Philips Healthcare manufactures, manages and maintains expensive and sophisticated IT machines for the healthcare sector all over the world.  And when their machines break down, repair is usually complex, requiring skilled engineers, and above all very costly.

Far better then to implement early warning systems that detect and forecast impending machine failure and notify an engineer that something is wrong?

That’s exactly what Philips Healthcare has done with its fleet of some 150 MRI scanners across the US, Germany and India using SMS as the delivery system.

However, when the chiller fails, so do all the other parts. A rise in temperature will turn the liquid to gas, which expands rapidly and causes a cascade that shuts down the cryogenics system. When this happens, engineers need to restore the machine at room temperature. Such failures can cost up to $100,000 to resolve.

Finding fault

John Romero, a national support specialist (MRI) at Philips Healthcare, was thinking about this problem in 2014. He says: “In the past, customers would call when a machine went down. But you can’t expect a customer to anticipate a fault. Their priority is getting the patient image done, not the functioning of the machine, so they would keep working till it broke down. We had to figure out a way for the machine’s thermostat itself to tell the engineers when something was wrong.”

Eventually, Romero rigged up a module using a Raspberry Pi, which triggered a text alert when there was any change in temperature. Later he called in messaging specialist OpenMarket to handle the routing of the texts to the engineers. He says: “We considered email, but the point about text is that everyone has a phone, everyone can read SMS and messages get through instantly. Meanwhile working with the OpenMarket API meant we didn’t have to worry about network coverage or permissions.”

Success helps

It took Romero six months to build the prototype, but the tests were a success and now 150 machines across the US, Germany and India have the ability to send ‘help’ messages. In two years, they’ve sent thousands resulting in that $250,000 saving.

Interestingly, though the project has kept machines running, it has actually required more man hours to maintain. However, Romero says this is delivering unexpected benefits. He says: “With all the extra data we have now on machine failures, we can build a more holistic view of what works. We can start thinking about replacing parts that we sense may be about to expire, for example, rather than waiting for an alert saying they are already failing.”

Philips Healthcare is now looking to embed messaging functions into 3,000 machines worldwide, and is considering extending the concept to other departments for use in nuclear medicine, cat scan and cardio vascular systems.