GoAutomate brings AI, financial-sector tech know-how to healthcare

TORONTO – GoAutomate is developing AI-based solutions that are designed to improve workflow in diagnostic imaging departments, which often get bogged down with faxes and paper-based processes. They’re also slowed down by mistakes in documents that are sent electronically. All of these documents regularly contain missing or incorrect data, and staff must fill in the blanks and fix the incorrect information – a time-consuming and tedious process.

Correcting DI requisitions can lead to delays in getting patients into the imaging suite. “Our goal is to reduce patient booking times from two-to-three weeks to one or two days, improving overall patient care,” said Jag Basrai, chief executive officer of GoAutomate.ai.

He added, “Diagnostic imaging can be a stressful environment. We’re trying to alleviate the stress and make sure everyone is as efficient as possible. We want to take away the more tedious tasks.”

The company is devising an AI solution that can automatically check forms, for example, to ensure that a patient’s Ontario Health number is valid, and fixing it when it’s incorrect. It can also check that the name of the referring physician is right, along with his or her contact details – important data for reporting and follow-ups.

Using AI, the system can refer to other documents to fill in missing information. “It’s not ChatGPT, as we’re training our own large language models,” asserted GoAutomate’s chief technology officer, Jason Daly. He said that the company is already testing the solution.

The company has been working on projects at several Ontario hospitals, automating different paper-based processes. “Our users are giving us feedback,” said Daly, helping to improve the product. The deployments allow GoAutomate to use and refine the models in real world scenarios.

Moreover, the AI system can even check to see that the requested exam protocol is appropriate for the condition of the patient, something that staff and radiologists often spend time doing.

The referral can then be booked more quickly.

As Basrai noted, “We’re eliminating a lot of the tedious steps, including re-keying information.” He explained that radiologists will still review the referral and chosen protocols. But for them too, the process has been improved.

Instead of requiring writing or keystrokes, GoAutomate’s solution will let them “tick off boxes”, a much faster and easier process.

Once the exam is approved, the system can search and bring up prior exams for the radiologist. It will automatically contact the patient to remind him or her about the appointment – reducing the number of no-shows. And it will enable the referring physician to log-in to see where the patient is in the diagnostic process.

Daly said that many members of the GoAutomate team – himself included – have backgrounds in the financial industry, where they created bots and other automated solutions to improve workflows.

They’re now bringing these secure technologies to the healthcare sector.

In a separate application, GoAutomate is using AI to blur-out patient names, numbers and other identifiers in diagnostic images when privacy is needed for certain procedures – such as organ transplants.

The company is currently at work outside of Ontario on a large project of this kind. “It’s a new process created by GoAutomate,” said Daly. He said it could be used in research studies, too, when privacy of the patients must be protected, or anonymity is needed for blind studies.

“We know there’s hundreds of use cases this technology could address,” he added.

Basrai mentioned that patients and their families often contact several hospitals or health organizations to get on the list for a wide range of procedures.

That means clerical staff at these hospitals are essentially receiving the same information, and two or three clerical teams could all be keying in the same data. Instead, AI could determine if an application has already been made, and could duplicate the data at the different institutions, saving staff time and eliminating a tedious chore.

“If a large language model does the work, it frees up people to do other things,” said Basrai.

The company says it could even build ‘bots’ that could search across the entire provincial healthcare system when physicians need access to reports and images. While Ontario has been creating repositories that physicians can enter when looking for patient information, it’s still a time-consuming process to go into various systems.

“Even if they’re quick, it could still take a clinician two minutes to search,” said Basrai. “AI does it in 20 seconds.”

Daly, who spent 20 years in the banking sector developing technological solutions, said the healthcare industry is ripe for this kind of “intelligent automation”. And while bots can quickly obtain the desired data, when AI is added, it takes things to a higher level of usefulness.

“The bots can bring the data in, and the large language models can process it,” said Daly. In this way, an AI-powered system can continue to do more of the tiresome tasks in healthcare, enabling humans to focus on higher value and more meaningful work.