“Patient neglect kills. This AI could help stop it” — Fast Company

In the vast majority of hospitals, nurses are required to check on patients at least once an hour. This practice, called hourly rounding, is designed to reduce the number of patient falls and pressure ulcers, which happen when bedridden people don’t move enough. But there’s no way to know whether nurses are checking every hour–or even at all.

Patient neglect is a nationwide problem. According to a 2016 study, medical errors–which include lapses in caregiving that lead to falls and injuries–are the third leading cause of death in the country, causing more than 250,000 deaths per year. “In nursing homes and hospitals, we hear horror stories about neglect,” says Michael Wang, an entrepreneur and registered nurse who worked in the cardiothoracic wing of New York Presbyterian Hospital. “Patients become injured or they die for the very simple fact that no one checked on them.”

Wang’s career began in the military, and he enrolled in Columbia’s nursing program to start his life as a civilian. After earning his degree as an RN, he worked in New York Presbyterian’s cardiothoracic unit at night while tackling an MBA from Columbia during the day. “That’s when I started to really combine what I learned in school and some of the practical issues I witnessed as a practicing bedside nurse,” he says. “I realized that there are huge gaps in the patient care process.”

In 2016, Wang founded the startup Inspiren with the goal of providing data about what actually goes on in hospital rooms. The company’s first product is a device that monitors everything that happens in a patient’s room, paired with an analytics platform to help nurses and hospitals understand how well they’re taking care of patients.

Called iN, the oval-shaped device sits on the wall in every hospital room and uses sensors to detect when a staff member is there. It also uses machine learning algorithms to understand what they’re doing–like turning a patient to prevent them from getting pressure ulcers. The device can recognize when patients are out of bed or if they fall, and raise the alarm.

It sounds a bit like Big Brother, hospital-style. But iN doesn’t use facial recognition to determine who is in the room. Instead, it senses motion and then measures the unique ratio of the hospital staff’s limb lengths, as if they were stick figures. Then, it looks them up in the system to identify who is in the room. To ensure that the device is HIPAA compliant, patients are identified only by their room number and bed number, and the processing happens on the device before the data itself, which is 95% accurate, is beamed up to the cloud.

The company went through 72 different iterations of the design, which is carefully crafted to mitigate the uncanny feeling of being watched. Initially iN was circular, but nurses told Wang and his team that it reminded them too much of an eye that was constantly staring down at them–so he tweaked it to be a friendlier oval. LED lights around its edge indicate a patient’s status–green for all good or orange if a patient hasn’t been seen. Based on feedback from nurses and doctors, the LEDs adjust to the amount of light in the room, ensuring that they’re not too bright when patients are trying to sleep.

Along with the iN monitor, Inspiren’s system comes with an app to help doctors and nurses keep track of which patients they’ve seen and who needs more attention. When a patient hasn’t been seen for an hour or more, nurses get notification reminders to make sure they’re doing hourly rounding effectively.

Kyle Mushet, a nurse at New York Presbyterian, hasn’t tested iN but says the system would help him make sure that no one falls through the cracks. “If we get busy or something’s happening, we don’t get to check in on our patients once an hour like we’d like,” Mushet says. “It’s going to revolutionize the way we’re going to interact with our patients.”

While the device aims to help nurses do their jobs better and save patients’ lives, it’s also useful for hospitals. “Hospitals don’t know what is going on inside the patients’ rooms,” Wang says. “There’s a big hole on real-life information that hospitals cannot keep track of. This translates to cost.”

That’s because when patients fall, develop ulcers, or have any kind of problem due to something the hospital staff has done, it’s the hospital that has to pay out of pocket for it. Insurance companies have skin in this game too–hospitals charge based on man-hours and the complexity of care received, but insurers don’t know if hospitals are accurately reporting how much time doctors and nurses have spent with a patient. iN has the potential to give hospital management the intel into what’s happening inside their walls, preventing neglect, reducing costs, and making sure that people are charged the right amount of money once they leave.

Currently, Inspiren’s system is on trial at a hospital in Queens, and Carolyn Sun, a researcher at Columbia, is working on an initial clinical study to determine its effectiveness. While Sun is still working on gathering baseline data about the current state of hourly rounding, she’s certain that knowing more about what’s going on in hospital rooms will help the patients inside them. “The evidence shows that the more time nurses have with patients, the better the patient outcomes,” she says. “We don’t even know if nurses truly do or don’t do these things. There’s no great data about that. It’s going to be really amazing to have this 360-degree view about what’s going on with the patient.”

Sun sees even greater potential in machine learning intelligence. “If you had all the data, in theory, even if a patient couldn’t tell you, you could predict whether they’re having pains by the look on their face or their heart rate,” she says. “You could have care that was preventative in a whole new way that we’ve never seen before.”

Article by Katharine Schwab.

Read the original article on Fast Company by clicking here.