Collaborating to improve care: A conversation with Smith’s Healthcare Analytics Initiative lead

The Canadian healthcare system continues to be in crisis. Rising costs, talent shortages and the increasing need for care are just a few of the ongoing challenges plaguing the nation’s universal healthcare system.

Since 2023, the federal and provincial governments have been working in lock step to improve areas of Canada’s healthcare system and agree that the availability and use of strong data is crucial to their efforts.

Researchers at Smith have also joined the fight to fix the Canada’s healthcare woes. The business school’s Healthcare Analytics Initiative focuses on applying data analytics techniques to improve healthcare delivery.

“Our ultimate objective is in one way or another to improve healthcare systems,” explains Beste Kucukyazici, E. Marie Shantz Fellow and associate professor of Management Analytics and lead of Smith's Healthcare Analytics Initiative. “It could be cost optimization while keeping the same health outcomes or increasing patient satisfaction, or improving healthcare provider satisfaction so that they don’t leave their jobs.”

To do this, Kucukyazici and other members of the Initiative go straight to the source, working with healthcare providers, hospitals and institutions, to effect real change. Here, she shares the potential of analytics for improving healthcare systems, few of the projects the Healthcare Analytics Initiative has been collaborating on, and plans to extend its impact.

What obstacles do government and healthcare institutions encounter when striving to improve the system?

There are many challenges, but a big one is cost. Whenever we are talking about some healthcare related problem, especially from the management perspective, of course the easiest, or the most intuitive, solution is to add more resources. But in the current system, given that healthcare costs are already very high, that it's at the alarming rate for most of the healthcare systems, adding resources may not be the best option or the ideal option.

What opportunities exist for coming up with solutions that utilize existing resources?

The Initiative team strives for identifying ways to improve the efficiency of resource usage – process optimization, better resource allocation, et cetera.

For instance, in one of the projects that professors Elaheh Fata and Vedat Verter have been collaborating on with Kingston General Hospital and Hotel Dieu, we are trying to improve the access to diagnostic imaging, in particular the CT scanners, by improving the patient scheduling system.

Most patients in Kingston wait for a CT Scan for more than a year. This is quite problematic because, as a tertiary hospital, many patients have life-threatening conditions and timely access to this equipment can be quite critical.

Considering the multiplicity and cost of the needed resources – technicians, radiologists, and equipment – just adding more scanners to the system is not the ideal solution. However, because appointments are scheduled so far out, there are no-show rates that result in wasted time and resources, so we are using historical data to come up with a prediction model to estimate and incorporate no-show rates into the scheduling problem to improve patient scheduling and reduce wait times. 

Many people can be hesitant about the role of AI in healthcare and its possible effect on quality of care. What are your thoughts in this area? 

We are trying to develop AI-based tools to provide extra information to the healthcare provider to be considered.

For example, another project we are working on looks at nurse-to-patient assignment decision-making. These assignments are done manually by nurse managers at the beginning of each shift based on the number of admitted patients and number of nurses on staff. But there are a lot of other variables at play: level of required direct and indirect care, patient distribution and other duties, which can impact nurse workload and patient outcomes.

So, Dr, Pooya Hoseinpour, professor Verter and I have developed a framework to quantify the nurses' workload, taking into consideration the number of patients and these other variables. Combined with RIFD tracking data collected on patients and nurses, we were able develop an optimization tool than can help assign existing patients to the nurses in a given shift. The nurse mangers can make better, more informed decisions that will improve the wellbeing of both workers and patients. The person is still the main decision maker, but we are providing extra data that would otherwise be very difficult for the human brain to incorporate into the decision-making process.

It sounds like there is a lot of momentum behind the Healthcare Analytics Initiative. What’s next?

We currently have a big push underway to expand the initiative to include a lab.

In all our projects, we have been working with undergrads, master’s, PhD students and postdocs. They’ve approached us to be involved in our projects and it’s been great to see these young people, especially at the undergrad level, interested in the application of artificial intelligence and management analytics in healthcare settings.

For example, the CT scan scheduling project (mentioned above) involves an undergrad student to analyze the data and to develop the prediction model for the no-show rate, and a doctoral student who uses those findings and incorporates them in an optimization model to improve patient scheduling and reduce wait times. 

These students are key drivers of our projects. We love working with them, so we’re hoping we can increase the funding for the initiative to further bridge our research into teaching at Smith.