottle switching errors with an accuracy of up to 99.6%. It remains only to determine the appropriate way to alert doctors, whether via voice or visual notification, before applying for FDA approval for actual use in medical institutions. “Just a small sound may prompt you to stop for a moment and check what you are doing,” says Dr. Kelly Michaelsen. Dr. John Wiederspan, who tested the device, confirms that he sees a promising future in it to improve patient safety, although he pointed out that the size of the device is large. A new medical innovation.. Artificial intelligence-powered glasses help reduce medical errors Dr. Kelly Michaelsen and John Wiederspan
Opportunities and challenges associated with the use of artificial intelligence to reduce medical errors
Many experts believe that artificial intelligence can make a real difference in reducing medical errors, just as self-driving cars have contributed to reducing road accidents. But they also emphasize that this technique is not a panacea. Technology is a powerful tool, but it is not enough. Errors are often the result of a series of gaps, not just one factor. In high-pressure environments such as operating rooms and emergency departments, AI tools may be more effective than paper lists or traditional alerts because they silently monitor and reduce the burden on medical staff.
Kelly Michaelsen is considering expanding the capabilities of the device to include calculating the dose volume inside the syringe, which is extremely important, especially in Pediatrics; dose sizes vary greatly depending on the weight of the child. John Wiederspan hopes that the technology will later be developed to also include oral medications, especially in departments where patients are taking several pills at once.