Robots take the guesswork out of doctors’ handwriting at NZ’s Mercy Radiology

The use of robotic process automation has been key in digitising mundane processes and better engaging customers—all without costing human jobs.

Doctor prepared to examine the patient    105770502
Thinkstock

That doctors’ handwriting is notoriously difficult read has become something of a truism, but robotic process automation (RPA) can make short work of even the most indecipherable scrawl. That’s the experience of Mercy Radiology, a network of 13 clinics, where the technology is achieving 98% accuracy rates when processing clinical referrals.

Mercy Radiology CEO Lloyd McCann says it’s often been difficult for staff to transcribe clinician notes, but giving the job to the robots resulted in fewer transcription errors and a process that has “sped up remarkably”.

McCann says that the first robot, Matilda, was on-premises, but it and the additional Rob-e robot are now in the cloud, which has enabled more functionality, such as the use of optical character recognition technology to process referrals.

“As robots can consume large amounts of data, we scan all of our referrals and so it gives the AI all of that data and it keeps learning. It keeps improving at a very rapid rate. So, the way somebody writes an ‘i’ or an ‘e’ or an ‘r’, once the application has seen that a number of times it is then very good at recognising that that is actually the character that person has written down. It’s that continuous learning loop and the ability to ingest huge amounts of data that makes it quite accurate,” he says

The Matilda robot now handles 98% of invoices automatically, and what previously took human workers eight to 10 hours, Matilda completes in two hours. This has allowed Mercy to invoice daily, instead of weekly, which has enhanced cash flow by $200,000 per month.

To continue reading this article register now

Download CIO's Roadmap Report: 5G in the Enterprise