From no coding skills to AI-ready in three days: Introducing Vector’s Excel to AI for T-CAIREM
October 11, 2022
October 11, 2022
By Jonathan Woods
October 11, 2022
An exciting future for AI in healthcare is coming into focus. In Ontario hospitals, AI has already been used to predict emergency room admissions, automate nursing assignments, and optimize nurse schedules, and there’s much more to come. But beyond datasets and algorithms, there’s another element that’s vital to making that future real for healthcare: specialized training for health researchers, students, and clinicians so that they can understand AI and use it to enhance their work.
Vector’s new pilot program, Excel to AI, was designed to get health experts ready to do just that.
Over three days in June, Excel to AI brought health researchers up to speed on foundational AI skills and concepts – even if they had no previous coding experience. The program equipped Microsoft Excel mavens working in health with practical skills in Python, the programming language of choice for many AI applications, in a matter of hours. The aim: to enable them to perform better data analysis on health projects and take their first steps into the world of AI.
Vector designed the Excel to AI pilot in partnership with The Temerty Centre for Artificial Intelligence Research and Education in Medicine (T-CAIREM). T-CAIREM is a University of Toronto-based organization whose vision is to transform healthcare through AI research, education, and infrastructure, leading to “more efficient treatment methods, lower health costs and create better health outcomes for Canadians.”
In three two-hour sessions held over a week, six T-CAIREM researchers went from learning Python’s basics to generating custom visualizations, creating data pipelines, and exploring how traditional machine learning libraries can solve business problems.
“We started from the very basics,” said Yinka Oladimeji, Excel to AI instructor and Technical Education Specialist at Vector. “We started building up gradually and gave them assignments at the end of every day. Day One and Day Two, we introduced them to what Python can do. By Day 3, they were already building data pipelines and introduced to the fundamentals of AI, including some simple models.”
Figure 1. Excel to AI syllabus. Students quickly progress from Python basics to exploring how machine learning can solve real problems in their work.
Daily assignments focused on performing health-related data tasks using Python. These included:
In each lesson, the program prioritized practicality. “These medical students and scientists can start using what they learned in the program immediately,” said Oladimeji. “They regularly have to do a lot of sampling and exploratory data analysis. Now they’re able to use Python to, for example, split their data into test and training datasets or run basic regression.”
Excel to AI’s final assignment required participants to consider Python applications in their regular work.
Faraz Honarvar, a medical student at Queen’s University’s School of Medicine, is currently working on a project that uses AI to identify biomarkers in medical resonance imagery to predict carotid artery disease. The project’s dataset comprises images from over 300 people, with each biomarker requiring manual identification. The project’s goal, he said, “is to develop machine learning algorithms that could identify these biomarkers more accurately, and, as a result, predict carotid artery disease more efficiently.”
Lessons from the program can simplify some research tasks involved. “For someone with little background in AI and coding, Python becomes the ideal easy-to-use tool for me, which can help integrate Excel data into AI modalities seamlessly,” he said.
Another participant, Darla, is an Undergraduate Research Fellow with The Centre for Addiction and Mental Health. Reslan is involved in a project that could use AI to understand the effect that annual clinical metrics target-setting has on hospital performance. The project currently involves structuring data from hospital quality improvement plans in Excel, but Reslan also sees an important role for Python.
“I think Python will be a great way to explore the data and analyze some trends of interest,” Reslan said. “How have institutional performances changed over time? How do changes in the targets [or] measures selected for use by the institutions affect performance?”… Python will be a great tool for this: once the data is structured, I can easily explore the relationship between different variables.”
Reslan also feels that Excel to AI has improved her ability to collaborate with colleagues that specialize in machine learning.
“[A] research project I am working on now is specifically concerned with building machine learning for emergency psychiatric health care,” she said. “I imagine that having this Introduction to Python for data analysis and machine learning will be extremely useful in allowing me to understand my team members’ work and better perform my research tasks.”
Using AI to improve patient care and lower healthcare costs is a priority for both Vector and T-CAIREM. The Excel to AI pilot is an important step on the journey toward that end, demonstrating that health experts can rapidly gain the foundational skills needed to learn AI, and more importantly, find ways to employ those skills in their everyday research and work.
According to Oladimeji, it’s also lit a fire among participants to build AI proficiency.
“On the third day, we started talking how you could use Python to work with AI models. That got everyone excited. It stirred up a lot of interest,” he said. “And they’re now actually ready to learn a lot more about AI.”
Sign up for Vector’s newsletter today to learn more about Vector professional development programs!