VHA tests AI-powered translation at point-of-care

TORONTO – Language barriers can be a significant challenge for home care service providers when visiting clients. Often staff rely on options such as informal translators, like family members. Family translators can be very helpful, but they aren’t always available and it may not be a client’s first choice to have a family member be part of what could be a highly sensitive conversation.

As a result, solutions such as this can risk inconsistency, create less equitable care and can place undue burden on clients and their loved ones during vulnerable moments.

With the advancing maturity of translation technology, VHA Home HealthCare (VHA)’s Innovation team saw an opportunity to investigate the use of AI-based tools and discover the viability of implementing an AI translation solution for use in the field by personal support workers (PSWs), nurses and rehab providers. They implemented a study investigating three AI translation tools, supported with funding from the CABHI Discover & Adopt program, to assess how accurate and user-friendly these tools would be in a real-world care setting.

From June to August 2025, VHA’s Innovation Team launched a pilot to evaluate the three tools with nurses, PSWs and rehab providers across areas of Ontario serviced by VHA.

Each tool was assessed for usability, accuracy, workflow impact and feasibility for potential organizational adoption. Each tool supported the top five languages among the client population (Cantonese, Mandarin, Hindi, Russian, and Italian). Feedback from team members during the pilot was used to make updates and improvements, including the addition of four languages.

“This is a core element of VHA’s Innovation portfolio – acting as a testbed for promising technologies for real-world deployment prior to full commercialization,” said Jordan D’Souza (pictured), head of innovation at VHA. “The user feedback not only helped guide VHA on which tool to use in the next phase of the trial but also delivered valuable feedback to developers to help make their product better for others.”

Based on user feedback on ease of use, reliability and functionality, Mabel was selected as the preferred tool to continue to the next phase of testing.

“What I liked about it [Mabel] is that it’s specifically designed for medical professionals,” said a VHA physiotherapist. “That makes it helpful because it has a much better understanding of the terminology we commonly use in our field. It’s not generic, which is a big advantage.”

The next stage of testing will evaluate performance of the system in a greater number of diverse care situations, with approximately 150 VHA rehab and nursing team members involved, to gain a fuller understanding of accuracy in high-demand languages and long-term feasibility.

Embracing an AI translation solution at the point of care can improve equitable care and make communication more reliable for a wide range of multicultural clients in a home care setting. Findings from this study may help provide guidance for hospital and community care partners gathering similar data for their own adoption trials. To learn more and speak to members of VHA’s Innovation team, please reach out to innovation@vha.ca.