CPR Pattern Recognition
Tools & Methods
Axure
Machine Learning
Team
Designer (me)
Developer
My Contribution
Idea
How it works
UI
Impact
Invented a new method to recognize who administered a specific set of CPR using machine learning to save reviewers time and improve care.
Problem
QI Reviewers are unable to view the CPR performance of an individual paramedic or EMT because there is no way to capture who did it at the moment of care. At best, EMS agencies have to assess clinicians as a homogenous group, meaning the crew of 2-3 people, based on collective CPR performance.
Solution
By using Artificial Intelligence/Machine Learning (AI/ML) we can analyze the unique combination of CPR rate, depth, and release velocity post-case to identify a unique CPR "signature" of an individual.
1. How it generally works
By using the employee information from the PCR we can have a baseline to run calculations off of. We can rely on it to tell whether or not there is a new employee, or if our calculations find a person that is unaccounted for, we can disregard them as they are likely a firefighter or another helping hand on scene.
At this point, the user then can challenge the system's guess as to who gave a particular set of CPR and we will remodel accordingly so, future guesses will be more accurate.
2. What users see
Impact
This solution makes it much quicker to review EMTs and identify which individuals might need more training, therefore improving level of care and the possibility of saving someone's life.