Karen Ung and Mani Kang stand in an office hallway.

Karen Ung and Mani Kang

Sign up for the Unity Health Toronto newsletter, a monthly update on the latest news, stories, patient voices and research emailed directly to subscribers. If you haven’t subscribed yet, you can do that by clicking here.

Unity Health Toronto is one of the first hospitals in Canada to work with Google Cloud to develop a generative artificial intelligence (Gen AI) tool to speed up the process of reviewing complex research contracts. The tool aims to reduce turn-around times for clinical trial contracts, resulting in a faster launch of groundbreaking health research and the development of interventions and treatments aiming to improve patient care.

Every year, Unity Health’s research contracts team receives roughly 1,400 research contracts. Of these, about 250 are clinical trial agreements, which are the most complex.

Before patients can enroll in a clinical trial, there are a number of approvals required, including ensuring that the clinical trial agreement is negotiated and signed between Unity Health Toronto and the external party. A clinical trial agreement is a legally binding agreement that governs the conduct of a particular study and outlines the obligations of each party. These include complex legal terms, insurance requirements, privacy obligations and study drug warranties to name a few.

The process of reviewing a 40-50 page clinical trial agreement involves multiple reviews, back-and-forth communication between Unity Health and the other party, as well as consultation with legal counsel. As such, it can take a highly-specialized reviewer up to 14-18 hours to complete a first review, says Manager of Research Contracts, Karen Ung.

Right now, the process of reviewing these agreements is all manual, with a team of skilled reviewers relying on their expertise, guidance documents, principles, and policies to revise and redline the agreement, Ung said. Anything redlined is then proposed to the other party for further discussion, negotiation and editing.

“Clinical trial agreements come with the most amount of complexity but they also provide significant benefits to Unity Health Toronto. Most importantly, they provide access to treatments – sometimes potentially life-saving treatments– that would otherwise be unavailable for patients,” said Ung. “What we’re hoping to do is create a tool that completes the first round of redlines for us, which allows the reviewer to go through the draft, fine tune it, and ensure everything is complete before sending it to the external party. Ideally, we’d like to get to the points of contention with the other party faster, so we can review and mitigate them, sign the contract and get the trial up and running.”

By using Google Cloud’s AI to complete the first redlines, the tool will free up valuable time for research contracts staff, so they can do other critical tasks, adds Senior Director of Research Operations Mani Kang.

“By completing that first step faster, staff can do other things like negotiate with counterparties, which AI is not going to be able to do,” Kang said. “This is a tool that will aid our highly-trained staff to do what they do best. We see this as augmentation, not automation.”

The tool is being built with Google Cloud’s Vertex AI platform and leverages Google’s large language models. Unity’s model will be fine-tuned in a secure and confidential environment using its own administrative data, policies and parameters, said Kang.

There will be a learning period, where Unity Health will supply the developers, Quantiphi, with several hundred data points. The development team will also receive Unity Health’s guidance documents and checklists. This learning period will be followed by a 24-month period, where the tool will be continuously optimized and fine-tuned through daily usage by Unity Health research staff.

At the end of this optimization period, the tool will have evolved from the original foundational model to better match the needs of Unity Health’s research institute, said Ung. 

“Some may see this investment in technology as risky, because we’re not buying an off-the-shelf product, and therefore we don’t know exactly what we’re going to get,” Ung said. “But we’re confident that we can develop something that will ultimately help the team save time while ensuring that we are agreeing to the best terms possible for Unity Health and our patients.”

Kang agrees.

“There’s definitely a level of excitement around this technology, but this project is grounded in the principles that make our organization tick,” Kang said. “We’re doing this for a reason, and that reason is to get faster access to the latest research and innovation for our patients.”

By Marlene Leung

Photo by Eduardo Lima