Hospitals are experimenting with machine learning to predict patient emergencies


Hospitals, like much anything else in the modern world, run on machines. Besides the computers that schedule appointments and keep track of occupied rooms, there are a vast array of monitoring devices that read out patients’ vital signs — blood pressure, heart rate, body temperature, breathing rate, and countless other factors. When something goes wrong with any of those vital signs, an alarm goes off. Ideally, this would lead to a doctor or nurse coming around to assess the problem — but in many hospitals, these devices lead to “tens of thousands of alarms” every day, Stat News reports.

So hospitals are turning to artificial intelligence in order to provide the most patients with the most efficient care.

Many hospitals have a command center, which enables a few technicians to monitor patients’ alarms and let hospital staff know when something serious is going wrong. But with hundreds or thousands of patients to keep track of, computers are able to do a much better job at predicting who needs the most immediate care. So by training an AI to pinpoint the warning signs in somebody’s vitals, a hospital’s command center can become much more effective.

Hospitals around the country are already experimenting with training machines to do this life-saving work: At one Cleveland hospital, workers made a breathtaking 77,000 calls to doctors and nurses over the course of just a month. While “most calls are routine,” Stat News explained, some are an indicator of a serious emergency, one where that phone call makes the difference between life and death.

The eventual goal is to give hospital workers more than a few moments’ notice for problems ranging from infections to “serious cardiac events.” But the difference being made already by learning AIs is a promising start.

Source:- theweek