Dr. Frank Rudzicz, Dr. Noah Crampton, Dr. Andrew Pinto

Prof. Frank Rudzicz, Prof. Noah Crampton, Prof. Andrew Pinto
Affiliation: Department of Computer Science, U of T
Research project: Artificial Intelligence Automation to Improve Family Medicine Workflow
Co-Investigators: Omri Nachmani, Hanu Chaudhari, Stephanie Garies, Jane Zhao, Christopher Meaney
Award: Family Medicine (FAFM & CFPC)-Temerty Innovation Grant - $100,000 CAD

Family physicians are increasingly overwhelmed by the use of electronic medical records (EMR). But a new research project, funded with a $100,000 grant funded by T-CAIREM, FAFM, and College of Family Physicians of Canada could change this while improving patient care in the future.

When introduced to primary care, EMRs were touted as powerful tools that would increase efficiency, reduce medical errors, and improve patient care. Instead, primary care providers report higher clerical burdens and damaged patient relationships from screen fatigue and increased work hours. But an AI-based automation workflow tool could help overcome these obstacles and improve the lives of patients and physicians.

The winning research team, led by Co-Principal Investigators Prof. Frank Rudzicz, Prof. Noah Crampton, and Prof. Andrew Pinto , proposed a three-part project to establish the foundation for developing future workflow automation tools.

First, the team plans on conducting a time-motion study with physicians in real-world practice to pinpoint parts of routine EMR tasks that could benefit from an automated AI workflow.

Second, using a novel healthcare dataset extracted from the POPLAR database of 1.8 million Ontarians that matches physicians’ progress notes to corresponding EMR actions, the team will train and validate a state-of-the-art EMR action prediction tool that extracts action intention from progress notes. 

For example, physicians could use the AI medical assistant to streamline their workflow by automating laboratory test requisitions, prescriptions, and referral letters. 

Third, the team will pilot this tool in three family medicine clinics to measure EMR interaction statistics, administrative time saved, and physician satisfaction.

The researchers hope their work leads to a tool that facilitates better interactions with patients, empowers providers to focus more on patients, reduces daily work hours, and lowers administrative costs. 

Furthermore, the researchers hope their project lays the foundation for a digital transformation of primary care by enabling future researchers to conduct workflow automation studies using their new framework and rich dataset.