AI may illuminate the path to value-based health care

Key takeaways:

• Unity Health Toronto has developed over 50 AI tools to provide value to patients, administrators and clinicians.
• The speaker said the next step will be to engage the private sector to expand their impact.

ORLANDO — AI can be harnessed to provide value to patients, health care administrators and clinicians, said a presenter at the AIMed24 Annual Meeting.

According to Muhammad Mamdani, MPH, PharmD, MA, vice president of data science and advanced analytics at Unity Health Toronto, many surgeons struggle defining value in health care because “it is a bit obscure... there is no standard definition.”

However, Mamdani said that he and his colleagues at Unity Health Toronto have developed AI tools that intend to both define and prioritize value in care.

“The unique thing about us is that we are the only health care organization in Canada, and I would argue one of the few in the world, that has formally declared AI as a core strategic pillar,” Mamdani said. “From the board of directors to the CEO of the entire organization, AI is one of those core pillars.”

Tools

At Unity Health Toronto, Mamdani leads a data science and advanced analytics team, which drives AI development in the hospital system. He said his team has developed and deployed over 50 AI solutions in clinical practice.

Among the solutions his team has advanced, Mamdani highlighted an AI nurse staffing optimization tool that tracks nursing absence patterns and predicts how many nurses an institution will need to hire.

“We were able to document a million-dollar annual cost reduction,” Mamdani said. “It took about 6 weeks to develop. It can be done.”

In addition, Mamdani said his team developed an AI scribe that listens to patient conversations and transcribes the notes into patient records. After deploying the tool in two of the Unity Health Toronto hospitals, “on average, 3 to 4 hours per week were freed up, and over 80% said they would continue using the AI scribe,” Mamdani said.

He also noted the development of an AI early warning system called ChartWatch. The program runs hourly, using between 150 and 170 different parameters to provide alerts when a patient is at high risk for death or admittance to the ICU within the next 48 hours.

According to Mamdani, recent results from a study he published in the Canadian Medical Association Journal linked ChartWatch with a 26% reduction in unexpected mortality. In addition, he said the program had reduced false-positive rates compared with other early warning systems in the industry.

“We are obsessed with making sure that we actually have solutions that perform well to give value to our clinicians,” Mamdani said.

Next steps

Mamdani said his team’s work is not finished. He noted that the next step will be to expand the scale of their impact.

“We have to engage the private sector, startups and those sorts of groups who know how to do this well and can take on responsibility,” Mamdani said. “Now, we are creating an ecosystem to engage the private sector to co-create with us, so that we not only develop solutions that will help our patients but that will also help patients far beyond our four walls.”

He closed by asking the audience to ponder the following when considering a technological solution, such as those driven by AI.

“The first thing I would ask when you are trying to determine what program you should have, what you should buy, or what you should build, is what do you value? And let that drive your journey.”